42 changed files with 996 additions and 5092 deletions
--- a/.gitignore
+++ b/.gitignore
@ -1,9 +1,4 @@
 *~
 .eggs/
 dist/
 wheelhouse/
 cosmotool.egg-info/
 build/
 *.o
 *.prog
 *.pyc
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@ -42,17 +42,6 @@ ENDIF(BUILD_PYTHON)
 MESSAGE(STATUS "Using the target ${CosmoTool_local} to build python module")
 find_library(ZLIB_LIBRARY z)
 find_library(DL_LIBRARY dl)
 find_library(RT_LIBRARY rt)
 if (RT_LIBRARY)
  SET(RT_DEP rt)
  message(STATUS "RT found, linking against it")
 else()
  SET(RT_DEP)
 endif()
 find_library(MATH_LIBRARY m)
 include(${CMAKE_SOURCE_DIR}/external/external_build.cmake)
 IF(YORICK_SUPPORT)
@ -66,6 +55,9 @@ IF(YORICK_SUPPORT)
 ENDIF(YORICK_SUPPORT)
 find_program(CYTHON NAMES cython3 cython)
 find_library(ZLIB_LIBRARY z)
 find_library(DL_LIBRARY dl)
 find_library(MATH_LIBRARY m)
 set(NETCDF_FIND_REQUIRED ${YORICK_SUPPORT})
 set(GSL_FIND_REQUIRED TRUE)
@ -78,8 +70,8 @@ SET(CPACK_PACKAGE_DESCRIPTION_SUMMARY "A toolbox for impatient cosmologists")
 SET(CPACK_PACKAGE_VENDOR "Guilhem Lavaux")
 SET(CPACK_RESOURCE_FILE_LICENSE "${CMAKE_CURRENT_SOURCE_DIR}/LICENCE_CeCILL_V2")
 SET(CPACK_PACKAGE_VERSION_MAJOR "1")
-SET(CPACK_PACKAGE_VERSION_MINOR "3")
+SET(CPACK_PACKAGE_VERSION_MINOR "0")
-SET(CPACK_PACKAGE_VERSION_PATCH "4${EXTRA_VERSION}")
+SET(CPACK_PACKAGE_VERSION_PATCH "0${EXTRA_VERSION}")
 SET(CPACK_PACKAGE_INSTALL_DIRECTORY "CosmoToolbox-${CPACK_PACKAGE_VERSION_MAJOR}.${CPACK_PACKAGE_VERSION_MINOR}")
 SET(CPACK_STRIP_FILES "lib/libCosmoTool.so")
 SET(CPACK_SOURCE_IGNORE_FILES 
--- a/MANIFEST.in
+++ b/MANIFEST.in
@ -1,5 +1,4 @@
 include  .gitignore
 include  pyproject.toml
 include  CMakeLists.txt
 include  FindNumPy.cmake
 include  FindPyLibs.cmake
@ -17,17 +16,14 @@ include  doc/source/cpplibrary.rst
 include  doc/source/index.rst
 include  doc/source/intro.rst
 include  doc/source/pythonmodule.rst
 include  external/config.guess
 include  external/config.sub
 include  external/external_build.cmake
-include  external/libsharp-8d51946.tar.gz
+include  external/libsharp-6077806.tar.gz
 include  external/omptl-20120422.tar.bz2
 include  external/patch-omptl
 include  python/CMakeLists.txt
 include  python/_cosmo_bispectrum.cpp
 include  python/_cosmo_cic.pyx
 include  python/_cosmo_power.pyx
 include  python/_cosmomath.pyx
 include  python/_cosmotool.pyx
 include  python/_fast_interp.pyx
 include  python/_project.pyx
@ -90,14 +86,12 @@ include  sample/testSmooth.cpp
 include  sample/test_cosmopower.cpp
 include  sample/test_fft_calls.cpp
 include  sample/test_healpix_calls.cpp
 include  sample/test_special.cpp
 include  sample/testkd.cpp
 include  sample/testkd2.cpp
 include  sample/testkd3.cpp
 include  setup.py
 include  src/CMakeLists.txt
 include  src/algo.hpp
 include  src/numpy_adaptors.hpp
 include  src/bqueue.hpp
 include  src/bqueue.tcc
 include  src/bsp_simple.hpp
@ -173,4 +167,3 @@ include  src/tf_fit.hpp
 include  src/yorick.hpp
 include  src/yorick_nc3.cpp
 include  src/yorick_nc4.cpp
 include  src/special_math.hpp
--- a/builder/build-wheels.sh
+++ b/builder/build-wheels.sh
@ -8,32 +8,28 @@ export CC CXX
 # Install a system package required by our library
 #yum install -y atlas-devel
-yum install -y cmake3 gsl-devel zlib-devel fftw3-devel libffi-devel hdf5 hdf5-devel
+yum install -y cmake3 gsl-devel zlib-devel
 ln -fs /usr/bin/cmake3 /usr/bin/cmake
-test -d /io/wheelhouse || mkdir /io/wheelhouse
+ALL_PYTHON="cp36-cp36m cp37-cp37m cp38-cp38" 
 test -d /io/wheelhouse/fix || mkdir /io/wheelhouse/fix
 ALL_PYTHON="cp39-cp39 cp310-cp310"
 # Compile wheels
 for pkg in $ALL_PYTHON; do
    PYBIN=/opt/python/${pkg}/bin
 #    "${PYBIN}/pip" install -r /io/dev-requirements.txt
    "${PYBIN}/pip" install setuptools wheel Cython 
    "${PYBIN}/pip" install -r /io/requirements.txt
-    "${PYBIN}/pip" wheel -vvv /io/ -w /io/wheelhouse/
+    "${PYBIN}/pip" wheel -vvv /io/ -w wheelhouse/
 done
 # Bundle external shared libraries into the wheels
-for whl in /io/wheelhouse/cosmotool*linux*.whl; do
+for whl in wheelhouse/*.whl; do
-    auditwheel repair "$whl" --plat $PLAT -w /io/wheelhouse/fix
+    auditwheel repair "$whl" --plat $PLAT -w /io/wheelhouse/
 done
 # Install packages and test
-#for pkg in $ALL_PYTHON; do
+for pkg in $ALL_PYTHON; do
-#    PYBIN=/opt/python/${pkg}/bin
+    PYBIN=/opt/python/${pkg}/bin
-#    "${PYBIN}/pip" install cosmotool --no-index -f /io/wheelhouse
+    "${PYBIN}/pip" install cosmotool --no-index -f /io/wheelhouse
-#done
+done
--- a/builder/start.sh
+++ b/builder/start.sh
@ -9,4 +9,4 @@ if ! [ -e ${d}/setup.py ] ; then
  exit 1
 fi
-podman run -ti  --rm -e PLAT=manylinux2014_x86_64 -v ${d}:/io:Z quay.io/pypa/manylinux2014_x86_64 /io/builder/build-wheels.sh
+podman run -ti  --rm -e PLAT=manylinux2010_x86_64 -v ${d}:/io:Z quay.io/pypa/manylinux2010_x86_64 /io/builder/build-wheels.sh
--- a/conda/build.sh
+++ b/conda/build.sh
@ -1,4 +1 @@
 export CC=$(basename ${CC})
 export CXX=$(basename ${CXX})
 $PYTHON setup.py install
--- a/conda/conda_build_config.yaml
+++ b/conda/conda_build_config.yaml
@ -1,10 +1,7 @@
 python:
  - 3.9
  - 3.8
  - 3.7
  - 3.8
 numpy:
  - 1.11
  - 1.19
 gsl:
  - 2.4
--- a/conda/meta.yaml
+++ b/conda/meta.yaml
@ -1,50 +1,38 @@
 package:
  name: cosmotool
-  version: "1.3.0"
+  version: "1.0.0a7"
 source:
-  git_rev: a86c9a8
+  git_rev: 7fce73e
  git_url: https://bitbucket.org/glavaux/cosmotool
 requirements:
  build:
-    - python                                 # [build_platform != target_platform]
+    - numpy >=1.11
-    - cross-python_{{ target_platform }}     # [build_platform != target_platform]
+    - gcc_linux-64
-    - cython                                 # [build_platform != target_platform]
+    - gxx_linux-64
    - numpy                                  # [build_platform != target_platform]
    - {{ compiler('c') }}
    - {{ compiler('cxx') }}
    - llvm-openmp                            # [osx]
    - libgomp                                # [linux]
  host:
    - python
-    - pip
+    - setuptools
    - numpy
    - pkgconfig
    - numexpr
    - cython
    - healpy
    - numexpr
    - cffi
    - pyfftw
-    - gsl {{ gsl }}
+    - gsl
    - h5py
  run:
    - numpy
    - python
    - {{ pin_compatible('numpy') }}
    - healpy
    - numexpr
    - cffi
    - pyfftw
    - h5py
    - {{ pin_compatible('gsl') }}
    - llvm-openmp
 test:
  imports:
    - cosmotool
  requires:
    - pip
 about:
  home: https://bitbucket.org/glavaux/cosmotool
--- a/external/config.guess
+++ b/external/config.guess
--- a/external/config.sub
+++ b/external/config.sub
--- a/external/external_build.cmake
+++ b/external/external_build.cmake
@ -2,19 +2,16 @@ include(FindOpenMP)
 OPTION(ENABLE_OPENMP "Set to Yes if Healpix and/or you need openMP" OFF)
 SET(SOURCE_PREFIX ${CMAKE_SOURCE_DIR})
 SET(FFTW_URL "http://www.fftw.org/fftw-3.3.3.tar.gz" CACHE STRING "URL to download FFTW from")
 SET(EIGEN_URL "https://gitlab.com/libeigen/eigen/-/archive/3.3.7/eigen-3.3.7.tar.bz2" CACHE STRING "URL to download Eigen from")
 SET(GENGETOPT_URL "ftp://ftp.gnu.org/gnu/gengetopt/gengetopt-2.22.5.tar.gz" CACHE STRING "URL to download gengetopt from")
-SET(HDF5_URL "https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-1.12/hdf5-1.12.2/src/hdf5-1.12.2.tar.gz" CACHE STRING "URL to download HDF5 from")
+SET(HDF5_URL "https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-1.8/hdf5-1.8.18/src/hdf5-1.8.18.tar.bz2" CACHE STRING "URL to download HDF5 from")
-SET(NETCDF_URL "https://downloads.unidata.ucar.edu/netcdf-c/4.9.2/netcdf-c-4.9.2.tar.gz" CACHE STRING "URL to download NetCDF from")
+SET(NETCDF_URL "ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf-4.5.0.tar.gz" CACHE STRING "URL to download NetCDF from")
-SET(NETCDFCXX_URL "https://github.com/Unidata/netcdf-cxx4/archive/v4.3.1.tar.gz" CACHE STRING "URL to download NetCDF-C++ from")
+SET(NETCDFCXX_URL "https://github.com/Unidata/netcdf-cxx4/archive/v4.3.0.tar.gz" CACHE STRING "URL to download NetCDF-C++ from")
-SET(BOOST_URL "https://boostorg.jfrog.io/artifactory/main/release/1.82.0/source/boost_1_82_0.tar.gz" CACHE STRING "URL to download Boost from")
+SET(BOOST_URL "http://sourceforge.net/projects/boost/files/boost/1.61.0/boost_1_61_0.tar.gz/download" CACHE STRING "URL to download Boost from")
-SET(GSL_URL "https://ftpmirror.gnu.org/gsl/gsl-2.7.tar.gz" CACHE STRING "URL to download GSL from ")
+SET(GSL_URL "ftp://ftp.gnu.org/gnu/gsl/gsl-1.15.tar.gz" CACHE STRING "URL to download GSL from ")
 mark_as_advanced(FFTW_URL EIGEN_URL HDF5_URL NETCDF_URL BOOST_URL GSL_URL)
 file(MAKE_DIRECTORY ${SOURCE_PREFIX}/downloads)
 SET(all_deps)
 MACRO(CHECK_CHANGE_STATE VAR) 
@ -39,13 +36,12 @@ CHECK_CHANGE_STATE(INTERNAL_DLIB DLIB_INCLUDE_DIR DLIB_LIBRARIES)
 IF(ENABLE_OPENMP)
  IF (NOT OPENMP_FOUND)
-    MESSAGE(NOTICE "No known compiler option for enabling OpenMP")
+    MESSAGE(ERROR "No known compiler option for enabling OpenMP")
  ELSE()
    SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${OpenMP_C_FLAGS}")
    SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}")
    SET(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${OpenMP_C_FLAGS}")
  ENDIF(NOT OPENMP_FOUND)
  SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${OpenMP_C_FLAGS}")
  SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}")
  SET(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} ${OpenMP_C_FLAGS}")
 ENDIF(ENABLE_OPENMP)
@ -59,25 +55,17 @@ if (ENABLE_SHARP)
  SET(DEP_BUILD ${BUILD_PREFIX}/sharp-prefix/src/sharp/auto)
  IF(NOT ENABLE_OPENMP)
    SET(SHARP_OPENMP --disable-openmp)
  ELSE()
    SET(SHARP_OPENMP)
  ENDIF()
  SET(CUTILS_LIBRARY ${DEP_BUILD}/lib/libc_utils.a)
  SET(FFTPACK_LIBRARY ${DEP_BUILD}/lib/libfftpack.a)
  SET(SHARP_LIBRARY ${DEP_BUILD}/lib/libsharp.a)
  SET(SHARP_LIBRARIES ${SHARP_LIBRARY} ${FFTPACK_LIBRARY} ${CUTILS_LIBRARY})
  SET(SHARP_INCLUDE_PATH ${DEP_BUILD}/include)
  message(STATUS "Flags: ${CMAKE_C_FLAGS}")
  ExternalProject_Add(sharp
-    URL ${CMAKE_SOURCE_DIR}/external/libsharp-8d51946.tar.gz
+    URL ${CMAKE_SOURCE_DIR}/external/libsharp-6077806.tar.gz
    PREFIX ${BUILD_PREFIX}/sharp-prefix
    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
    BUILD_IN_SOURCE 1 
-    CONFIGURE_COMMAND
+    CONFIGURE_COMMAND autoconf && ./configure "CC=${CMAKE_C_COMPILER}" "CXX=${CMAKE_CXX_COMPILER}" --prefix=${DEP_BUILD} ${SHARP_OPENMP}
          cp -f ${CMAKE_SOURCE_DIR}/external/config.guess . &&
          cp -f ${CMAKE_SOURCE_DIR}/external/config.sub . &&
          autoconf &&
          ./configure "CC=${CMAKE_C_COMPILER}" "CXX=${CMAKE_CXX_COMPILER}" "CFLAGS=${CMAKE_C_FLAGS}" "LDFLAGS=${CMAKE_EXE_LINKER_FLAGS}" --prefix=${DEP_BUILD} ${SHARP_OPENMP}
    INSTALL_COMMAND echo "No install"
    BUILD_BYPRODUCTS ${SHARP_LIBRARIES}
  )
@ -94,23 +82,27 @@ if (INTERNAL_HDF5)
  ExternalProject_Add(hdf5
    PREFIX ${BUILD_PREFIX}/hdf5-prefix
    URL ${HDF5_URL}
-    URL_HASH MD5=30172c75e436d7f2180e274071a4ca97
+    URL_HASH MD5=29117bf488887f89888f9304c8ebea0b
-    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
+    CMAKE_ARGS
-    CONFIGURE_COMMAND
+       -DCMAKE_INSTALL_PREFIX=${EXT_INSTALL}
-        ${HDF5_SOURCE_DIR}/configure "CC=${CMAKE_C_COMPILER}" "CXX=${CMAKE_CXX_COMPILER}" "CFLAGS=${CMAKE_C_FLAGS}" "LDFLAGS=${CMAKE_EXE_LINKER_FLAGS}" --prefix=${HDF5_BIN_DIR} --disable-shared  --enable-cxx --enable-hl --enable-tools --with-pic
+       -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
-    INSTALL_COMMAND make install
+       -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
       -DHDF5_BUILD_CPP_LIB=ON
       -DHDF5_BUILD_TOOLS=ON
       -DHDF5_BUILD_HL_LIB=ON
       -DBUILD_SHARED_LIBS=OFF
  )
  SET(cosmotool_DEPS ${cosmotool_DEPS} hdf5)
  SET(hdf5_built hdf5)
  SET(ENV{HDF5_ROOT} ${HDF5_BIN_DIR})
  SET(HDF5_ROOTDIR ${HDF5_BIN_DIR})
  SET(CONFIGURE_LDFLAGS "${CONFIGURE_LDFLAGS} -L${HDF5_BIN_DIR}/lib")
-  SET(CONFIGURE_LIBS "${CONFIGURE_LIBS} -ldl ${RT_LIBRARY}")
+  SET(CONFIGURE_LIBS "${CONFIGURE_LIBS} -ldl")
-  set(HDF5_C_STATIC_LIBRARY ${HDF5_BIN_DIR}/lib/libhdf5.a)
+  set(HDF5_C_STATIC_LIBRARY ${HDF5_BIN_DIR}/lib/libhdf5-static.a)
-  set(HDF5_HL_STATIC_LIBRARY ${HDF5_BIN_DIR}/lib/libhdf5_hl.a)
+  set(HDF5_HL_STATIC_LIBRARY ${HDF5_BIN_DIR}/lib/libhdf5_hl-static.a)
-  set(HDF5_LIBRARIES ${HDF5_BIN_DIR}/lib/libhdf5.a CACHE STRING "HDF5 lib" FORCE)
+  set(HDF5_LIBRARIES ${HDF5_BIN_DIR}/lib/libhdf5-static.a CACHE STRING "HDF5 lib" FORCE)
-  set(HDF5_HL_LIBRARIES ${HDF5_BIN_DIR}/lib/libhdf5_hl.a CACHE STRING "HDF5 HL lib" FORCE)
+  set(HDF5_HL_LIBRARIES ${HDF5_BIN_DIR}/lib/libhdf5_hl-static.a CACHE STRING "HDF5 HL lib" FORCE)
-  set(HDF5_CXX_LIBRARIES ${HDF5_BIN_DIR}/lib/libhdf5_cpp.a CACHE STRING "HDF5 C++ lib" FORCE)
+  set(HDF5_CXX_LIBRARIES ${HDF5_BIN_DIR}/lib/libhdf5_cpp-static.a CACHE STRING "HDF5 C++ lib" FORCE)
  SET(HDF5_INCLUDE_DIRS ${HDF5_BIN_DIR}/include CACHE STRING "HDF5 include path" FORCE)
  mark_as_advanced(HDF5_LIBRARIES HDF5_CXX_LIBRARIES HDF5_INCLUDE_DIRS)
@ -168,52 +160,43 @@ if (INTERNAL_NETCDF)
  SET(NETCDF_CONFIG_COMMAND ${NETCDF_SOURCE_DIR}/configure
         --prefix=${NETCDF_BIN_DIR} --libdir=${NETCDF_BIN_DIR}/lib
         --enable-netcdf-4  --with-pic --disable-shared --disable-dap 
-         --disable-byterange --disable-cdmremote --disable-rpc --enable-cxx-4 
+         --disable-cdmremote --disable-rpc --enable-cxx-4 
         --disable-examples ${EXTRA_NC_FLAGS} CC=${CMAKE_C_COMPILER}
         CXX=${CMAKE_CXX_COMPILER})
  list(INSERT CMAKE_PREFIX_PATH 0 ${EXT_INSTALL})
  string(REPLACE ";" "|" CMAKE_PREFIX_PATH_ALT_SEP "${CMAKE_PREFIX_PATH}")
  ExternalProject_Add(netcdf
    DEPENDS ${hdf5_built}
    URL_HASH MD5=f48ee01534365006934f0c63d4055ea0
    PREFIX ${BUILD_PREFIX}/netcdf-prefix
    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
    URL ${NETCDF_URL}
-    LIST_SEPARATOR |
+  LIST_SEPARATOR |
-    CMAKE_ARGS
+	CMAKE_ARGS
 		-DCMAKE_PREFIX_PATH=${CMAKE_PREFIX_PATH_ALT_SEP}
 		-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
 		-DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
                -DNC_EXTRA_DEPS=${RT_DEP}
 		-DBUILD_SHARED_LIBS=OFF
 		-DBUILD_TESTING=OFF
 		-DCMAKE_BUILD_TYPE=Release
 		-DENABLE_NETCDF4=ON
-                -DENABLE_BYTERANGE=FALSE
+    -DCMAKE_POSITION_INDEPENDENT_CODE=ON
-		-DCMAKE_POSITION_INDEPENDENT_CODE=ON
+    -DENABLE_DAP=OFF
 		-DENABLE_DAP=OFF
 		-DCMAKE_INSTALL_PREFIX=${NETCDF_BIN_DIR}
                -DHDF5_C_LIBRARY=${HDF5_C_STATIC_LIBRARY}
                -DHDF5_HL_LIBRARY=${HDF5_HL_STATIC_LIBRARY}
                -DHDF5_INCLUDE_DIR=${HDF5_INCLUDE_DIRS}
-		-DCMAKE_INSTALL_LIBDIR=lib
+    -DCMAKE_INSTALL_LIBDIR=lib
  )
  SET(NETCDFCXX_SOURCE_DIR ${BUILD_PREFIX}/netcdf-c++-prefix/src/netcdf-c++)
  ExternalProject_Add(netcdf-c++
    DEPENDS ${hdf5_built} netcdf
    PREFIX ${BUILD_PREFIX}/netcdf-c++-prefix
    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
    URL ${NETCDFCXX_URL}
 	CMAKE_ARGS
 		-DCMAKE_PREFIX_PATH=${CMAKE_PREFIX_PATH_ALT_SEP}
 		-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
 		-DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
 		-DBUILD_SHARED_LIBS=OFF
-                -DCMAKE_POSITION_INDEPENDENT_CODE=ON
+    -DCMAKE_POSITION_INDEPENDENT_CODE=ON
                -DHDF5_C_LIBRARY=${HDF5_C_STATIC_LIBRARY}
                -DHDF5_HL_LIBRARY=${HDF5_HL_STATIC_LIBRARY}
                -DHDF5_INCLUDE_DIR=${HDF5_INCLUDE_DIRS}
 		-DBUILD_TESTING=OFF
 		-DCMAKE_BUILD_TYPE=Release
 		-DCMAKE_INSTALL_PREFIX=${NETCDF_BIN_DIR}
@ -251,8 +234,6 @@ if (INTERNAL_BOOST)
  ExternalProject_Add(boost
    URL ${BOOST_URL}
    PREFIX ${BUILD_PREFIX}/boost-prefix
    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
    URL_HASH MD5=f7050f554a65f6a42ece221eaeec1660
    CONFIGURE_COMMAND 
           ${BOOST_SOURCE_DIR}/bootstrap.sh --prefix=${CMAKE_BINARY_DIR}/ext_build/boost
    BUILD_IN_SOURCE 1
@ -295,7 +276,6 @@ IF(INTERNAL_GSL)
  ExternalProject_Add(gsl
    URL ${GSL_URL}
    PREFIX ${BUILD_PREFIX}/gsl-prefix
    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
    CONFIGURE_COMMAND ${GSL_SOURCE_DIR}/configure
           --prefix=${EXT_INSTALL} --disable-shared
           --with-pic
@ -348,7 +328,6 @@ IF(INTERNAL_FFTW)
  ExternalProject_Add(fftw
     URL ${FFTW_URL}
     PREFIX ${BUILD_PREFIX}/fftw-prefix
    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
     CONFIGURE_COMMAND 
           ${FFTW_SOURCE}/configure
                 --prefix=${EXT_INSTALL}
@ -378,7 +357,6 @@ IF (INTERNAL_EIGEN)
  ExternalProject_Add(eigen
     URL ${EIGEN_URL}
     URL_HASH MD5=b9e98a200d2455f06db9c661c5610496
    DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
     PREFIX ${BUILD_PREFIX}/eigen-prefix
     CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${EXT_INSTALL}
       -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
@ -411,7 +389,6 @@ SET(cosmotool_DEPS ${cosmotool_DEPS} omptl)
 SET(OMPTL_BUILD_DIR ${BUILD_PREFIX}/omptl-prefix/src/omptl)
 ExternalProject_Add(omptl
  PREFIX ${BUILD_PREFIX}/omptl-prefix
  DOWNLOAD_DIR ${SOURCE_PREFIX}/downloads
  URL ${CMAKE_SOURCE_DIR}/external/omptl-20120422.tar.bz2
  CONFIGURE_COMMAND echo "No configure"
  BUILD_COMMAND echo "No build"
--- a/external/libsharp-8d51946.tar.gz
+++ b/external/libsharp-8d51946.tar.gz
--- a/external/patch-omptl
+++ b/external/patch-omptl
@ -1,6 +1,6 @@
-diff -ur omptl.old/omptl_algorithm omptl/omptl_algorithm
+diff -ur omptl.orig/omptl_algorithm omptl/omptl_algorithm
--- omptl.old/omptl_algorithm	2022-06-19 08:21:39.815498672 +0200
+--- omptl.orig/omptl_algorithm	2017-01-16 14:58:37.996690639 +0100
-+++ omptl/omptl_algorithm	2022-06-19 08:21:52.953544672 +0200
+++ omptl/omptl_algorithm	2017-01-16 15:00:26.678641720 +0100
@@ -20,7 +20,7 @@
 #define OMPTL_ALGORITHM 1
@ -22,14 +22,12 @@ diff -ur omptl.old/omptl_algorithm omptl/omptl_algorithm
 #endif
 #endif /* OMPTL_ALGORITHM */
-diff -ur omptl.old/omptl_algorithm_par.h omptl/omptl_algorithm_par.h
+diff -ur omptl.orig/omptl_algorithm_par.h omptl/omptl_algorithm_par.h
--- omptl.old/omptl_algorithm_par.h	2022-06-19 08:21:39.816498675 +0200
+--- omptl.orig/omptl_algorithm_par.h	2017-01-16 14:58:37.996690639 +0100
-+++ omptl/omptl_algorithm_par.h	2022-06-19 08:23:50.705956964 +0200
+++ omptl/omptl_algorithm_par.h	2017-01-16 14:59:57.974126410 +0100
-@@ -20,9 +20,10 @@
+@@ -21,8 +21,8 @@
 #include <utility>
 #include <cmath>
 #include <cstdlib>
 +#include <random>
 -#include <omptl/omptl_tools.h>
 -#include <omptl/omptl_numeric>
@ -38,136 +36,9 @@ diff -ur omptl.old/omptl_algorithm_par.h omptl/omptl_algorithm_par.h
 #include <iterator>
-@@ -510,7 +511,7 @@
+diff -ur omptl.orig/omptl_numeric omptl/omptl_numeric
- 
+--- omptl.orig/omptl_numeric	2017-01-16 14:58:37.996690639 +0100
- 		typename std::vector<Iterator>::iterator result =
+++ omptl/omptl_numeric	2017-01-16 15:00:57.051186974 +0100
 		    std::find_if(results.begin(),results.end(),
 -			std::bind2nd(std::not_equal_to<Iterator>(), last) );
 +			std::bind(std::not_equal_to<Iterator>(), std::placeholders::_1, last) );
 		if ( result != results.end() )
 			return *result;
@@ -569,7 +570,7 @@
 		const typename std::vector<Iterator>::iterator result
 			 = std::find_if(results.begin(), results.end(),
 -			std::bind2nd(std::not_equal_to<Iterator>(), last) );
 +			std::bind(std::not_equal_to<Iterator>(), std::placeholders::_1, last) );
 		if ( result != results.end() )
 			return *result;
@@ -654,7 +655,7 @@
 		const typename std::vector<Iterator>::iterator
 		result = std::find_if(results.begin(), results.end(),
 -					std::bind2nd(std::not_equal_to<Iterator>(), last1));
 +					std::bind(std::not_equal_to<Iterator>(), std::placeholders::_1, last1));
 		if ( result != results.end() )
 			return *result;
@@ -953,7 +954,7 @@
 		results[t] = std::lower_bound(partitions[t].first, partitions[t].second, value, comp);
 	const typename std::vector<ForwardIterator>::iterator result =
 -		std::find_if(results.begin(), results.end(), std::bind2nd(std::not_equal_to<ForwardIterator>(), last) );
 +		std::find_if(results.begin(), results.end(), std::bind(std::not_equal_to<ForwardIterator>(), std::placeholders::_1, last) );
 	if (result != results.end())
 		return *result;
@@ -1179,7 +1180,7 @@
 namespace detail
 {
 -	
 +
 template<typename Iterator, class StrictWeakOrdering>
 Iterator _pivot_range(Iterator first, Iterator last,
 	const typename std::iterator_traits<Iterator>::value_type pivot,
@@ -1309,14 +1310,14 @@
 void random_shuffle(RandomAccessIterator first, RandomAccessIterator last,
                     const unsigned P)
 {
 -	std::random_shuffle(first, last);
 +	std::shuffle(first, last, std::mt19937(std::random_device()()));
 }
 template <class RandomAccessIterator, class RandomNumberGenerator>
 void random_shuffle(RandomAccessIterator first, RandomAccessIterator last,
                     RandomNumberGenerator& rgen, const unsigned P)
 {
 -	std::random_shuffle(first, last, rgen);
 +	std::shuffle(first, last, std::mt19937(std::random_device()()));
 }
 // Not (yet) parallelized, not straightforward due to possible dependencies
@@ -1472,7 +1473,7 @@
                             const T& new_value, const unsigned P)
 {
 	return ::omptl::replace_copy_if(first, last, result,
 -			std::bind2nd(std::equal_to<T>(), old_value), new_value, P);
 +			std::bind(std::equal_to<T>(), std::placeholders::_1, old_value), new_value, P);
 }
 template <class ForwardIterator, class Predicate, class T>
@@ -1700,7 +1701,7 @@
 	std::vector<char> pivot_used(pivots.size(), false); // can't be bool due to parallel write
 -	const unsigned max_depth = std::floor(std::tr1::log2(P));
 +	const unsigned max_depth = unsigned(std::floor(std::log2(P)));
 	assert(1u << max_depth <= P);
 	for (unsigned i = 0; i < max_depth; ++i)
 	{
@@ -1781,7 +1782,7 @@
 std::cout  << std::endl;
 std::cout << borders.size() << " " << partitions.size() << " " << P << std::endl;
 -*/	
 +*/
 	// Round one: sort final partitions, split remaining
 	#pragma omp parallel for
 	for (int i = 0; i < int(partitions.size()); ++i)
@@ -1814,7 +1815,7 @@
 			const RandomAccessIterator begin  = partitions[i].first;
 			const RandomAccessIterator end    = partitions[i].second;
 -			
 +
 			const RandomAccessIterator middle =
 				detail::_pivot_range(begin, end, pivots[pivot_index], comp);
 			partitions[i  ] = std::make_pair(begin, middle);
 diff -ur omptl.old/omptl_algorithm_ser.h omptl/omptl_algorithm_ser.h
 --- omptl.old/omptl_algorithm_ser.h	2022-06-19 08:21:39.815498672 +0200
 +++ omptl/omptl_algorithm_ser.h	2022-06-19 08:21:52.960544697 +0200
@@ -14,7 +14,7 @@
 // You should have received a copy of the GNU Lesser General Public
 // License along with this library; if not, write to the Free Software
 // Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 -
 +#include <random>
 namespace omptl
 {
@@ -463,14 +463,14 @@
 void random_shuffle(RandomAccessIterator first, RandomAccessIterator last,
       const unsigned P)
 {
 -	return ::std::random_shuffle(first, last);
 +	return ::std::shuffle(first, last, std::mt19937(std::random_device()()));
 }
 template <class RandomAccessIterator, class RandomNumberGenerator>
 void random_shuffle(RandomAccessIterator first, RandomAccessIterator last,
                     RandomNumberGenerator &rgen, const unsigned P)
 {
 -	return ::std::random_shuffle(first, last, rgen);
 +	return ::std::shuffle(first, last, std::mt19937(std::random_device()()));
 }
 template <class ForwardIterator, class T>
 diff -ur omptl.old/omptl_numeric omptl/omptl_numeric
 --- omptl.old/omptl_numeric	2022-06-19 08:21:39.816498675 +0200
 +++ omptl/omptl_numeric	2022-06-19 08:21:52.955544679 +0200
@@ -19,7 +19,7 @@
 #define OMPTL_NUMERIC 1
@ -192,9 +63,9 @@ diff -ur omptl.old/omptl_numeric omptl/omptl_numeric
 +#include "omptl_numeric_extensions.h"
 #endif /* OMPTL_NUMERIC */
-diff -ur omptl.old/omptl_numeric_extensions.h omptl/omptl_numeric_extensions.h
+diff -ur omptl.orig/omptl_numeric_extensions.h omptl/omptl_numeric_extensions.h
--- omptl.old/omptl_numeric_extensions.h	2022-06-19 08:21:39.815498672 +0200
+--- omptl.orig/omptl_numeric_extensions.h	2017-01-16 14:58:37.996690639 +0100
-+++ omptl/omptl_numeric_extensions.h	2022-06-19 08:21:52.956544683 +0200
+++ omptl/omptl_numeric_extensions.h	2017-01-16 14:59:21.549472508 +0100
@@ -51,9 +51,9 @@
 } // namespace
@ -207,9 +78,9 @@ diff -ur omptl.old/omptl_numeric_extensions.h omptl/omptl_numeric_extensions.h
 #endif
 namespace omptl
-diff -ur omptl.old/omptl_numeric_par.h omptl/omptl_numeric_par.h
+diff -ur omptl.orig/omptl_numeric_par.h omptl/omptl_numeric_par.h
--- omptl.old/omptl_numeric_par.h	2022-06-19 08:21:39.816498675 +0200
+--- omptl.orig/omptl_numeric_par.h	2017-01-16 14:58:37.996690639 +0100
-+++ omptl/omptl_numeric_par.h	2022-06-19 08:21:52.957544686 +0200
+++ omptl/omptl_numeric_par.h	2017-01-16 14:59:36.397739066 +0100
@@ -23,8 +23,8 @@
 #include <functional>
 #include <iterator>
@ -221,15 +92,3 @@ diff -ur omptl.old/omptl_numeric_par.h omptl/omptl_numeric_par.h
 namespace omptl
 {
 diff -ur omptl.old/omptl_tools.h omptl/omptl_tools.h
 --- omptl.old/omptl_tools.h	2022-06-19 08:21:39.816498675 +0200
 +++ omptl/omptl_tools.h	2022-06-19 08:21:52.957544686 +0200
@@ -25,7 +25,7 @@
 #include <climits>
 #include <iterator>
 -#include <tr1/cmath>
 +#include <cmath>
 namespace omptl
 {
--- a/pyproject.toml
+++ b/pyproject.toml
@ -1,4 +0,0 @@
 [build-system]
 requires = ["setuptools","wheel","cython"]
 build-backend = "setuptools.build_meta"
--- a/python/CMakeLists.txt
+++ b/python/CMakeLists.txt
@ -2,38 +2,38 @@ set(CMAKE_SHARED_MODULE_PREFIX)
 set(PYTHON_INCLUDES ${NUMPY_INCLUDE_DIRS} ${PYTHON_INCLUDE_PATH} ${CMAKE_SOURCE_DIR}/python)
-include_directories(${CMAKE_SOURCE_DIR}/python ${CMAKE_SOURCE_DIR}/src ${CMAKE_BINARY_DIR}/src)
+include_directories(${CMAKE_SOURCE_DIR}/src ${CMAKE_BINARY_DIR}/src)
 IF(CYTHON)
  add_custom_command(
    OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/_cosmotool.cpp
    COMMAND ${CYTHON} --cplus -o ${CMAKE_CURRENT_BINARY_DIR}/_cosmotool.cpp ${CMAKE_CURRENT_SOURCE_DIR}/_cosmotool.pyx
-    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmotool.pyx)
+    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmotool.pyx) 
  add_custom_command(
    OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/_cosmo_power.cpp
    COMMAND ${CYTHON} --cplus -o ${CMAKE_CURRENT_BINARY_DIR}/_cosmo_power.cpp ${CMAKE_CURRENT_SOURCE_DIR}/_cosmo_power.pyx
-    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmo_power.pyx)
+    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmo_power.pyx) 
  add_custom_command(
    OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/_fast_interp.cpp
    COMMAND ${CYTHON} --cplus -o ${CMAKE_CURRENT_BINARY_DIR}/_fast_interp.cpp ${CMAKE_CURRENT_SOURCE_DIR}/_fast_interp.pyx
-    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_fast_interp.pyx)
+    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_fast_interp.pyx) 
  add_custom_command(
    OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/_cosmo_cic.cpp
    COMMAND ${CYTHON} --cplus -o ${CMAKE_CURRENT_BINARY_DIR}/_cosmo_cic.cpp ${CMAKE_CURRENT_SOURCE_DIR}/_cosmo_cic.pyx
-    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmo_cic.pyx)
+    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmo_cic.pyx) 
  add_custom_command(
    OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/_project.cpp
    COMMAND ${CYTHON} --cplus -o ${CMAKE_CURRENT_BINARY_DIR}/_project.cpp ${CMAKE_CURRENT_SOURCE_DIR}/_project.pyx
-    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_project.pyx ${CMAKE_CURRENT_SOURCE_DIR}/project_tool.hpp )
+    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_project.pyx ${CMAKE_CURRENT_SOURCE_DIR}/project_tool.hpp ) 
  add_custom_command(
    OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/_cosmomath.cpp
    COMMAND ${CYTHON} --cplus -o ${CMAKE_CURRENT_BINARY_DIR}/_cosmomath.cpp ${CMAKE_CURRENT_SOURCE_DIR}/_cosmomath.pyx
-    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmomath.pyx )
+    DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/_cosmomath.pyx ) 
 ENDIF(CYTHON)
@ -43,21 +43,17 @@ add_library(_cosmo_cic MODULE ${CMAKE_CURRENT_BINARY_DIR}/_cosmo_cic.cpp)
 add_library(_fast_interp MODULE ${CMAKE_CURRENT_BINARY_DIR}/_fast_interp.cpp)
 add_library(_project MODULE ${CMAKE_CURRENT_BINARY_DIR}/_project.cpp)
 add_library(_cosmomath MODULE ${CMAKE_CURRENT_BINARY_DIR}/_cosmomath.cpp)
-target_include_directories(_cosmotool PRIVATE ${PYTHON_INCLUDES})
+target_include_directories(_cosmotool PRIVATE ${PYTHON_INCLUDES}) 
-target_include_directories(_cosmo_power PRIVATE ${PYTHON_INCLUDES})
+target_include_directories(_cosmo_power PRIVATE ${PYTHON_INCLUDES}) 
-target_include_directories(_cosmo_cic PRIVATE ${PYTHON_INCLUDES})
+target_include_directories(_cosmo_cic PRIVATE ${PYTHON_INCLUDES}) 
-target_include_directories(_fast_interp PRIVATE ${PYTHON_INCLUDES})
+target_include_directories(_fast_interp PRIVATE ${PYTHON_INCLUDES}) 
-target_include_directories(_project PRIVATE ${PYTHON_INCLUDES})
+target_include_directories(_project PRIVATE ${PYTHON_INCLUDES}) 
 target_include_directories(_cosmomath PRIVATE ${PYTHON_INCLUDES})
 SET(CMAKE_MODULE_LINKER_FLAGS "${CMAKE_MODULE_LINKER_FLAGS} -Bsymbolic-functions")
 if(APPLE)
  set(CMAKE_MODULE_LINKER_FLAGS "-undefined dynamic_lookup")
 endif()
 IF(NOT APPLE)
  set(CMAKE_MODULE_LINKER_FLAGS "-Wl,--version-script=${CMAKE_CURRENT_SOURCE_DIR}/cosmotool.version")
 ENDIF()
 target_link_libraries(_cosmotool PRIVATE ${CosmoTool_local}  ${GSL_LIBRARIES})
 target_link_libraries(_cosmo_power PRIVATE ${CosmoTool_local} ${GSL_LIBRARIES})
@ -69,10 +65,13 @@ target_link_libraries(_fast_interp PRIVATE ${CosmoTool_local} )
 SET(ct_TARGETS   _cosmotool _project _cosmo_power _cosmo_cic _fast_interp _cosmomath)
 if (Boost_FOUND)
-  message(STATUS "Building bispectrum support")
+  message(STATUS "Building bispectrum support (path = ${Boost_INCLUDE_DIRS})")
  include_directories(${Boost_INCLUDE_DIRS})
  add_library(_cosmo_bispectrum MODULE _cosmo_bispectrum.cpp)
  target_link_libraries(_cosmo_bispectrum ${MATH_LIBRARY})
  if(ENABLE_OPENMP)
    set_target_properties(_cosmo_bispectrum PROPERTIES COMPILE_FLAGS "${OpenMP_CXX_FLAGS}" LINK_FLAGS "${OpenMP_CXX_FLAGS}")
  endif()
  if (Boost_DEP)
    add_dependencies(_cosmo_bispectrum ${Boost_DEP})
  endif()
@ -111,19 +110,10 @@ endif (WIN32 AND NOT CYGWIN)
 configure_file(${CMAKE_CURRENT_SOURCE_DIR}/cosmotool/config.py.in ${CMAKE_CURRENT_BINARY_DIR}/cosmotool/config.py @ONLY)
-INSTALL(TARGETS
+INSTALL(TARGETS 
  ${ct_TARGETS}
  COMPONENT python
  LIBRARY DESTINATION ${PYTHON_SITE_PACKAGES}/cosmotool
 )
-INSTALL(DIRECTORY cosmotool ${CMAKE_CURRENT_BINARY_DIR}/cosmotool
+INSTALL(DIRECTORY cosmotool ${CMAKE_CURRENT_BINARY_DIR}/cosmotool DESTINATION ${PYTHON_SITE_PACKAGES} 
        COMPONENT python DESTINATION ${PYTHON_SITE_PACKAGES}
        FILES_MATCHING PATTERN "*.py")
 add_custom_target(
    python-install
    DEPENDS ${ct_TARGETS}
    COMMAND "${CMAKE_COMMAND}" -DCMAKE_INSTALL_COMPONENT=python -P
            "${CMAKE_BINARY_DIR}/cmake_install.cmake")
--- a/python/_cosmomath.pyx
+++ b/python/_cosmomath.pyx
@ -1,4 +1,3 @@
 #cython: language_level=3
 import numpy as np
 cimport numpy as np
@ -8,21 +7,15 @@ np.import_ufunc()
 cdef extern from "sys/types.h":
   ctypedef np.int64_t int64_t
 cdef extern from "numpy/npy_common.h":
   ctypedef npy_intp
 cdef extern from "special_math.hpp" namespace "CosmoTool":
-   T log_modified_bessel_first_kind[T](T v, T z) except + nogil
+   T log_modified_bessel_first_kind[T](T v, T z) nogil except +
 cdef extern from "numpy_adaptors.hpp" namespace "CosmoTool":
   void parallel_ufunc_dd_d[T,IT](char **args, IT* dimensions, IT* steps, void *func)
 cdef np.PyUFuncGenericFunction loop_func[1]
 cdef char input_output_types[3]
 cdef void *elementwise_funcs[1]
-loop_func[0] = <np.PyUFuncGenericFunction>parallel_ufunc_dd_d[double,npy_intp]
+loop_func[0] = np.PyUFunc_dd_d
 input_output_types[0] = np.NPY_DOUBLE
 input_output_types[1] = np.NPY_DOUBLE
--- a/python/_cosmotool.pyx
+++ b/python/_cosmotool.pyx
@ -31,7 +31,7 @@ cdef extern from "loadSimu.hpp" namespace "CosmoTool":
    bool noAuto
-  cdef const int NEED_GADGET_ID
+  cdef const int NEED_GADGET_ID 
  cdef const int NEED_POSITION
  cdef const int NEED_VELOCITY
  cdef const int NEED_TYPE
@ -39,7 +39,7 @@ cdef extern from "loadSimu.hpp" namespace "CosmoTool":
 cdef extern from "loadGadget.hpp" namespace "CosmoTool":
-  SimuData *loadGadgetMulti(const char *fname, int id, int flags, int gformat) except + nogil
+  SimuData *loadGadgetMulti(const char *fname, int id, int flags, int gformat) nogil except +
  void cxx_writeGadget "CosmoTool::writeGadget" (const char * s, SimuData *data) except +
 cdef extern from "safe_gadget.hpp":
@ -57,36 +57,36 @@ class PySimulationBase(object):
  """
  This is the base class to representation Simulation in CosmoTool/python.
  """
-
+  
  def getPositions(self):
    """
    getPositions(self)
-
+    
    Returns:
-        A list of three arrays holding the positions of the particles.
+        A list of three arrays holding the positions of the particles. 
        The i-th element is the i-th coordinate of each particle.
        It may be None if the positions were not requested.
    """
    raise NotImplementedError("getPositions is not implemented")
-
+    
  def getVelocities(self):
    """
    getVelocities(self)
-
+    
    Returns:
-        A list of three arrays holding the velocities of the particles.
+        A list of three arrays holding the velocities of the particles. 
        The i-th element is the i-th coordinate of each particle.
        It may be None if the velocities were not requested.
    """
    raise NotImplementedError("getVelocities is not implemented")
-
+    
  def getIdentifiers(self):
    """
    getIdentifiers(self)
-
+    
    Returns:
        Returns an integer array that hold the unique identifiers of
-        each particle.
+        each particle. 
        It may be None if the identifiers were not requested.
    """
    raise NotImplementedError("getIdentifiers is not implemented")
@ -94,10 +94,10 @@ class PySimulationBase(object):
  def getTypes(self):
    """
    getTypes(self)
-
+    
    Returns:
        Returns an integer array that hold the type of
-        each particle.
+        each particle. 
        It may be None if the types were not requested.
    """
    raise NotImplementedError("getTypes is not implemented")
@ -105,27 +105,27 @@ class PySimulationBase(object):
  def getOmega_M(self):
    """
    getOmega_M(self)
-
+    
    Returns:
        the mean matter density in the simulation, with respect
        to the critical density.
-    """
+    """ 
    raise NotImplementedError("getOmega_M is not implemented")
-
+  
  def getOmega_Lambda(self):
    """
    getOmega_Lambda(self)
-
+    
    Returns:
        the mean dark energy density in the simulation, with respect
        to the critical density.
-    """
+    """ 
    raise NotImplementedError("getOmega_Lambda is not implemented")
  def getTime(self):
    """
    getTime(self)
-
+    
    Returns:
        the time the snapshot was taken in the simulation. It can
        have various units depending on the file format.
@ -135,7 +135,7 @@ class PySimulationBase(object):
  def getHubble(self):
    """
    getHubble(self)
-
+    
    Returns:
        the hubble constant in unit of 100 km/s/Mpc
    """
@ -144,7 +144,7 @@ class PySimulationBase(object):
  def getBoxsize(self):
    """
    getBoxsize(self)
-
+    
    Returns:
        the size of the simulation box. The length unit is not fixed,
        though it is customary to have it in Mpc/h if the loader has
@ -155,7 +155,7 @@ class PySimulationBase(object):
  def getMasses(self):
    """
    getMasses(self)
-
+    
    Returns:
        an array with the masses of each particles, in unspecified unit that
        depend on the loader.
@ -165,7 +165,7 @@ class PySimulationBase(object):
 cdef class Simulation:
    """
    Simulation()
-
+    
    Class that directly manages internal loaded data obtained from a loader
    """
@ -180,7 +180,7 @@ cdef class Simulation:
    property BoxSize:
        def __get__(Simulation self):
            return self.data.BoxSize
-
+        
    property time:
        def __get__(Simulation self):
            return self.data.time
@ -192,15 +192,15 @@ cdef class Simulation:
    property Omega_M:
        def __get__(Simulation self):
            return self.data.Omega_M
-
+        
    property Omega_Lambda:
        def __get__(Simulation self):
            return self.data.Omega_Lambda
-
+        
    property positions:
        def __get__(Simulation self):
            return self.positions
-
+        
    property velocities:
        def __get__(Simulation self):
            return self.velocities
@ -216,7 +216,7 @@ cdef class Simulation:
    property masses:
        def __get__(Simulation self):
            return self.masses
-
+        
    property numParticles:
        def __get__(Simulation self):
            return self.data.NumPart
@ -228,7 +228,7 @@ cdef class Simulation:
    def __cinit__(Simulation self):
        self.data = <SimuData *>0
-
+  
    def __dealloc__(Simulation self):
        if self.data != <SimuData *>0:
          del self.data
@ -237,43 +237,40 @@ cdef class Simulation:
 class PySimulationAdaptor(PySimulationBase):
  """
  PySimulationAdaptor(PySimulationBase_)
-
+  
  This class is an adaptor for an internal type to the loader. It defines
  all the methods of PySimulationBase.
-
+  
  Attributes:
    simu: a Simulation_ object
  """
  def __init__(self,sim):
    self.simu = sim
  def getNumParticles(self):
    return self.simu.numParticles
  def getBoxsize(self):
    return self.simu.BoxSize
-
+    
  def getPositions(self):
    return self.simu.positions
  def getTypes(self):
    return self.simu.types
-
+    
  def getVelocities(self):
    return self.simu.velocities
-
+    
  def getIdentifiers(self):
    return self.simu.identifiers
  def getTime(self):
    return self.simu.time
-
+    
  def getHubble(self):
    return self.simu.Hubble
-
+    
  def getOmega_M(self):
    return self.simu.Omega_M
-
+    
  def getOmega_Lambda(self):
    return self.simu.Omega_Lambda
@ -284,7 +281,7 @@ cdef class ArrayWrapper:
    cdef void* data_ptr
    cdef np.uint64_t size
    cdef int type_array
-
+    
    cdef set_data(self, np.uint64_t size, int type_array, void* data_ptr):
        """ Set the data of the array
            This cannot be done in the constructor as it must recieve C-level
@ -297,29 +294,22 @@ cdef class ArrayWrapper:
        self.data_ptr = data_ptr
        self.size = size
        self.type_array = type_array
-
+     
    def __array__(self):
        """ Here we use the __array__ method, that is called when numpy
 tries to get an array from the object."""
        cdef np.npy_intp shape[1]
-
+        
        shape[0] = <np.npy_intp> self.size
        # Create a 1D array, of length 'size'
        ndarray = np.PyArray_SimpleNewFromData(1, shape, self.type_array, self.data_ptr)
        return ndarray
-
+     
    def __dealloc__(self):
        """ Frees the array. This is called by Python when all the
 references to the object are gone. """
        pass
-
+ 
 cdef extern from "numpy/arrayobject.h":
    # a little bit awkward: the reference to obj will be stolen
    # using PyObject*  to signal that Cython cannot handle it automatically
    int PyArray_SetBaseObject(np.ndarray arr, PyObject *obj) except -1 # -1 means there was an error
 cdef object wrap_array(void *p, np.uint64_t s, int typ):
    cdef np.ndarray ndarray
    cdef ArrayWrapper wrapper
@ -327,10 +317,9 @@ cdef object wrap_array(void *p, np.uint64_t s, int typ):
    wrapper = ArrayWrapper()
    wrapper.set_data(s, typ, p)
    ndarray = np.array(wrapper, copy=False)
-    #ndarray.base = <PyObject*> wrapper
+    ndarray.base = <PyObject*> wrapper
    PyArray_SetBaseObject(ndarray, <PyObject*> wrapper)
    Py_INCREF(wrapper)
-
+    
    return ndarray
@ -379,7 +368,7 @@ def loadGadget(str filename, int snapshot_id, int gadgetFormat = 1, bool loadPos
    """loadGadget(filename, snapshot_id, gadgetFormat = 1, loadPosition=True, loadVelocity=False, loadId=False, loadType=False)
    This function loads Gadget-1 snapshot format.
-
+    
    If snapshot_id is negative then the snapshot is considered not to be part of
    a set of snapshots written by different cpu. Otherwise the filename is modified
    to reflect the indicated snapshot_id.
@ -387,16 +376,16 @@ def loadGadget(str filename, int snapshot_id, int gadgetFormat = 1, bool loadPos
    Arguments:
        filename (str): input filename
        snapshot_id (int): identifier of the gadget file if it is a multi-file snapshot
-
+    
    Keyword arguments:
        loadPosition (bool): whether to load positions
        loadVelocity (bool): whether to load velocities
        loadId (bool): whether to load unique identifiers
        loadType (bool): whether to set types to particles
        loadMass (bool): whether to set the mass of particles
-
+    
    Returns:
-        an PySimulationAdaptor instance.
+        an PySimulationAdaptor instance. 
 """
    cdef int flags
@ -421,7 +410,7 @@ def loadGadget(str filename, int snapshot_id, int gadgetFormat = 1, bool loadPos
    with nogil:
      data = loadGadgetMulti(filename_bs, snapshot_id, flags, gadgetFormat)
    if data == <SimuData*>0:
-       raise RuntimeError("File could not be read")
+        return None
    return PySimulationAdaptor(wrap_simudata(data, flags))
@ -430,13 +419,13 @@ def loadParallelGadget(object filename_list, int gadgetFormat = 1, bool loadPosi
  Arguments:
    filename (list): a list or tuple of filenames to load in parallel
-
+  
  Keyword arguments:
    loadPosition (bool): indicate to load positions
    loadVelocity (bool): indicate to load velocities
    loadId (bool): indicate to load id
    loadType (bool): indicate to load particle types
-
+    
  Returns:
    It loads a gadget-1 snapshot and return a cosmotool.PySimulationBase_ object.
  """
@ -464,16 +453,16 @@ def loadParallelGadget(object filename_list, int gadgetFormat = 1, bool loadPosi
  data = alloc_simudata(num_files)
  for i,l in enumerate(filename_list):
    filenames[i] = l.encode('utf-8')
-
+  
  with nogil:
    for i in prange(num_files):
      local_data = loadGadgetMulti_safe(filenames[i], flags, gadgetFormat)
      data[i] = local_data
 #      data[i] = loadGadgetMulti(filenames[i].c_str(), -1, flags)
-
+  
  out_arrays = []
  for i in xrange(num_files):
-    if data[i] == <SimuData*>0:
+    if data[i] == <SimuData*>0:        
      out_arrays.append(None)
    else:
      out_arrays.append(PySimulationAdaptor(wrap_simudata(data[i], flags)))
@ -484,10 +473,10 @@ def loadParallelGadget(object filename_list, int gadgetFormat = 1, bool loadPosi
 def writeGadget(str filename, object simulation):
  """writeGadget(filename, simulation)
-
+  
-  This function attempts to write the content of the simulation object into
+  This function attempts to write the content of the simulation object into 
  a file named `filename` using a Gadget-1 file format.
-
+  
  Arguments:
    filename (str): output filename
    simulation (PySimulationBase): a simulation object
@ -497,23 +486,23 @@ def writeGadget(str filename, object simulation):
  cdef np.ndarray[np.int64_t, ndim=1] ids
  cdef np.int64_t NumPart
  cdef int j
-
+  
  if not isinstance(simulation,PySimulationBase):
    raise TypeError("Second argument must be of type SimulationBase")
-
+  
  NumPart = simulation.positions[0].size
  simdata.noAuto = True
-
+  
  for j in xrange(3):
    pos = simulation.getPositions()[j]
    vel = simulation.getVelocities()[j]
-
+    
    if pos.size != NumPart or vel.size != NumPart:
      raise ValueError("Invalid number of particles")
-
+    
    simdata.Pos[j] = <float *>pos.data
    simdata.Vel[j] = <float *>vel.data
-
+  
  ids = simulation.getIdentifiers()
  simdata.Id = <int64_t *>ids.data
  simdata.BoxSize = simulation.getBoxsize()
@ -530,21 +519,21 @@ def writeGadget(str filename, object simulation):
 def loadRamses(str basepath, int snapshot_id, int cpu_id, bool doublePrecision = False, bool loadPosition = True, bool loadVelocity = False, bool loadId = False, bool loadMass = False):
  """ loadRamses(basepath, snapshot_id, cpu_id, doublePrecision = False, loadPosition = True, loadVelocity = False)
  Loads the indicated snapshot based on the cpu id, snapshot id and basepath. It is important to specify the correct precision in doublePrecision otherwise the loading will fail. There is no way of auto-detecting properly the precision of the snapshot file.
-
+  
  Args:
    basepath (str): the base directory of the snapshot
    snapshot_id (int): the snapshot id
    cpu_id (int): the cpu id of the file to load
-
+  
  Keyword args:
    doublePrecision (bool): By default it is False, thus singlePrecision
    loadPosition (bool): Whether to load positions
    loadVelocity (bool): Whether to load velocities
    loadId (bool): Whether to load identifiers
-    loadMass (bool): Whether to load mass value
+    loadMass (bool): Whether to load mass value 
-
+    
  Returns:
-    An object derived from PySimulationBase_.
+    An object derived from PySimulationBase_. 
 """
  cdef int flags
  cdef SimuData *data
@ -560,7 +549,7 @@ def loadRamses(str basepath, int snapshot_id, int cpu_id, bool doublePrecision =
  if loadMass:
     flags |= NEED_MASS
-  encpath = basepath.encode('utf-8')
+  encpath = basepath.encode('utf-8') 
  try:
    data = loadRamsesSimu(encpath, snapshot_id, cpu_id, doublePrecision, flags)
    if data == <SimuData*>0:
@ -569,4 +558,4 @@ def loadRamses(str basepath, int snapshot_id, int cpu_id, bool doublePrecision =
    raise RuntimeError(str(e) + ' (check the float precision in snapshot)')
  return PySimulationAdaptor(wrap_simudata(data, flags))
-
+  
--- a/python/_project.pyx
+++ b/python/_project.pyx
@ -25,8 +25,8 @@ cdef extern from "openmp.hpp" namespace "CosmoTool":
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
-cdef void interp3d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y,
+cdef void interp3d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y, 
-                                        DTYPE_t z,
+                                        DTYPE_t z, 
                                        DTYPE_t[:,:,:] d, DTYPE_t Lbox, DTYPE_t *retval) nogil:
    cdef int Ngrid = d.shape[0]
@ -84,8 +84,8 @@ cdef void interp3d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y,
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
-cdef void ngp3d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y,
+cdef void ngp3d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y, 
-                                 DTYPE_t z,
+                                 DTYPE_t z, 
                                 DTYPE_t[:,:,:] d, DTYPE_t Lbox, DTYPE_t *retval) nogil:
    cdef int Ngrid = d.shape[0]
@ -108,14 +108,14 @@ cdef void ngp3d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y,
    iy = iy%Ngrid
    iz = iz%Ngrid
-    retval[0] = d[ix ,iy ,iz ]
+    retval[0] = d[ix ,iy ,iz ] 
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
-cdef void ngp3d_INTERNAL(DTYPE_t x, DTYPE_t y,
+cdef void ngp3d_INTERNAL(DTYPE_t x, DTYPE_t y, 
-                         DTYPE_t z,
+                         DTYPE_t z, 
                         DTYPE_t[:,:,:] d, DTYPE_t Lbox, DTYPE_t *retval, DTYPE_t inval) nogil:
    cdef int Ngrid = d.shape[0]
@ -137,16 +137,16 @@ cdef void ngp3d_INTERNAL(DTYPE_t x, DTYPE_t y,
      retval[0] = inval
      return
-    retval[0] = d[ix ,iy ,iz ]
+    retval[0] = d[ix ,iy ,iz ] 
@cython.boundscheck(False)
@cython.cdivision(True)
@cython.wraparound(False)
-cdef void interp3d_INTERNAL(DTYPE_t x, DTYPE_t y,
+cdef void interp3d_INTERNAL(DTYPE_t x, DTYPE_t y, 
-                               DTYPE_t z,
+                               DTYPE_t z, 
                               DTYPE_t[:,:,:] d, DTYPE_t Lbox, DTYPE_t *retval, DTYPE_t inval) nogil:
-
+    
    cdef int Ngrid = d.shape[0]
    cdef DTYPE_t inv_delta = Ngrid/Lbox
    cdef int ix, iy, iz
@ -193,13 +193,13 @@ cdef void interp3d_INTERNAL(DTYPE_t x, DTYPE_t y,
        d[ix+1,iy+1,iz+1] * f[1][1][1]
@cython.boundscheck(False)
-def interp3d(x not None, y not None,
+def interp3d(x not None, y not None, 
             z not None,
             npx.ndarray[DTYPE_t, ndim=3] d not None, DTYPE_t Lbox,
             bool periodic=False, bool centered=True, bool ngp=False, DTYPE_t inval = 0):
    """ interp3d(x,y,z,d,Lbox,periodic=False,centered=True,ngp=False) -> interpolated values
-
+    
-    Compute the tri-linear interpolation of the given field (d) at the given position (x,y,z). It assumes that they are box-centered coordinates. So (x,y,z) == (0,0,0) is equivalent to the pixel at (Nx/2,Ny/2,Nz/2) with Nx,Ny,Nz = d.shape. If periodic is set, it assumes the box is periodic
+    Compute the tri-linear interpolation of the given field (d) at the given position (x,y,z). It assumes that they are box-centered coordinates. So (x,y,z) == (0,0,0) is equivalent to the pixel at (Nx/2,Ny/2,Nz/2) with Nx,Ny,Nz = d.shape. If periodic is set, it assumes the box is periodic 
 """
    cdef npx.ndarray[DTYPE_t] out
    cdef DTYPE_t[:] out_slice
@ -227,12 +227,12 @@ def interp3d(x not None, y not None,
    if type(x) == np.ndarray or type(y) == np.ndarray or type(z) == np.ndarray:
        if type(x) != np.ndarray or type(y) != np.ndarray or type(z) != np.ndarray:
            raise ValueError("All or no array. No partial arguments")
-
+        
        ax = x
        ay = y
        az = z
        assert ax.size == ay.size and ax.size == az.size
-
+        
        out = np.empty(x.shape, dtype=DTYPE)
        out_slice = out
        in_slice = d
@ -280,10 +280,10 @@ cdef DTYPE_t interp2d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y,
    rx = (inv_delta*x + Ngrid/2)
    ry = (inv_delta*y + Ngrid/2)
-
+    
    ix = int(floor(rx))
    iy = int(floor(ry))
-
+    
    rx -= ix
    ry -= iy
@ -291,13 +291,13 @@ cdef DTYPE_t interp2d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y,
      ix += Ngrid
    while iy < 0:
      iy += Ngrid
-
+    
    jx = (ix+1)%Ngrid
    jy = (iy+1)%Ngrid
-
+    
    assert ((ix >= 0) and ((jx) < Ngrid))
    assert ((iy >= 0) and ((jy) < Ngrid))
-
+    
    f[0][0] = (1-rx)*(1-ry)
    f[1][0] = (  rx)*(1-ry)
    f[0][1] = (1-rx)*(  ry)
@ -314,7 +314,7 @@ cdef DTYPE_t interp2d_INTERNAL_periodic(DTYPE_t x, DTYPE_t y,
@cython.cdivision(True)
 cdef DTYPE_t interp2d_INTERNAL(DTYPE_t x, DTYPE_t y,
                               npx.ndarray[DTYPE_t, ndim=2] d, DTYPE_t Lbox) except? 0:
-
+    
    cdef int Ngrid = d.shape[0]
    cdef DTYPE_t inv_delta = Ngrid/Lbox
    cdef int ix, iy
@ -348,7 +348,7 @@ cdef DTYPE_t interp2d_INTERNAL(DTYPE_t x, DTYPE_t y,
        d[ix+1,iy  ] * f[1][0] + \
        d[ix  ,iy+1] * f[0][1] + \
        d[ix+1,iy+1] * f[1][1]
-
+        
 def interp2d(x not None, y not None,
             npx.ndarray[DTYPE_t, ndim=2] d not None, DTYPE_t Lbox,
             bool periodic=False):
@ -362,11 +362,11 @@ def interp2d(x not None, y not None,
    if type(x) == np.ndarray or type(y) == np.ndarray:
        if type(x) != np.ndarray or type(y) != np.ndarray:
            raise ValueError("All or no array. No partial arguments")
-
+        
        ax = x
        ay = y
-        assert ax.size == ay.size
+        assert ax.size == ay.size 
-
+        
        out = np.empty(x.shape, dtype=DTYPE)
        if periodic:
          for i in range(ax.size):
@ -381,8 +381,8 @@ def interp2d(x not None, y not None,
        return interp2d_INTERNAL_periodic(x, y, d, Lbox)
      else:
        return interp2d_INTERNAL(x, y, d, Lbox)
-
+        
-
+ 
@cython.boundscheck(False)
@cython.cdivision(True)
 cdef void INTERNAL_project_cic_no_mass(DTYPE_t[:,:,:] g,
@ -450,7 +450,7 @@ cdef void INTERNAL_project_cic_no_mass_periodic(DTYPE_t[:,:,:] g,
        ag[b1[0],b[1],b[2]] += a[0]*c[1]*c[2]
        ag[b[0],b1[1],b[2]] += c[0]*a[1]*c[2]
        ag[b1[0],b1[1],b[2]] += a[0]*a[1]*c[2]
-
+        
        ag[b[0],b[1],b1[2]] += c[0]*c[1]*a[2]
        ag[b1[0],b[1],b1[2]] += a[0]*c[1]*a[2]
        ag[b[0],b1[1],b1[2]] += c[0]*a[1]*a[2]
@ -525,21 +525,20 @@ cdef void INTERNAL_project_cic_with_mass_periodic(DTYPE_t[:,:,:] g,
        g[b1[0],b[1],b[2]] += a[0]*c[1]*c[2]*m0
        g[b[0],b1[1],b[2]] += c[0]*a[1]*c[2]*m0
        g[b1[0],b1[1],b[2]] += a[0]*a[1]*c[2]*m0
-
+        
        g[b[0],b[1],b1[2]] += c[0]*c[1]*a[2]*m0
        g[b1[0],b[1],b1[2]] += a[0]*c[1]*a[2]*m0
        g[b[0],b1[1],b1[2]] += c[0]*a[1]*a[2]*m0
        g[b1[0],b1[1],b1[2]] += a[0]*a[1]*a[2]*m0
-
+       
 def project_cic(npx.ndarray[DTYPE_t, ndim=2] x not None, npx.ndarray[DTYPE_t, ndim=1] mass, int Ngrid,
-                double Lbox, bool periodic = False, centered=True, output=None):
+                double Lbox, bool periodic = False, centered=True):
    """
-    project_cic(x array (N,3), mass (may be None), Ngrid, Lbox, periodict, centered=True, output=None)
+    project_cic(x array (N,3), mass (may be None), Ngrid, Lbox, periodict, centered=True)
-    This function does a Cloud-In-Cell projection of a 3d unstructured dataset. First argument is a Nx3 array of coordinates.
+    This function does a Cloud-In-Cell projection of a 3d unstructured dataset. First argument is a Nx3 array of coordinates. 
    Second argument is an optinal mass. Ngrid is the size output grid and Lbox is the physical size of the grid.
    if output is not None, it must be a numpy array with dimension NgridxNgridxNgrid. The result will be accumulated in that array.
    """
    cdef npx.ndarray[DTYPE_t, ndim=3] g
    cdef double shifter
@ -559,13 +558,7 @@ def project_cic(npx.ndarray[DTYPE_t, ndim=2] x not None, npx.ndarray[DTYPE_t, nd
    if mass is not None and mass.shape[0] != x.shape[0]:
        raise ValueError("Mass array and coordinate array must have the same number of elements")
-    if output is None:
+    g = np.zeros((Ngrid,Ngrid,Ngrid),dtype=DTYPE)
        g = np.zeros((Ngrid,Ngrid,Ngrid),dtype=DTYPE)
    else:
        if type(output) != np.ndarray:
            raise ValueError("Invalid array type")
        g = output
    cdef DTYPE_t[:,:,:] d_g = g
    cdef DTYPE_t[:,:] d_x = x
@ -576,7 +569,7 @@ def project_cic(npx.ndarray[DTYPE_t, ndim=2] x not None, npx.ndarray[DTYPE_t, nd
       else:
         d_mass = mass
         with nogil:
-           INTERNAL_project_cic_with_mass(d_g, d_x, d_mass, Ngrid, Lbox, shifter)
+           INTERNAL_project_cic_with_mass(d_g, d_x, d_mass, Ngrid, Lbox, shifter)       
    else:
       if mass is None:
         with nogil:
@ -584,13 +577,13 @@ def project_cic(npx.ndarray[DTYPE_t, ndim=2] x not None, npx.ndarray[DTYPE_t, nd
       else:
         d_mass = mass
         with nogil:
-          INTERNAL_project_cic_with_mass_periodic(d_g, d_x, d_mass, Ngrid, Lbox, shifter)
+          INTERNAL_project_cic_with_mass_periodic(d_g, d_x, d_mass, Ngrid, Lbox, shifter)       
-
+         
    return g
 def tophat_fourier_internal(npx.ndarray[DTYPE_t, ndim=1] x not None):
    cdef int i
-    cdef npx.ndarray[DTYPE_t] y
+    cdef npx.ndarray[DTYPE_t] y    
    cdef DTYPE_t x0
    y = np.empty(x.size, dtype=DTYPE)
@ -616,7 +609,7 @@ def tophat_fourier(x not None):
    return b.reshape(x.shape)
-
+    
@cython.boundscheck(False)
@cython.cdivision(True)
@ -666,25 +659,25 @@ cdef DTYPE_t cube_integral_trilin(DTYPE_t u[3], DTYPE_t u0[3], int r[1], DTYPE_t
        if tmp_a < alpha_max:
            alpha_max = tmp_a
            j = i
-
+     
    I = compute_projection(vertex_value, u, u0, alpha_max)
-
+    
    for i in xrange(3):
        u0[i] += u[i]*alpha_max
    # alpha_max is the integration length
    # we integrate between 0 and alpha_max (curvilinear coordinates)
    r[0] = j
-
+    
    return I
@cython.boundscheck(False)
 cdef DTYPE_t integrator0(DTYPE_t[:,:,:] density,
                         DTYPE_t u[3], DTYPE_t u0[3], int u_delta[3], int iu0[3], int jumper[1], DTYPE_t alpha_max) nogil:
    cdef DTYPE_t d
-
+    
    d = density[iu0[0], iu0[1], iu0[2]]
-
+    
    return cube_integral(u, u0, jumper, alpha_max)*d
@cython.boundscheck(False)
@ -694,7 +687,7 @@ cdef DTYPE_t integrator1(DTYPE_t[:,:,:] density,
    cdef DTYPE_t d
    cdef int a[3][2]
    cdef int i
-
+    
    for i in xrange(3):
      a[i][0] = iu0[i]
      a[i][1] = iu0[i]+1
@ -712,14 +705,14 @@ cdef DTYPE_t integrator1(DTYPE_t[:,:,:] density,
    return cube_integral_trilin(u, u0, jumper, vertex_value, alpha_max)
-
+    
@cython.boundscheck(False)
 cdef DTYPE_t C_line_of_sight_projection(DTYPE_t[:,:,:] density,
                             DTYPE_t a_u[3],
                             DTYPE_t min_distance,
                             DTYPE_t max_distance, DTYPE_t[:] shifter, int integrator_id) nogil except? 0:
-    cdef DTYPE_t u[3]
+    cdef DTYPE_t u[3] 
    cdef DTYPE_t ifu0[3]
    cdef DTYPE_t u0[3]
    cdef DTYPE_t utot[3]
@ -727,11 +720,11 @@ cdef DTYPE_t C_line_of_sight_projection(DTYPE_t[:,:,:] density,
    cdef int iu0[3]
    cdef int i
    cdef int N = density.shape[0]
-    cdef int half_N = density.shape[0]//2
+    cdef int half_N = density.shape[0]/2
    cdef int completed
    cdef DTYPE_t I0, d, dist2, delta, s, max_distance2
    cdef int jumper[1]
-
+    
    cdef DTYPE_t (*integrator)(DTYPE_t[:,:,:],
                               DTYPE_t u[3], DTYPE_t u0[3], int u_delta[3], int iu0[3], int jumper[1], DTYPE_t alpha_max) nogil
@ -754,7 +747,7 @@ cdef DTYPE_t C_line_of_sight_projection(DTYPE_t[:,:,:] density,
        if (not ((iu0[i]>= 0) and (iu0[i] < N))):
          with gil:
            raise RuntimeError("iu0[%d] = %d !!" % (i,iu0[i]))
-
+            
        if (not (u0[i]>=0 and u0[i]<=1)):
          with gil:
            raise RuntimeError("u0[%d] = %g !" % (i,u0[i]))
@ -763,7 +756,7 @@ cdef DTYPE_t C_line_of_sight_projection(DTYPE_t[:,:,:] density,
    if ((iu0[0] >= N-1) or (iu0[0] <= 0) or
        (iu0[1] >= N-1) or (iu0[1] <= 0) or
        (iu0[2] >= N-1) or (iu0[2] <= 0)):
-      completed = 1
+      completed = 1 
    I0 = 0
    jumper[0] = 0
@ -778,8 +771,8 @@ cdef DTYPE_t C_line_of_sight_projection(DTYPE_t[:,:,:] density,
            iu0[jumper[0]] += 1
            u0[jumper[0]] = 0
-
+            
-        if ((iu0[0] >= N-1) or (iu0[0] <= 0) or
+        if ((iu0[0] >= N-1) or (iu0[0] <= 0) or 
            (iu0[1] >= N-1) or (iu0[1] <= 0) or
            (iu0[2] >= N-1) or (iu0[2] <= 0)):
            completed = 1
@ -794,7 +787,7 @@ cdef DTYPE_t C_line_of_sight_projection(DTYPE_t[:,:,:] density,
            #delta = sqrt(dist2) - max_distance
            #I0 -= d*delta
            completed = 1
-
+            
    return I0
 def line_of_sight_projection(DTYPE_t[:,:,:] density not None,
@ -802,18 +795,18 @@ def line_of_sight_projection(DTYPE_t[:,:,:] density not None,
                             DTYPE_t min_distance,
                             DTYPE_t max_distance, DTYPE_t[:] shifter not None, int integrator_id=0):
  cdef DTYPE_t u[3]
-
+  
  u[0] = a_u[0]
  u[1] = a_u[1]
  u[2] = a_u[2]
-
+  
  return C_line_of_sight_projection(density,
                             u,
                             min_distance,
                             max_distance, shifter, integrator_id)
-cdef double _spherical_projloop(double theta, double phi, DTYPE_t[:,:,:] density,
+cdef double _spherical_projloop(double theta, double phi, DTYPE_t[:,:,:] density, 
-                double min_distance, double max_distance,
+                double min_distance, double max_distance, 
                DTYPE_t[:] shifter, int integrator_id) nogil:
  cdef DTYPE_t u0[3]
@ -821,7 +814,7 @@ cdef double _spherical_projloop(double theta, double phi, DTYPE_t[:,:,:] density
  u0[0] = cos(phi)*stheta
  u0[1] = sin(phi)*stheta
  u0[2] = cos(theta)
-
+ 
  return C_line_of_sight_projection(density, u0, min_distance, max_distance, shifter, integrator_id)
@ -832,20 +825,20 @@ def spherical_projection(int Nside,
                         DTYPE_t max_distance, int progress=1, int integrator_id=0, DTYPE_t[:] shifter = None, int booster=-1):
    """
    spherical_projection(Nside, density, min_distance, max_distance, progress=1, integrator_id=0, shifter=None, booster=-1)
-
+    
    Keyword arguments:
      progress (int): show progress if it is equal to 1
      integrator_id (int): specify the order of integration along the line of shift
      shifter (DTYPE_t array): this is an array of size 3. It specifies the amount of shift to apply to the center, in unit of voxel
      booster (int): what is the frequency of refreshment of the progress bar. Small number decreases performance by locking the GIL.
-
+      
    Arguments:
      Nside (int): Nside of the returned map
      density (NxNxN array): this is the density field, expressed as a cubic array
      min_distance (float): lower bound of the integration
      max_distance (float): upper bound of the integration
-
+      
    Returns:
      an healpix map, as a 1-dimensional array.
    """
@ -860,11 +853,11 @@ def spherical_projection(int Nside,
    cdef long N, N0
    cdef double stheta
    cdef int tid
-
+    
    if shifter is None:
        shifter = view.array(shape=(3,), format=FORMAT_DTYPE, itemsize=sizeof(DTYPE_t))
        shifter[:] = 0
-
+        
    print("allocating map")
    outm_array = np.empty(hp.nside2npix(Nside),dtype=DTYPE)
    print("initializing views")
@ -877,10 +870,10 @@ def spherical_projection(int Nside,
    N = smp_get_max_threads()
    N0 = outm.size
-
+    
    if booster < 0:
      booster = 1#000
-
+    
    job_done = view.array(shape=(N,), format="i", itemsize=sizeof(int))
    job_done[:] = 0
    theta,phi = hp.pix2ang(Nside, np.arange(N0))
--- a/python/cosmotool.version
+++ b/python/cosmotool.version
@ -1,7 +0,0 @@
 CODEABI_1.0 {
  global:
    PyInit_*;
    _init;
    _fini;
  local: *;
 };
--- a/requirements.txt
+++ b/requirements.txt
@ -1,4 +1,5 @@
-numpy<1.22
+numpy
 cffi
 numexpr
-cython<3
+pyfftw
 cython
--- a/sample/CMakeLists.txt
+++ b/sample/CMakeLists.txt
@ -1,8 +1,4 @@
 SET(tolink  ${CosmoTool_local} ${CosmoTool_LIBS} ${GSL_LIBRARIES} ${DL_LIBRARY})
 if (RT_LIBRARY)
  SET(tolink ${tolink} ${RT_LIBRARY})
 endif()
 include_directories(${CMAKE_SOURCE_DIR}/src)
 include_directories(${FFTW3_INCLUDE_DIRS} ${EIGEN3_INCLUDE_DIRS} ${GSL_INCLUDE_PATH})
 if(YORICK_SUPPORT)
--- a/sample/simple3DFilter.cpp
+++ b/sample/simple3DFilter.cpp
@ -1,22 +1,22 @@
 #include "hdf5_array.hpp"
 #include "miniargs.hpp"
 #include "mykdtree.hpp"
 #include "omptl/algorithm"
 #include "openmp.hpp"
-#include "sphSmooth.hpp"
+#include "omptl/algorithm"
 #include "yorick.hpp"
 #include <H5Cpp.h>
 #include <boost/bind.hpp>
 #include <boost/format.hpp>
 #include <cassert>
 #include "yorick.hpp"
 #include "sphSmooth.hpp"
 #include "mykdtree.hpp"
 #include "miniargs.hpp"
 #include <H5Cpp.h>
 #include "hdf5_array.hpp"
 #include <iostream>
 #include <boost/format.hpp>
 #include <boost/bind.hpp>
 using namespace std;
 using namespace CosmoTool;
 #define N_SPH 32
-struct VCoord {
+struct VCoord{
  float v[3];
  float mass;
 };
@ -27,186 +27,192 @@ typedef boost::multi_array<float, 2> array_type;
 typedef boost::multi_array<float, 3> array3_type;
 typedef boost::multi_array<float, 4> array4_type;
-ComputePrecision getVelocity(const VCoord &v, int i) { return v.mass * v.v[i]; }
+ComputePrecision getVelocity(const VCoord& v, int i)
 {
  return v.mass * v.v[i];
 }
-ComputePrecision getMass(const VCoord &v) { return v.mass; }
+ComputePrecision getMass(const VCoord& v)
 {
  return v.mass;
 }
 typedef SPHSmooth<VCoord> MySmooth;
 typedef MySmooth::SPHTree MyTree;
 typedef MyTree::Cell MyCell;
-template <typename FuncT>
+template<typename FuncT>
-void computeInterpolatedField(MyTree *tree1, double boxsize, int Nres,
+void computeInterpolatedField(MyTree *tree1, double boxsize, int Nres, double cx, double cy, double cz,
-                              double cx, double cy, double cz,
+                              array3_type& bins, array3_type& arr, FuncT func, double rLimit2)
-                              array3_type &bins, array3_type &arr, FuncT func,
+{
-                              double rLimit2) {
+#pragma omp parallel
  int rz_max = 0;
 #pragma omp parallel shared(rz_max)
  {
    MySmooth smooth1(tree1, N_SPH);
 #pragma omp for schedule(dynamic) 
    for (int rz = 0; rz < Nres; rz++)
      {
        double pz = (rz)*boxsize/Nres-cz;
-#pragma omp for collapse(3) schedule(dynamic)
+        cout << format("[%d] %d / %d") % smp_get_thread_id() % rz % Nres << endl;
-    for (int rz = 0; rz < Nres; rz++) {
+        for (int ry = 0; ry < Nres; ry++)
          {
            double py = (ry)*boxsize/Nres-cy;
            for (int rx = 0; rx < Nres; rx++)
              {
                double px = (rx)*boxsize/Nres-cx;
                MyTree::coords c = { float(px), float(py), float(pz) };
-      for (int ry = 0; ry < Nres; ry++) {
+                double r2 = c[0]*c[0]+c[1]*c[1]+c[2]*c[2];
-        for (int rx = 0; rx < Nres; rx++) {
+                if (r2 > rLimit2)
-          if (rz > rz_max) {
+                  {
-            rz_max = rz;
+                    arr[rx][ry][rz] = 0;
-            cout << format("[%d] %d / %d") % smp_get_thread_id() % rz % Nres
+                    continue;
-                 << endl;
+                  }
                uint32_t numInCell = bins[rx][ry][rz];
                if (numInCell > N_SPH)
                  smooth1.fetchNeighbours(c, numInCell);
                else
                  smooth1.fetchNeighbours(c);
                arr[rx][ry][rz] = smooth1.computeSmoothedValue(c, func);
              }
          }
          double px = (rx)*boxsize / Nres - cx;
          double py = (ry)*boxsize / Nres - cy;
          double pz = (rz)*boxsize / Nres - cz;
          MyTree::coords c = {float(px), float(py), float(pz)};
          double r2 = c[0] * c[0] + c[1] * c[1] + c[2] * c[2];
          if (r2 > rLimit2) {
            arr[rx][ry][rz] = 0;
            continue;
          }
          uint32_t numInCell = bins[rx][ry][rz];
          if (numInCell > N_SPH)
            smooth1.fetchNeighbours(c, numInCell);
          else
            smooth1.fetchNeighbours(c);
          arr[rx][ry][rz] = smooth1.computeSmoothedValue(c, func);
        }
      }
    }
  }
 }
-int main(int argc, char **argv) {
+int main(int argc, char **argv)
-  char *fname1, *outFile;
+{
  char *fname1, *fname2;
  double rLimit, boxsize, rLimit2, cx, cy, cz;
  int Nres;
  int periodic;
-  MiniArgDesc args[] = {{"INPUT DATA1", &fname1, MINIARG_STRING},
+  MiniArgDesc args[] = {
-                        {"RADIUS LIMIT", &rLimit, MINIARG_DOUBLE},
+    { "INPUT DATA1", &fname1, MINIARG_STRING },
-                        {"BOXSIZE", &boxsize, MINIARG_DOUBLE},
+    { "RADIUS LIMIT", &rLimit, MINIARG_DOUBLE },
-                        {"RESOLUTION", &Nres, MINIARG_INT},
+    { "BOXSIZE", &boxsize, MINIARG_DOUBLE },
-                        {"CX", &cx, MINIARG_DOUBLE},
+    { "RESOLUTION", &Nres, MINIARG_INT },
-                        {"CY", &cy, MINIARG_DOUBLE},
+    { "CX", &cx, MINIARG_DOUBLE },
-                        {"CZ", &cz, MINIARG_DOUBLE},
+    { "CY", &cy, MINIARG_DOUBLE },
-                        {"OUTPUT FILE", &outFile, MINIARG_STRING},
+    { "CZ", &cz, MINIARG_DOUBLE },
-                        {"PERIODIC", &periodic, MINIARG_INT},
+    { 0, 0, MINIARG_NULL }
-                        {0, 0, MINIARG_NULL}};
+  };
  if (!parseMiniArgs(argc, argv, args))
    return 1;
  H5::H5File in_f(fname1, 0);
-  H5::H5File out_f(outFile, H5F_ACC_TRUNC);
+  H5::H5File out_f("fields.h5", H5F_ACC_TRUNC);
  array_type v1_data;
-  uint64_t N1_points, N2_points;
+  uint32_t N1_points, N2_points;
-
+  
  array3_type bins(boost::extents[Nres][Nres][Nres]);
-  rLimit2 = rLimit * rLimit;
+  rLimit2 = rLimit*rLimit;
  hdf5_read_array(in_f, "particles", v1_data);
  assert(v1_data.shape()[1] == 7);
  N1_points = v1_data.shape()[0];
-
+  
  cout << "Got " << N1_points << " in the first file." << endl;
  MyCell *allCells_1 = new MyCell[N1_points];
-
+  
 #pragma omp parallel for schedule(static)
-  for (uint32_t i = 0; i < Nres * Nres * Nres; i++)
+  for (long i = 0; i < Nres*Nres*Nres; i++)
    bins.data()[i] = 0;
  cout << "Shuffling data in cells..." << endl;
 #pragma omp parallel for schedule(static)
-  for (uint64_t i = 0; i < N1_points; i++) {
+  for (int i = 0 ; i < N1_points; i++)
-    for (int j = 0; j < 3; j++)
+    {
-      allCells_1[i].coord[j] = v1_data[i][j];
+      for (int j = 0; j < 3; j++)
-    for (int k = 0; k < 3; k++)
+        allCells_1[i].coord[j] = v1_data[i][j];
-      allCells_1[i].val.pValue.v[k] = v1_data[i][3 + k];
+      for (int k = 0; k < 3; k++)
-    allCells_1[i].val.pValue.mass = v1_data[i][6];
+        allCells_1[i].val.pValue.v[k] = v1_data[i][3+k];
-    allCells_1[i].active = true;
+      allCells_1[i].val.pValue.mass = v1_data[i][6];
-    allCells_1[i].val.weight = 0.0;
+      allCells_1[i].active = true;
      allCells_1[i].val.weight = 0.0;
-    long rx = floor((allCells_1[i].coord[0] + cx) * Nres / boxsize + 0.5);
+      long rx = floor((allCells_1[i].coord[0]+cx)*Nres/boxsize+0.5);
-    long ry = floor((allCells_1[i].coord[1] + cy) * Nres / boxsize + 0.5);
+      long ry = floor((allCells_1[i].coord[1]+cy)*Nres/boxsize+0.5);
-    long rz = floor((allCells_1[i].coord[2] + cz) * Nres / boxsize + 0.5);
+      long rz = floor((allCells_1[i].coord[2]+cz)*Nres/boxsize+0.5);
-
+      
-    if (rx < 0 || rx >= Nres || ry < 0 || ry >= Nres || rz < 0 || rz >= Nres)
+      if (rx < 0 || rx >= Nres || ry < 0 || ry >= Nres || rz < 0 || rz >= Nres)
-      continue;
+        continue;
-
+	  
-    auto &b = bins[rx][ry][rz];
+//#pragma omp atomic update
-#pragma omp atomic
+      bins[rx][ry][rz]++;
-    b++;
+    }
  }
  v1_data.resize(boost::extents[1][1]);
-
+  
  hdf5_write_array(out_f, "num_in_cell", bins);
  cout << "Building trees..." << endl;
  MyTree tree1(allCells_1, N1_points);
  tree1.setPeriodic(periodic != 0, boxsize);
  cout << "Creating smoothing filter..." << endl;
-  //  array3_type out_rad_1(boost::extents[Nres][Nres][Nres]);
+//  array3_type out_rad_1(boost::extents[Nres][Nres][Nres]);
-
+ 
  cout << "Weighing..." << endl;
-  int rz_max = 0;
+#pragma omp parallel
 #pragma omp parallel shared(rz_max)
  {
    MySmooth smooth1(&tree1, N_SPH);
 #pragma omp for schedule(dynamic) 
    for (int rz = 0; rz < Nres; rz++)
     {
        double pz = (rz)*boxsize/Nres-cz;
-#pragma omp for collapse(3) schedule(dynamic, 8)
+        (cout << rz << " / " << Nres << endl).flush();
-    for (int rz = 0; rz < Nres; rz++) {
+        for (int ry = 0; ry < Nres; ry++)
-      for (int ry = 0; ry < Nres; ry++) {
+          {
-        for (int rx = 0; rx < Nres; rx++) {
+            double py = (ry)*boxsize/Nres-cy;
-          if (rz > rz_max) {
+            for (int rx = 0; rx < Nres; rx++)
-            rz_max = rz;
+              {
-            (cout << rz << " / " << Nres << endl).flush();
+                double px = (rx)*boxsize/Nres-cx;
                MyTree::coords c = { float(px), float(py), float(pz) };
                double r2 = c[0]*c[0]+c[1]*c[1]+c[2]*c[2];
                if (r2 > rLimit2)
                  {
                    continue;
                  }
                uint32_t numInCell = bins[rx][ry][rz];
                if (numInCell > N_SPH)
                  smooth1.fetchNeighbours(c, numInCell);
                else
                  smooth1.fetchNeighbours(c);
 #pragma omp critical
                smooth1.addGridSite(c);
              }
          }
-          double pz = (rz)*boxsize / Nres - cz;
+        (cout << " Done " << rz << endl).flush();
-          double py = (ry)*boxsize / Nres - cy;
+      }  
          double px = (rx)*boxsize / Nres - cx;
          MyTree::coords c = {float(px), float(py), float(pz)};
          double r2 = c[0] * c[0] + c[1] * c[1] + c[2] * c[2];
          if (r2 > rLimit2) {
            continue;
          }
          uint32_t numInCell = bins[rx][ry][rz];
          if (numInCell > N_SPH)
            smooth1.fetchNeighbours(c, numInCell);
          else
            smooth1.fetchNeighbours(c);
          smooth1.addGridSite(c);
        }
      }
    }
  }
-
+  
  cout << "Interpolating..." << endl;
  array3_type interpolated(boost::extents[Nres][Nres][Nres]);
-
+  
-  computeInterpolatedField(&tree1, boxsize, Nres, cx, cy, cz, bins,
+  computeInterpolatedField(&tree1, boxsize, Nres, cx, cy, cz,
-                           interpolated, getMass, rLimit2);
+                           bins, interpolated, getMass, rLimit2);
  hdf5_write_array(out_f, "density", interpolated);
-  // out_f.flush();
+  //out_f.flush();
  for (int i = 0; i < 3; i++) {
-    computeInterpolatedField(&tree1, boxsize, Nres, cx, cy, cz, bins,
+      computeInterpolatedField(&tree1, boxsize, Nres, cx, cy, cz,
-                             interpolated, boost::bind(getVelocity, _1, i),
+                               bins, interpolated, boost::bind(getVelocity, _1, i), rLimit2);
-                             rLimit2);
+      hdf5_write_array(out_f, str(format("p%d") % i), interpolated);
    hdf5_write_array(out_f, str(format("p%d") % i), interpolated);
  }
-
+  
  return 0;
 };
--- a/setup.py
+++ b/setup.py
@ -2,7 +2,6 @@ import stat
 import os
 import sys
 import shutil
 from sysconfig import get_config_var
 from distutils.command.install_data import install_data
 import pathlib
 from setuptools import find_packages, setup, Extension
@ -77,9 +76,9 @@ class InstallCMakeLibs(install_lib):
 #        # your files are moved to the appropriate location when the installation
 #        # is run
 #
-#        libs = [os.path.join(bin_dir, _lib) for _lib in
+#        libs = [os.path.join(bin_dir, _lib) for _lib in 
-#                os.listdir(bin_dir) if
+#                os.listdir(bin_dir) if 
-#                os.path.isfile(os.path.join(bin_dir, _lib)) and
+#                os.path.isfile(os.path.join(bin_dir, _lib)) and 
 #                os.path.splitext(_lib)[1] in [".dll", ".so"]
 #                and not (_lib.startswith("python") or _lib.startswith(PACKAGE_NAME))]
 #
@ -88,16 +87,16 @@ class InstallCMakeLibs(install_lib):
 #            shutil.move(lib, os.path.join(self.build_dir,
 #                                          os.path.basename(lib)))
 #
-#        # Mark the libs for installation, adding them to
+#        # Mark the libs for installation, adding them to 
-#        # distribution.data_files seems to ensure that setuptools' record
+#        # distribution.data_files seems to ensure that setuptools' record 
 #        # writer appends them to installed-files.txt in the package's egg-info
 #        #
-#        # Also tried adding the libraries to the distribution.libraries list,
+#        # Also tried adding the libraries to the distribution.libraries list, 
-#        # but that never seemed to add them to the installed-files.txt in the
+#        # but that never seemed to add them to the installed-files.txt in the 
-#        # egg-info, and the online recommendation seems to be adding libraries
+#        # egg-info, and the online recommendation seems to be adding libraries 
-#        # into eager_resources in the call to setup(), which I think puts them
+#        # into eager_resources in the call to setup(), which I think puts them 
-#        # in data_files anyways.
+#        # in data_files anyways. 
-#        #
+#        # 
 #        # What is the best way?
 #
 #        # These are the additional installation files that should be
@ -105,7 +104,7 @@ class InstallCMakeLibs(install_lib):
 #        # step; depending on the files that are generated from your cmake
 #        # build chain, you may need to modify the below code
 #
-#        self.distribution.data_files = [os.path.join(self.install_dir,
+#        self.distribution.data_files = [os.path.join(self.install_dir, 
 #                                                     os.path.basename(lib))
 #                                        for lib in libs]
 #        print(self.distribution.data_files)
@ -168,33 +167,15 @@ class BuildCMakeExt(build_ext):
        # Change your cmake arguments below as necessary
        # Below is just an example set of arguments for building Blender as a Python module
        c_compiler=os.environ.get('CC', get_config_var("CC"))
        cxx_compiler=os.environ.get('CXX', get_config_var("CXX"))
        compilers=[]
        fill_up_settings=[
              ("CMAKE_C_COMPILER", "CC", get_config_var("CC")),
              ("CMAKE_CXX_COMPILER", "CXX", get_config_var("CXX")),
              ("CMAKE_C_FLAGS", "CFLAGS", None),
              ("CMAKE_CXX_FLAGS", "CXXFLAGS", None),
              ("CMAKE_EXE_LINKER_FLAGS_INIT", "LDFLAGS", None),
              ("CMAKE_SHARED_LINKER_FLAGS_INIT", "LDFLAGS", None),
              ("CMAKE_MODULE_LINKER_FLAGS_INIT", "LDFLAGS", None)]
        for cmake_flag, os_flag, default_flag in fill_up_settings:
          if os_flag in os.environ or default_flag is not None:
             c = os.environ.get(os_flag, default_flag)
             if not c is None:
               compilers.append(f"-D{cmake_flag}={c}")
        print(compilers)
        self.spawn(['cmake', '-H'+SOURCE_DIR, '-B'+self.build_temp,
                    '-DCMAKE_C_COMPILER=' + os.environ["CC"], "-DCMAKE_CXX_COMPILER=" + os.environ["CXX"],
                    '-DENABLE_OPENMP=ON','-DINTERNAL_BOOST=ON','-DINTERNAL_EIGEN=ON',
                    '-DINTERNAL_HDF5=ON','-DINTERNAL_NETCDF=ON',
                    '-DBUILD_PYTHON=ON', '-DINSTALL_PYTHON_LOCAL=OFF',
                    '-DCOSMOTOOL_PYTHON_PACKAGING=ON',
                    f"-DCYTHON={cython_code}",
                    f"-DPYTHON_SITE_PACKAGES={build_dir.absolute()}/private_install",
-                    f"-DPYTHON_EXECUTABLE={sys.executable}"] + compilers)
+                    f"-DPYTHON_EXECUTABLE={sys.executable}"])
        self.announce("Building binaries", level=3)
@ -221,7 +202,7 @@ class BuildCMakeExt(build_ext):
               if  _pyd_top[0].startswith(PACKAGE_NAME):
                   if os.path.splitext(_pyd)[1] in [".pyd", ".so"] or _pyd_top[-1] == 'config.py':
                     pyd_path.append((_pyd_top,_pyd))
-
+               
        for top,p in pyd_path:
           _,n = os.path.split(p)
@ -237,10 +218,10 @@ class BuildCMakeExt(build_ext):
 CosmoTool_extension = CMakeExtension(name="cosmotool")
 setup(name='cosmotool',
-      version='1.3.6',
+      version='1.1.0',
      packages=["cosmotool"],
      package_dir={'cosmotool': 'python/cosmotool'},
-      install_requires=['cython','numpy','cffi','numexpr','h5py'],
+      install_requires=['numpy','cffi','numexpr','pyfftw','h5py'],
      setup_requires=['cython','cffi','numpy','numexpr'],
      ext_modules=[CosmoTool_extension],
      description='A small cosmotool box of useful functions',
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@ -38,7 +38,6 @@ if (HDF5_FOUND)
    h5_readFlash.cpp
    loadFlash.cpp
    )
  add_dependencies(CosmoHDF5 ${cosmotool_DEPS})
  set_property(TARGET CosmoHDF5 PROPERTY POSITION_INDEPENDENT_CODE ${BUILD_SHARED_LIBS})
  target_include_directories(CosmoHDF5 BEFORE PRIVATE ${HDF5_INCLUDE_DIR})
 else(HDF5_FOUND)
--- a/src/cosmopower.cpp
+++ b/src/cosmopower.cpp
@ -212,30 +212,6 @@ double CosmoPower::powerHuWiggles(double k)
  return normPower * pow(k,n) * T_k * T_k;
 }
 double CosmoPower::BAO_Tk(double k){
  double no_wiggle_tk = noWiggleTk(k);
  double A = 0;
  double r_s = 10;
  double k_D = 2 * M_PI / 100;
  //sqrt as we want to make the parameterization part of the transfer function
  double param = sqrt(1 + A * sin(k * r_s) * exp(- k / k_D)); 
  return no_wiggle_tk * param;
 }
 double CosmoPower::sample_BAO(double k)
 {
  // BAO wiggle parameterization for reconstruction
  // Babic et al. 2022, https://arxiv.org/abs/2203.06177
  double T_k = BAO_Tk(k); 
  return normPower * pow(k,n) * T_k * T_k;
  }
 double CosmoPower::primordialPowerSpectrum(double k)
 {
  //Primordial power spectrum, needed for PNG
@ -273,19 +249,6 @@ double CosmoPower::matterTransferFunctionHu(double k)
  return T_k;
 }
 double CosmoPower::noWiggleTk(double k)
 {
  double s = 44.5 * log(9.83 / OmegaEff) / (sqrt(1 + 10 * pow(OMEGA_B * h * h, 0.75)));
  double alpha_Gamma = 1 - 0.328 * log(431 * OmegaEff) * OMEGA_B / OMEGA_0 + 0.38 * log(22.3 * OmegaEff) * pow(OMEGA_B / OMEGA_0, 2);
  double GammaEff = OMEGA_0 * h * (alpha_Gamma + (1 - alpha_Gamma)/(1 + pow(0.43 * k * s, 4)));
  double q = k/(h*GammaEff) * pow(Theta_27, 2);
  double L_0 = log(2 * M_E + 1.8 * q);
  double C_0 = 14.2 + 731 / (1 + 62.5 * q);
  double T0 = L_0 / (L_0 + C_0 * q * q);
  return T0;
 }
 double CosmoPower::powerHuBaryons(double k)
 {
  double s = 44.5 * log(9.83 / OmegaEff) / (sqrt(1 + 10 * pow(OMEGA_B * h * h, 0.75)));
@ -379,14 +342,12 @@ void CosmoPower::normalize(double k_min, double k_max)
  normPower = 1;
 #if 0
  ofstream ff("PP_k.txt");
  for (int i = 0; i < 100; i++)
    {
       double k = pow(10.0, 8.0*i/100.-4);
       ff << k << " " << power(k) << endl;
    }
 #endif
 // gsl_integration_qagiu(&f, 0, 0, TOLERANCE, NUM_ITERATION, w, &normVal, &abserr);
  gsl_integration_qag(&f, x_min, x_max, 0, TOLERANCE, NUM_ITERATION, GSL_INTEG_GAUSS61, w, &normVal, &abserr);
@ -481,18 +442,12 @@ void CosmoPower::setFunction(CosmoFunction f)
    case POWER_SUGIYAMA:
      eval = &CosmoPower::powerSugiyama;
      break;
    case SAMPLE_WIGGLES:
      eval = &CosmoPower::sample_BAO;
      break;
    case POWER_BDM:
      eval = &CosmoPower::powerBDM;
      break;
    case POWER_TEST:
      eval = &CosmoPower::powerTest;
      break;
    case BAO_TK:
      eval = &CosmoPower::BAO_Tk;
      break;
    default:
      abort();
    }
--- a/src/cosmopower.hpp
+++ b/src/cosmopower.hpp
@ -89,9 +89,7 @@ namespace CosmoTool {
      POWER_SUGIYAMA,
      POWER_BDM,
      POWER_TEST,
-      HU_WIGGLES_ORIGINAL,
+      HU_WIGGLES_ORIGINAL
      SAMPLE_WIGGLES,
      BAO_TK
    };
    CosmoPower();
@ -124,9 +122,6 @@ namespace CosmoTool {
    double powerBDM(double k);
    double powerTest(double k);
    double powerHuWigglesOriginal(double k);
    double sample_BAO(double k);
    double noWiggleTk(double k);
    double BAO_Tk(double k);
  };
 };
--- a/src/fortran.cpp
+++ b/src/fortran.cpp
@ -139,22 +139,14 @@ void UnformattedRead::beginCheckpoint(bool bufferRecord)
 void UnformattedRead::endCheckpoint(bool autodrop)
 {
  bool always_fail = false;
  if (recordBuffer != 0) {
    delete[] recordBuffer;
    recordBuffer = 0;
  }
  if (cSize == Check_32bits) {
    if (checkPointAccum >= 1UL<<32UL) {
      always_fail = true;
      checkPointAccum %= (1UL<<32UL);
    }
  } 
  if (checkPointRef != checkPointAccum)
    {
-      if (always_fail || !autodrop || checkPointAccum > checkPointRef) {
+      if (!autodrop || checkPointAccum > checkPointRef) {
 	throw InvalidUnformattedAccess();
      }
      f->seekg(checkPointRef-checkPointAccum, ios::cur);
--- a/src/fourier/fft/fftw_calls.hpp
+++ b/src/fourier/fft/fftw_calls.hpp
@ -39,112 +39,91 @@ knowledge of the CeCILL license and that you accept its terms.
 #include <fftw3.h>
 #include <complex>
-namespace CosmoTool {
+namespace CosmoTool
 {
-  static inline void init_fftw_wisdom() {
+static inline void init_fftw_wisdom()
-    fftw_import_system_wisdom();
+{
-    fftw_import_wisdom_from_filename("fft_wisdom");
+  fftw_import_system_wisdom();
-  }
+  fftw_import_wisdom_from_filename("fft_wisdom");
 }
-  static inline void save_fftw_wisdom() {
+static inline void save_fftw_wisdom()
-    fftw_export_wisdom_to_filename("fft_wisdom");
+{
-  }
+  fftw_export_wisdom_to_filename("fft_wisdom");
 }
-  template <typename T>
+template<typename T> class FFTW_Calls {};
  class FFTW_Calls {};
 #define FFTW_CALLS_BASE(rtype, prefix)                                         \
  template <>                                                                  \
  class FFTW_Calls<rtype> {                                                    \
  public:                                                                      \
    typedef rtype real_type;                                                   \
    typedef prefix##_complex complex_type;                                     \
    typedef prefix##_plan plan_type;                                           \
                                                                               \
    static complex_type *alloc_complex(size_t N) {                             \
      return prefix##_alloc_complex(N);                                        \
    }                                                                          \
    static real_type *alloc_real(size_t N) { return prefix##_alloc_real(N); }  \
    static void free(void *p) { fftw_free(p); }                                \
                                                                               \
    static void execute(plan_type p) { prefix##_execute(p); }                  \
    static void execute_r2c(plan_type p, real_type *in, complex_type *out) {   \
      prefix##_execute_dft_r2c(p, in, out);                                    \
    }                                                                          \
    static void execute_c2r(plan_type p, complex_type *in, real_type *out) {   \
      prefix##_execute_dft_c2r(p, in, out);                                    \
    }                                                                          \
    static void                                                                \
    execute_r2c(plan_type p, real_type *in, std::complex<real_type> *out) {    \
      prefix##_execute_dft_r2c(p, in, (complex_type *)out);                    \
    }                                                                          \
    static void                                                                \
    execute_c2r(plan_type p, std::complex<real_type> *in, real_type *out) {    \
      prefix##_execute_dft_c2r(p, (complex_type *)in, out);                    \
    }                                                                          \
    static void execute_c2c(                                                   \
        plan_type p, std::complex<real_type> *in,                              \
        std::complex<real_type> *out) {                                        \
      prefix##_execute_dft(p, (complex_type *)in, (complex_type *)out);        \
    }                                                                          \
    static plan_type plan_dft_r2c_1d(                                          \
        int Nx, real_type *in, complex_type *out, unsigned flags) {            \
      return prefix##_plan_dft_r2c_1d(Nx, in, out, flags);                     \
    }                                                                          \
    static plan_type plan_dft_c2r_1d(                                          \
        int Nx, complex_type *in, real_type *out, unsigned flags) {            \
      return prefix##_plan_dft_c2r_1d(Nx, in, out, flags);                     \
    }                                                                          \
    static plan_type plan_dft_r2c_2d(                                          \
        int Nx, int Ny, real_type *in, complex_type *out, unsigned flags) {    \
      return prefix##_plan_dft_r2c_2d(Nx, Ny, in, out, flags);                 \
    }                                                                          \
    static plan_type plan_dft_c2r_2d(                                          \
        int Nx, int Ny, complex_type *in, real_type *out, unsigned flags) {    \
      return prefix##_plan_dft_c2r_2d(Nx, Ny, in, out, flags);                 \
    }                                                                          \
    static plan_type plan_dft_r2c_3d(                                          \
        int Nx, int Ny, int Nz, real_type *in, complex_type *out,              \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_r2c_3d(Nx, Ny, Nz, in, out, flags);             \
    }                                                                          \
    static plan_type plan_dft_c2r_3d(                                          \
        int Nx, int Ny, int Nz, complex_type *in, real_type *out,              \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_c2r_3d(Nx, Ny, Nz, in, out, flags);             \
    }                                                                          \
                                                                               \
    static plan_type plan_dft_r2c(                                             \
        int rank, const int *n, real_type *in, complex_type *out,              \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_r2c(rank, n, in, out, flags);                   \
    }                                                                          \
    static plan_type plan_dft_c2r(                                             \
        int rank, const int *n, complex_type *in, real_type *out,              \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_c2r(rank, n, in, out, flags);                   \
    }                                                                          \
    static plan_type plan_dft_3d(                                              \
        int Nx, int Ny, int Nz, complex_type *in, complex_type *out, int sign, \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_3d(Nx, Ny, Nz, in, out, sign, flags);           \
    }                                                                          \
    static plan_type plan_dft_2d(                                              \
        int Nx, int Ny, complex_type *in, complex_type *out, int sign,         \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_2d(Nx, Ny, in, out, sign, flags);               \
    }                                                                          \
    static plan_type plan_dft_1d(                                              \
        int Nx, complex_type *in, complex_type *out, int sign,                 \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_1d(Nx, in, out, sign, flags);                   \
    }                                                                          \
    static void destroy_plan(plan_type plan) { prefix##_destroy_plan(plan); }  \
  }
-  FFTW_CALLS_BASE(double, fftw);
+#define FFTW_CALLS_BASE(rtype, prefix) \
-  FFTW_CALLS_BASE(float, fftwf);
+  template<>				\
 class FFTW_Calls<rtype> {		\
 public: \
  typedef rtype real_type; \
  typedef prefix ## _complex complex_type; \
  typedef prefix ## _plan plan_type; \
  \
  static complex_type *alloc_complex(size_t N) { return prefix ## _alloc_complex(N); } \
  static real_type *alloc_real(size_t N) { return prefix ## _alloc_real(N); } \
  static void free(void *p) { fftw_free(p); } \
 \
  static void execute(plan_type p) { prefix ## _execute(p); } \
  static void execute_r2c(plan_type p, real_type *in, complex_type *out) { prefix ## _execute_dft_r2c(p, in, out); } \
  static void execute_c2r(plan_type p, complex_type *in, real_type *out) { prefix ## _execute_dft_c2r(p, in, out); } \
  static void execute_r2c(plan_type p, real_type *in, std::complex<real_type> *out) { prefix ## _execute_dft_r2c(p, in, (complex_type*)out); } \
  static void execute_c2r(plan_type p, std::complex<real_type> *in, real_type *out) { prefix ## _execute_dft_c2r(p, (complex_type*) in, out); } \
  static void execute_c2c(plan_type p, std::complex<real_type> *in, std::complex<real_type> *out) { prefix ## _execute_dft(p, (complex_type *)in, (complex_type*)out); } \
  static plan_type plan_dft_r2c_2d(int Nx, int Ny,  \
 		       real_type *in, complex_type *out, \
 		       unsigned flags) \
  { \
    return prefix ## _plan_dft_r2c_2d(Nx, Ny, in, out, \
 				flags); \
  } \
  static plan_type plan_dft_c2r_2d(int Nx, int Ny,  \
                       complex_type *in, real_type *out, \
                       unsigned flags) \
  { \
    return prefix ## _plan_dft_c2r_2d(Nx, Ny, in, out, \
                                flags); \
  } \
  static plan_type plan_dft_r2c_3d(int Nx, int Ny, int Nz, \
                                   real_type *in, complex_type *out, \
                                   unsigned flags) \
  { \
    return prefix ## _plan_dft_r2c_3d(Nx, Ny, Nz, in, out, flags); \
  } \
  static plan_type plan_dft_c2r_3d(int Nx, int Ny, int Nz, \
                                   complex_type *in, real_type *out, \
                                   unsigned flags) \
  { \
    return prefix ## _plan_dft_c2r_3d(Nx, Ny, Nz, in, out, flags); \
  } \
 \
  static plan_type plan_dft_r2c(int rank, const int *n, real_type *in, \
                                complex_type *out, unsigned flags) \
  { \
    return prefix ## _plan_dft_r2c(rank, n, in, out, flags); \
  } \
  static plan_type plan_dft_c2r(int rank, const int *n, complex_type *in, \
                                real_type *out, unsigned flags) \
  { \
    return prefix ## _plan_dft_c2r(rank, n, in, out, flags); \
  } \
  static plan_type plan_dft_3d(int Nx, int Ny, int Nz, complex_type *in, complex_type *out, int sign, unsigned flags) { \
    return prefix ## _plan_dft_3d(Nx, Ny, Nz, in, out, sign, flags); \
  } \
  static plan_type plan_dft_2d(int Nx, int Ny, complex_type *in, complex_type *out, int sign, unsigned flags) { \
    return prefix ## _plan_dft_2d(Nx, Ny, in, out, sign, flags); \
  } \
  static void destroy_plan(plan_type plan) { prefix ## _destroy_plan(plan); } \
 }
 FFTW_CALLS_BASE(double, fftw);
 FFTW_CALLS_BASE(float, fftwf);
 #undef FFTW_CALLS_BASE
-}; // namespace CosmoTool
+};
 #endif
--- a/src/fourier/fft/fftw_calls_mpi.hpp
+++ b/src/fourier/fft/fftw_calls_mpi.hpp
@ -5,140 +5,110 @@
 #include <mpi.h>
 #include <fftw3-mpi.h>
-namespace CosmoTool {
+namespace CosmoTool
 {
-  static inline void init_fftw_mpi() { fftw_mpi_init(); }
+static inline void init_fftw_mpi()
 {
  fftw_mpi_init();
 }
-  static inline void done_fftw_mpi() { fftw_mpi_cleanup(); }
+static inline void done_fftw_mpi()
 {
  fftw_mpi_cleanup();
 }
-  template <typename T>
+template<typename T> class FFTW_MPI_Calls {};
  class FFTW_MPI_Calls {};
 #define FFTW_MPI_CALLS_BASE(rtype, prefix)                                     \
  template <>                                                                  \
  class FFTW_MPI_Calls<rtype> {                                                \
  public:                                                                      \
    typedef rtype real_type;                                                   \
    typedef prefix##_complex complex_type;                                     \
    typedef prefix##_plan plan_type;                                           \
                                                                               \
    static complex_type *alloc_complex(size_t N) {                             \
      return prefix##_alloc_complex(N);                                        \
    }                                                                          \
    static real_type *alloc_real(size_t N) { return prefix##_alloc_real(N); }  \
    static void free(void *p) { fftw_free(p); }                                \
                                                                               \
    template <size_t Nd>                                                       \
    static ptrdiff_t local_size(                                               \
        std::array<ptrdiff_t, Nd> const &N, MPI_Comm comm,                     \
        ptrdiff_t *local_n0, ptrdiff_t *local_0_start) {                       \
      return prefix##_mpi_local_size(                                          \
          Nd, N.data(), comm, local_n0, local_0_start);                        \
    }                                                                          \
    static ptrdiff_t local_size_2d(                                            \
        ptrdiff_t N0, ptrdiff_t N1, MPI_Comm comm, ptrdiff_t *local_n0,        \
        ptrdiff_t *local_0_start) {                                            \
      return prefix##_mpi_local_size_2d(                                       \
          N0, N1, comm, local_n0, local_0_start);                              \
    }                                                                          \
                                                                               \
    static ptrdiff_t local_size_3d(                                            \
        ptrdiff_t N0, ptrdiff_t N1, ptrdiff_t N2, MPI_Comm comm,               \
        ptrdiff_t *local_n0, ptrdiff_t *local_0_start) {                       \
      return prefix##_mpi_local_size_3d(                                       \
          N0, N1, N2, comm, local_n0, local_0_start);                          \
    }                                                                          \
                                                                               \
    static void execute(plan_type p) { prefix##_execute(p); }                  \
    static void                                                                \
    execute_c2c(plan_type p, complex_type *in, complex_type *out) {            \
      prefix##_mpi_execute_dft(p, in, out);                                    \
    }                                                                          \
    static void execute_c2c(                                                   \
        plan_type p, std::complex<real_type> *in,                              \
        std::complex<real_type> *out) {                                        \
      prefix##_mpi_execute_dft(p, (complex_type *)in, (complex_type *)out);    \
    }                                                                          \
    static void execute_r2c(plan_type p, real_type *in, complex_type *out) {   \
      prefix##_mpi_execute_dft_r2c(p, in, out);                                \
    }                                                                          \
    static void                                                                \
    execute_c2r(plan_type p, std::complex<real_type> *in, real_type *out) {    \
      prefix##_mpi_execute_dft_c2r(p, (complex_type *)in, out);                \
    }                                                                          \
    static void execute_c2r(plan_type p, complex_type *in, real_type *out) {   \
      prefix##_mpi_execute_dft_c2r(p, in, out);                                \
    }                                                                          \
    static void                                                                \
    execute_r2c(plan_type p, real_type *in, std::complex<real_type> *out) {    \
      prefix##_mpi_execute_dft_r2c(p, in, (complex_type *)out);                \
    }                                                                          \
                                                                               \
    static plan_type plan_dft_r2c_1d(                                          \
        int n, real_type *in, complex_type *out, MPI_Comm, unsigned flags) {   \
      return prefix##_plan_dft_r2c_1d(n, in, out, flags);                      \
    }                                                                          \
                                                                               \
    static plan_type plan_dft_r2c_2d(                                          \
        int Nx, int Ny, real_type *in, complex_type *out, MPI_Comm comm,       \
        unsigned flags) {                                                      \
      return prefix##_mpi_plan_dft_r2c_2d(Nx, Ny, in, out, comm, flags);       \
    }                                                                          \
                                                                               \
    static plan_type plan_dft_c2r_1d(                                          \
        int n, complex_type *in, real_type *out, MPI_Comm, unsigned flags) {   \
      return prefix##_plan_dft_c2r_1d(n, in, out, flags);                      \
    }                                                                          \
    static plan_type plan_dft_c2r_2d(                                          \
        int Nx, int Ny, complex_type *in, real_type *out, MPI_Comm comm,       \
        unsigned flags) {                                                      \
      return prefix##_mpi_plan_dft_c2r_2d(Nx, Ny, in, out, comm, flags);       \
    }                                                                          \
                                                                               \
    static plan_type plan_dft_r2c_3d(                                          \
        int Nx, int Ny, int Nz, real_type *in, complex_type *out,              \
        MPI_Comm comm, unsigned flags) {                                       \
      return prefix##_mpi_plan_dft_r2c_3d(Nx, Ny, Nz, in, out, comm, flags);   \
    }                                                                          \
    static plan_type plan_dft_c2r_3d(                                          \
        int Nx, int Ny, int Nz, complex_type *in, real_type *out,              \
        MPI_Comm comm, unsigned flags) {                                       \
      return prefix##_mpi_plan_dft_c2r_3d(Nx, Ny, Nz, in, out, comm, flags);   \
    }                                                                          \
                                                                               \
    static plan_type plan_dft_r2c(                                             \
        int rank, const ptrdiff_t *n, real_type *in, complex_type *out,        \
        MPI_Comm comm, unsigned flags) {                                       \
      return prefix##_mpi_plan_dft_r2c(rank, n, in, out, comm, flags);         \
    }                                                                          \
    static plan_type plan_dft_c2r(                                             \
        int rank, const ptrdiff_t *n, complex_type *in, real_type *out,        \
        MPI_Comm comm, unsigned flags) {                                       \
      return prefix##_mpi_plan_dft_c2r(rank, n, in, out, comm, flags);         \
    }                                                                          \
    static plan_type plan_dft_3d(                                              \
        int Nx, int Ny, int Nz, complex_type *in, complex_type *out,           \
        MPI_Comm comm, int sign, unsigned flags) {                             \
      return prefix##_mpi_plan_dft_3d(Nx, Ny, Nz, in, out, comm, sign, flags); \
    }                                                                          \
    static plan_type plan_dft_2d(                                              \
        int Nx, int Ny, complex_type *in, complex_type *out, MPI_Comm comm,    \
        int sign, unsigned flags) {                                            \
      return prefix##_mpi_plan_dft_2d(Nx, Ny, in, out, comm, sign, flags);     \
    }                                                                          \
    static plan_type plan_dft_1d(                                              \
        int Nx, complex_type *in, complex_type *out, MPI_Comm comm, int sign,  \
        unsigned flags) {                                                      \
      return prefix##_plan_dft_1d(Nx, in, out, sign, flags);                   \
    }                                                                          \
    static void destroy_plan(plan_type plan) { prefix##_destroy_plan(plan); }  \
  }
-  FFTW_MPI_CALLS_BASE(double, fftw);
+#define FFTW_MPI_CALLS_BASE(rtype, prefix) \
-  FFTW_MPI_CALLS_BASE(float, fftwf);
+  template<>				\
 class FFTW_MPI_Calls<rtype> {		\
 public: \
  typedef rtype real_type; \
  typedef prefix ## _complex complex_type; \
  typedef prefix ## _plan plan_type; \
  \
  static complex_type *alloc_complex(size_t N) { return prefix ## _alloc_complex(N); } \
  static real_type *alloc_real(size_t N) { return prefix ## _alloc_real(N); } \
  static void free(void *p) { fftw_free(p); } \
 \
  template<size_t Nd> \
  static ptrdiff_t local_size(std::array<ptrdiff_t,Nd> const& N, MPI_Comm comm, \
                              ptrdiff_t *local_n0, ptrdiff_t *local_0_start) { \
    return prefix ## _mpi_local_size(Nd, N.data(), comm, local_n0, local_0_start); \
  } \
  static ptrdiff_t local_size_2d(ptrdiff_t N0, ptrdiff_t N1, MPI_Comm comm, \
                                ptrdiff_t *local_n0, ptrdiff_t *local_0_start) { \
    return prefix ## _mpi_local_size_2d(N0, N1, comm, local_n0, local_0_start); \
  } \
 \
  static ptrdiff_t local_size_3d(ptrdiff_t N0, ptrdiff_t N1, ptrdiff_t N2, MPI_Comm comm, \
                                ptrdiff_t *local_n0, ptrdiff_t *local_0_start) { \
    return prefix ## _mpi_local_size_3d(N0, N1, N2, comm, local_n0, local_0_start); \
  } \
 \
  static void execute(plan_type p) { prefix ## _execute(p); } \
  static void execute_c2c(plan_type p, complex_type *in, complex_type *out) { prefix ## _mpi_execute_dft(p, in, out); } \
  static void execute_c2c(plan_type p, std::complex<real_type> *in, std::complex<real_type> *out) { prefix ## _mpi_execute_dft(p, (complex_type*)in, (complex_type*)out); } \
  static void execute_r2c(plan_type p, real_type *in, complex_type *out) { prefix ## _mpi_execute_dft_r2c(p, in, out); } \
  static void execute_c2r(plan_type p, std::complex<real_type> *in, real_type *out) { prefix ## _mpi_execute_dft_c2r(p, (complex_type*)in, out); } \
  static void execute_c2r(plan_type p, complex_type *in, real_type *out) { prefix ## _mpi_execute_dft_c2r(p, in, out); } \
  static void execute_r2c(plan_type p, real_type *in, std::complex<real_type> *out) { prefix ## _mpi_execute_dft_r2c(p, in, (complex_type*)out); } \
 \
  static plan_type plan_dft_r2c_2d(int Nx, int Ny,  \
 		       real_type *in, complex_type *out, \
 		       MPI_Comm comm, unsigned flags) \
  { \
    return prefix ## _mpi_plan_dft_r2c_2d(Nx, Ny, in, out, \
 				comm, flags); \
  } \
  static plan_type plan_dft_c2r_2d(int Nx, int Ny,  \
                       complex_type *in, real_type *out, \
                       MPI_Comm comm, unsigned flags) \
  { \
    return prefix ## _mpi_plan_dft_c2r_2d(Nx, Ny, in, out, \
                                comm, flags); \
  } \
  static plan_type plan_dft_r2c_3d(int Nx, int Ny, int Nz, \
                                   real_type *in, complex_type *out, \
                                   MPI_Comm comm, unsigned flags) \
  { \
    return prefix ## _mpi_plan_dft_r2c_3d(Nx, Ny, Nz, in, out, comm, flags); \
  } \
  static plan_type plan_dft_c2r_3d(int Nx, int Ny, int Nz, \
                                   complex_type *in, real_type *out, \
                                   MPI_Comm comm, \
                                   unsigned flags) \
  { \
    return prefix ## _mpi_plan_dft_c2r_3d(Nx, Ny, Nz, in, out, comm, flags); \
  } \
 \
  static plan_type plan_dft_r2c(int rank, const ptrdiff_t *n, real_type *in, \
                                complex_type *out, MPI_Comm comm, unsigned flags) \
  { \
    return prefix ## _mpi_plan_dft_r2c(rank, n, in, out, comm, flags); \
  } \
  static plan_type plan_dft_c2r(int rank, const ptrdiff_t *n, complex_type *in, \
                                real_type *out, MPI_Comm comm, unsigned flags) \
  { \
    return prefix ## _mpi_plan_dft_c2r(rank, n, in, out, comm, flags); \
  } \
  static plan_type plan_dft_3d(int Nx, int Ny, int Nz, complex_type *in, complex_type *out, MPI_Comm comm, int sign, unsigned flags) { \
    return prefix ## _mpi_plan_dft_3d(Nx, Ny, Nz, in, out, comm, sign, flags); \
  } \
  static plan_type plan_dft_2d(int Nx, int Ny, complex_type *in, complex_type *out, MPI_Comm comm, int sign, unsigned flags) { \
    return prefix ## _mpi_plan_dft_2d(Nx, Ny, in, out, comm, sign, flags); \
  } \
  static void destroy_plan(plan_type plan) { prefix ## _destroy_plan(plan); } \
 }
 FFTW_MPI_CALLS_BASE(double, fftw);
 FFTW_MPI_CALLS_BASE(float, fftwf);
 #undef FFTW_MPI_CALLS_BASE
-}; // namespace CosmoTool
+};
 #endif
--- a/src/h5_readFlash.cpp
+++ b/src/h5_readFlash.cpp
@ -7,16 +7,16 @@ This software is a computer program whose purpose is to provide a toolbox for co
 data analysis (e.g. filters, generalized Fourier transforms, power spectra, ...)
 This software is governed by the CeCILL license under French law and
-abiding by the rules of distribution of free software.  You can  use,
+abiding by the rules of distribution of free software.  You can  use, 
 modify and/ or redistribute the software under the terms of the CeCILL
 license as circulated by CEA, CNRS and INRIA at the following URL
-"http://www.cecill.info".
+"http://www.cecill.info". 
 As a counterpart to the access to the source code and  rights to copy,
 modify and redistribute granted by the license, users are provided only
 with a limited warranty  and the software's author,  the holder of the
 economic rights,  and the successive licensors  have only  limited
-liability.
+liability. 
 In this respect, the user's attention is drawn to the risks associated
 with loading,  using,  modifying and/or developing or reproducing the
@ -25,9 +25,9 @@ that may mean  that it is complicated to manipulate,  and  that  also
 therefore means  that it is reserved for developers  and  experienced
 professionals having in-depth computer knowledge. Users are therefore
 encouraged to load and test the software's suitability as regards their
-requirements in conditions enabling the security of their systems and/or
+requirements in conditions enabling the security of their systems and/or 
-data to be ensured and,  more generally, to use and operate it in the
+data to be ensured and,  more generally, to use and operate it in the 
-same conditions as regards security.
+same conditions as regards security. 
 The fact that you are presently reading this means that you have had
 knowledge of the CeCILL license and that you accept its terms.
@ -59,8 +59,8 @@ int ipvz_out   = 5;
 /* xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx */
-/*  n*_runtime_parameters should be set by the caller to
+/*  n*_runtime_parameters should be set by the caller to 
-    the maximum number of runtime parameters to read.
+    the maximum number of runtime parameters to read. 
 */
 void h5_read_runtime_parameters
@ -100,12 +100,12 @@ void h5_read_runtime_parameters
  nreal_runtime_parameters = MAX_PARM;
  /* integer runtime parameters */
-  int_rt_parms = (int_runtime_params_t *) malloc(nint_runtime_parameters * sizeof(int_runtime_params_t));
+  int_rt_parms = (int_runtime_params_t *) malloc(nint_runtime_parameters * sizeof(int_runtime_params_t)); 
  rank = 1;
  DataSet dataset = file->openDataSet("integer runtime parameters");
-
+  
-  IntType int_rt_type = dataset.getIntType();
+  IntType int_rt_type = dataset.getIntType();  
  //int_rt_type = H5Dget_type(dataset);
  DataSpace dataspace = dataset.getSpace();
@ -123,7 +123,10 @@ void h5_read_runtime_parameters
  DataSpace memspace(rank, &dimens_1d);
  //memspace = H5Screate_simple(rank, &dimens_1d, NULL);
-  dataset.read(int_rt_parms, int_rt_type, memspace, dataspace);
+  dataset.read(int_rt_parms, int_rt_type, memspace, dataspace,
 		    H5P_DEFAULT);
  //status = H5Dread(dataset, int_rt_type, memspace, dataspace,
 	//	    H5P_DEFAULT, int_rt_parms);
  for (i = 0; i < nint_runtime_parameters; i++) {
@ -142,14 +145,14 @@ void h5_read_runtime_parameters
  /* reals */
-  real_rt_parms = (real_runtime_params_t *) malloc(nreal_runtime_parameters * sizeof(real_runtime_params_t));
+  real_rt_parms = (real_runtime_params_t *) malloc(nreal_runtime_parameters * sizeof(real_runtime_params_t)); 
-
+  
  rank = 1;
  dataset = file->openDataSet("real runtime parameters");
-  //dataset = H5Dopen(*file_identifier, "real runtime parameters");
+  //dataset = H5Dopen(*file_identifier, "real runtime parameters"); 
-
+  
  dataspace = dataset.getSpace();
-  FloatType real_rt_type = dataset.getFloatType();
+  FloatType real_rt_type = dataset.getFloatType();  
  ndims = dataspace.getSimpleExtentDims(&dimens_1d, NULL);
  //dataspace = H5Dget_space(dataset);
  //real_rt_type = H5Dget_type(dataset);
@ -164,7 +167,10 @@ void h5_read_runtime_parameters
   memspace = DataSpace(rank, &dimens_1d);
  //memspace = H5Screate_simple(rank, &dimens_1d, NULL);
-  dataset.read(real_rt_parms, real_rt_type, memspace, dataspace);
+  dataset.read(real_rt_parms, real_rt_type, memspace, dataspace,
 		    H5P_DEFAULT);
  //status = H5Dread(dataset, real_rt_type, memspace, dataspace,
 //		    H5P_DEFAULT, real_rt_parms);
  for (i = 0; i < nreal_runtime_parameters; i++) {
@ -187,7 +193,7 @@ void h5_read_runtime_parameters
      *LBox = real_runtime_parameter_values[i];
    }
    if (strncmp(real_runtime_parameter_names[i],"hubbleconstant", 14) == 0 ) {
-      *hubble = real_runtime_parameter_values[i];
+      *hubble = real_runtime_parameter_values[i];      
    }
    if (strncmp(real_runtime_parameter_names[i],"omegamatter", 11) == 0 ) {
      *omegam = real_runtime_parameter_values[i];
@ -199,7 +205,7 @@ void h5_read_runtime_parameters
      *omegalambda = real_runtime_parameter_values[i];
    }
  }
-
+  
  for (i = 0; i < nint_runtime_parameters; i++) {
    if (strncmp(int_runtime_parameter_names[i],"pt_numx",7) == 0 ) {
      *numPart = int_runtime_parameter_values[i];
@ -208,7 +214,7 @@ void h5_read_runtime_parameters
  }
 }
-
+     
 /* xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx*/
 void h5_read_flash3_particles (H5File* file,
                        int* totalparticles,
@ -235,8 +241,8 @@ void h5_read_flash3_particles (H5File* file,
  char     *propName;
  double   *partBuffer;
  char    part_names[50][OUTPUT_PROP_LENGTH];
-  int     string_size;
+  int     string_size;  
-
+         
 //  char part_names[NPART_PROPS][OUTPUT_PROP_LENGTH];
  hsize_t dimens_2d[2], maxdimens_2d[2];
  hsize_t start_2d[2], count_2d[2], stride_2d[2];
@ -259,12 +265,12 @@ void h5_read_flash3_particles (H5File* file,
  //total number of particle properties
  numProps = dimens_2d[0];
-
+  
  string_size = OUTPUT_PROP_LENGTH;
  StrType string_type = H5Tcopy(H5T_C_S1);
  string_type.setSize(string_size);
  //status = H5Tset_size(string_type, string_size);
-
+  
  rank = 2;
  start_2d[0] = 0;
@ -276,20 +282,22 @@ void h5_read_flash3_particles (H5File* file,
  count_2d[0] = dimens_2d[0];
  count_2d[1] = dimens_2d[1];
-  dataspace.selectHyperslab(H5S_SELECT_SET, count_2d, start_2d);
+  dataspace.selectHyperslab(H5S_SELECT_SET, count_2d, start_2d); 
  //status = H5Sselect_hyperslab(dataspace, H5S_SELECT_SET, start_2d,
  //                      stride_2d, count_2d, NULL);
  DataSpace memspace(rank, dimens_2d);
  //memspace = H5Screate_simple(rank, dimens_2d, NULL);
-  dataset.read(part_names, string_type);
+  dataset.read(part_names, string_type, H5S_ALL, H5S_ALL, H5P_DEFAULT);
-
+  //status = H5Dread(dataset, string_type, H5S_ALL, H5S_ALL,
  //              H5P_DEFAULT, part_names);
  string_type.close();
  memspace.close();
  dataspace.close();
-  dataset.close();
+  dataset.close(); 
  //H5Tclose(string_type);
  //H5Sclose(memspace);
  //H5Sclose(dataspace);
@ -305,7 +313,7 @@ void h5_read_flash3_particles (H5File* file,
    if (strncmp(part_names[i], "velz", 4) == 0) { ipvz = i+1; }
  }
-  if ((iptag < 0) || (ipx < 0) || (ipy < 0) || (ipz < 0) || (ipvx < 0) ||
+  if ((iptag < 0) || (ipx < 0) || (ipy < 0) || (ipz < 0) || (ipvx < 0) || 
       (ipvy < 0) || (ipvz < 0) ) {
    printf("One or more required particle attributes not found in file!\n");
    return;
@ -317,7 +325,7 @@ void h5_read_flash3_particles (H5File* file,
  //read particles
  dataset = file->openDataSet("tracer particles");
  //dataset = H5Dopen(*file_identifier, "tracer particles");
-
+  
  FloatType datatype = dataset.getFloatType();
  //datatype = H5Dget_type(dataset);
@ -330,7 +338,7 @@ void h5_read_flash3_particles (H5File* file,
  ndims = dataspace.getSimpleExtentDims(dimens_2d, NULL);
  //dataspace = H5Dget_space(dataset);
  //H5Sget_simple_extent_dims(dataspace, dimens_2d, maxdimens_2d);
-
+ 
  /*insert particle properties (numPartBuffer) particles at a time*/
  pstack = (*localnp);
  poffset = 0;
@ -358,7 +366,7 @@ void h5_read_flash3_particles (H5File* file,
    dimens_2d[0] = (pcount);
    dimens_2d[1] = (numProps);
-    dataspace.selectHyperslab(H5S_SELECT_SET, count_2d, start_2d);
+    dataspace.selectHyperslab(H5S_SELECT_SET, count_2d, start_2d); 
    //status     = H5Sselect_hyperslab(dataspace, H5S_SELECT_SET, start_2d,
    //                                 stride_2d, count_2d, NULL);
@ -366,7 +374,8 @@ void h5_read_flash3_particles (H5File* file,
    //memspace   = H5Screate_simple(rank, dimens_2d, maxdimens_2d);
    /* read data from the dataset */
-   dataset.read(partBuffer, datatype, memspace, dataspace);
+   dataset.read(partBuffer, datatype, memspace, dataspace, H5P_DEFAULT);
   //status = H5Dread(dataset, datatype, memspace, dataspace, H5P_DEFAULT, partBuffer);
    /* convert buffer into particle struct */
@ -382,8 +391,8 @@ void h5_read_flash3_particles (H5File* file,
      pos3[p+poffset]  = (float) *(partBuffer+ipz-1+p*numProps);
    }
    }
-
+   
-
+    
    if (vel1 && vel2 && vel3) {
    for(p=0; p < (pcount); p++) {
      vel1[p+poffset]  = (float) *(partBuffer+ipvx-1+p*numProps);
@ -414,7 +423,7 @@ void h5_read_flash3_particles (H5File* file,
  //status = H5Sclose(dataspace);
  //status = H5Dclose(dataset);
 }
-
+ 
 /*xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx*/
@ -442,7 +451,7 @@ void h5_read_flash3_header_info(H5File* file,
  H5std_string DATASET_NAME;
-  string_type = H5Tcopy(H5T_C_S1);
+  string_type = H5Tcopy(H5T_C_S1);  
  H5Tset_size(string_type, MAX_STRING_LENGTH);
  DataSet dataset = file->openDataSet("real scalars");
@ -458,13 +467,13 @@ void h5_read_flash3_header_info(H5File* file,
  /* malloc a pointer to a list of real_list_t's */
  real_list = (real_list_t *) malloc(dimens_1d * sizeof(real_list_t));
-  // create a new simple dataspace of 1 dimension and size of 'dimens_1d'
+  // create a new simple dataspace of 1 dimension and size of 'dimens_1d' 
  DataSpace memspace(1, &dimens_1d);
-  // create an empty vessel sized to hold one real_list_t's worth of data
+  // create an empty vessel sized to hold one real_list_t's worth of data 
  CompType real_list_type( sizeof(real_list_t) );
-  // subdivide the empty vessel into its component sections (name and value)
+  // subdivide the empty vessel into its component sections (name and value) 
  real_list_type.insertMember(
          "name",
          HOFFSET(real_list_t, name),
@ -475,8 +484,9 @@ void h5_read_flash3_header_info(H5File* file,
          HOFFSET(real_list_t, value),
          PredType::NATIVE_DOUBLE);
-  // read the data into 'real_list'
+  // read the data into 'real_list' 
-  dataset.read( real_list, real_list_type, memspace, dataspace);
+  dataset.read( real_list, real_list_type, memspace, dataspace,
                H5P_DEFAULT);
  if (status < 0) {
--- a/src/hdf5_array.hpp
+++ b/src/hdf5_array.hpp
@ -144,8 +144,8 @@ namespace CosmoTool {
                        bool useBases = false)
  {
      std::vector<hsize_t> memdims(data.shape(), data.shape() + data.num_dimensions());
-      H5::DataSpace dataspace(int(dimensions.size()), dimensions.data());
+      H5::DataSpace dataspace(dimensions.size(), dimensions.data());
-      H5::DataSpace memspace(int(memdims.size()), memdims.data());
+      H5::DataSpace memspace(memdims.size(), memdims.data());
      if (useBases) {
        std::vector<hsize_t> offsets(data.index_bases(), data.index_bases() + data.num_dimensions());
@ -398,7 +398,7 @@ namespace CosmoTool {
            hdf5_weak_check_array(data, dimensions);
            std::vector<hsize_t> memdims(data.shape(), data.shape() + data.num_dimensions());
-            H5::DataSpace memspace(int(memdims.size()), memdims.data());
+            H5::DataSpace memspace(memdims.size(), memdims.data());
            std::vector<hsize_t> offsets(data.index_bases(), data.index_bases() + data.num_dimensions());            
            dataspace.selectHyperslab(H5S_SELECT_SET, memdims.data(), offsets.data());
@ -426,7 +426,7 @@ namespace CosmoTool {
 #define CTOOL_HDF5_INSERT_ELEMENT(r, STRUCT, element) \
     { \
       ::CosmoTool::get_hdf5_data_type<BOOST_PP_TUPLE_ELEM(2, 0, element)> t; \
-       long position = offsetof(STRUCT, BOOST_PP_TUPLE_ELEM(2, 1, element)); \
+       position = HOFFSET(STRUCT, BOOST_PP_TUPLE_ELEM(2, 1, element)); \
       const char *field_name = BOOST_PP_STRINGIZE(BOOST_PP_TUPLE_ELEM(2, 1, element)); \
       type.insertMember(field_name,  position, t.type()); \
     } 
@ -439,6 +439,7 @@ namespace CosmoTool { \
  \
      TNAME() : type(sizeof(STRUCT)) \
      {  \
        long position; \
        BOOST_PP_SEQ_FOR_EACH(CTOOL_HDF5_INSERT_ELEMENT, STRUCT, ATTRIBUTES) \
      }  \
      \
@ -470,6 +471,7 @@ namespace CosmoTool { \
  \
      TNAME() : type(sizeof(STRUCT)) \
      {  \
        long position; \
        BOOST_PP_SEQ_FOR_EACH(CTOOL_HDF5_INSERT_ENUM_ELEMENT, STRUCT, ATTRIBUTES) \
      }  \
      \
--- a/src/interpolate.cpp
+++ b/src/interpolate.cpp
@ -157,7 +157,6 @@ const Interpolate& Interpolate::operator=(const Interpolate& a)
  gsl_spline_init(spline, a.spline->x, a.spline->y, a.spline->size);
  logx = a.logx;
  logy = a.logy;
  return *this;
 }
 double Interpolate::getMaxX() const
--- a/src/loadGadget.cpp
+++ b/src/loadGadget.cpp
@ -221,7 +221,7 @@ SimuData *CosmoTool::loadGadgetMulti(const char *fname, int id,
      cerr << "Invalid format while reading header" << endl;
      delete data;
      delete f;
-      throw;
+      return 0;
    }
@ -275,7 +275,7 @@ SimuData *CosmoTool::loadGadgetMulti(const char *fname, int id,
        cerr << "Invalid format while reading positions" << endl;
        delete f;
        delete data;
-        throw;
+        return 0;
      }
  } else {
@ -292,7 +292,7 @@ SimuData *CosmoTool::loadGadgetMulti(const char *fname, int id,
 	  {
 	    delete f;
 	    delete data;
-            throw;
+	    return 0;
 	  }
      }
@ -317,7 +317,7 @@ SimuData *CosmoTool::loadGadgetMulti(const char *fname, int id,
 	cerr << "Invalid format while reading velocities" << endl;
 	delete f;
 	delete data;
-        throw;
+	return 0;
      }
    // THE VELOCITIES ARE IN PHYSICAL COORDINATES
@ -367,7 +367,7 @@ SimuData *CosmoTool::loadGadgetMulti(const char *fname, int id,
 	cerr << "Invalid unformatted access while reading ID" << endl;
 	delete f;
 	delete data;
-        throw;
+	return 0;
      }
  } else {
    f->skip(2*4);
--- a/src/mykdtree.hpp
+++ b/src/mykdtree.hpp
@ -1,5 +1,5 @@
 /*+
-This is CosmoTool (./src/mykdtree.hpp) -- Copyright (C) Guilhem Lavaux (2007-2022)
+This is CosmoTool (./src/mykdtree.hpp) -- Copyright (C) Guilhem Lavaux (2007-2014)
 guilhem.lavaux@gmail.com
@ -7,16 +7,16 @@ This software is a computer program whose purpose is to provide a toolbox for co
 data analysis (e.g. filters, generalized Fourier transforms, power spectra, ...)
 This software is governed by the CeCILL license under French law and
-abiding by the rules of distribution of free software.  You can  use,
+abiding by the rules of distribution of free software.  You can  use, 
 modify and/ or redistribute the software under the terms of the CeCILL
 license as circulated by CEA, CNRS and INRIA at the following URL
-"http://www.cecill.info".
+"http://www.cecill.info". 
 As a counterpart to the access to the source code and  rights to copy,
 modify and redistribute granted by the license, users are provided only
 with a limited warranty  and the software's author,  the holder of the
 economic rights,  and the successive licensors  have only  limited
-liability.
+liability. 
 In this respect, the user's attention is drawn to the risks associated
 with loading,  using,  modifying and/or developing or reproducing the
@ -25,9 +25,9 @@ that may mean  that it is complicated to manipulate,  and  that  also
 therefore means  that it is reserved for developers  and  experienced
 professionals having in-depth computer knowledge. Users are therefore
 encouraged to load and test the software's suitability as regards their
-requirements in conditions enabling the security of their systems and/or
+requirements in conditions enabling the security of their systems and/or 
-data to be ensured and,  more generally, to use and operate it in the
+data to be ensured and,  more generally, to use and operate it in the 
-same conditions as regards security.
+same conditions as regards security. 
 The fact that you are presently reading this means that you have had
 knowledge of the CeCILL license and that you accept its terms.
@ -48,13 +48,13 @@ namespace CosmoTool {
  typedef uint64_t NodeIntType;
-  template<int N, typename CType = ComputePrecision>
+  template<int N, typename CType = ComputePrecision> 
  struct KDDef
  {
    typedef CType CoordType;
    typedef float KDCoordinates[N];
  };
-
+  
  template<int N, typename ValType, typename CType = ComputePrecision>
  struct KDCell
  {
@ -99,10 +99,10 @@ namespace CosmoTool {
    typename KDDef<N,CType>::CoordType r, r2;
    KDCell<N, ValType,CType> **cells;
    typename KDDef<N,CType>::CoordType *distances;
-    uint64_t currentRank;
+    uint32_t currentRank;
-    uint64_t numCells;
+    uint32_t numCells;
  };
-
+  
  template<int N, typename ValType, typename CType = ComputePrecision>
  class RecursionMultipleInfo
@ -114,14 +114,14 @@ namespace CosmoTool {
    RecursionMultipleInfo(const typename KDDef<N,CType>::KDCoordinates& rx,
 			  KDCell<N,ValType,CType> **cells,
-			  uint64_t numCells)
+			  uint32_t numCells)
      : queue(cells, numCells, INFINITY),traversed(0)
    {
      std::copy(rx, rx+N, x);
    }
  };
-  template<int N, typename ValType, typename CType = ComputePrecision>
+  template<int N, typename ValType, typename CType = ComputePrecision> 
  struct KD_default_cell_splitter
  {
    void operator()(KDCell<N,ValType,CType> **cells, NodeIntType Ncells, NodeIntType& split_index, int axis, typename KDDef<N,CType>::KDCoordinates minBound, typename KDDef<N,CType>::KDCoordinates maxBound);
@ -135,7 +135,7 @@ namespace CosmoTool {
    typedef typename KDDef<N>::KDCoordinates coords;
    typedef KDCell<N,ValType,CType> Cell;
    typedef KDTreeNode<N,ValType,CType> Node;
-
+    
    CellSplitter splitter;
    KDTree(Cell *cells, NodeIntType Ncells);
@ -153,20 +153,20 @@ namespace CosmoTool {
       std::copy(replicate, replicate+N, this->replicate);
     }
-    uint64_t getIntersection(const coords& x, CoordType r,
+    uint32_t getIntersection(const coords& x, CoordType r, 
 			     Cell **cells,
-			     uint64_t numCells);
+			     uint32_t numCells);
-    uint64_t getIntersection(const coords& x, CoordType r,
+    uint32_t getIntersection(const coords& x, CoordType r, 
 			     Cell **cells,
 			     CoordType *distances,
-			     uint64_t numCells);
+			     uint32_t numCells);
-    uint64_t countCells(const coords& x, CoordType r);
+    uint32_t countCells(const coords& x, CoordType r);
    Cell *getNearestNeighbour(const coords& x);
-    void getNearestNeighbours(const coords& x, uint64_t NumCells,
+    void getNearestNeighbours(const coords& x, uint32_t NumCells,
 			      Cell **cells);
-    void getNearestNeighbours(const coords& x, uint64_t NumCells,
+    void getNearestNeighbours(const coords& x, uint32_t NumCells,
 			      Cell **cells,
 			      CoordType *distances);
@ -183,7 +183,7 @@ namespace CosmoTool {
    NodeIntType getNumberInNode(const Node *n) const { return n->numNodes; }
 #else
    NodeIntType getNumberInNode(const Node *n) const {
-      if (n == 0)
+      if (n == 0) 
        return 0;
      return 1+getNumberInNode(n->children[0])+getNumberInNode(n->children[1]);
    }
@ -211,7 +211,7 @@ namespace CosmoTool {
 		    uint32_t depth,
 		    coords minBound,
 		    coords maxBound);
-
+    
    template<bool justCount>
    void recursiveIntersectionCells(RecursionInfoCells<N,ValType, CType>& info,
 				    Node *node,
@ -224,7 +224,7 @@ namespace CosmoTool {
 			  CoordType& R2,
 			  Cell*& cell);
    void recursiveMultipleNearest(RecursionMultipleInfo<N,ValType,CType>& info, Node *node,
-				  int level);
+				  int level);    
  };
--- a/src/mykdtree.tcc
+++ b/src/mykdtree.tcc
@ -33,7 +33,7 @@ namespace CosmoTool {
  template<int N, typename ValType, typename CType, typename CellSplitter>
  KDTree<N,ValType,CType,CellSplitter>::KDTree(Cell *cells, NodeIntType Ncells)
  {
-    periodic = false;
+    periodic = false; 
    base_cell = cells;
    numNodes = Ncells;
@ -41,7 +41,7 @@ namespace CosmoTool {
    sortingHelper = new Cell *[Ncells];
    for (NodeIntType i = 0; i < Ncells; i++)
-	sortingHelper[i] = &cells[i];
+	sortingHelper[i] = &cells[i];    
    optimize();
  }
@ -73,16 +73,16 @@ namespace CosmoTool {
            absoluteMax[k] = cell->coord[k];
        }
      }
-
+    
    std::cout << "   rebuilding the tree..." << std::endl;
-    root = buildTree(sortingHelper, activeCells, 0, absoluteMin, absoluteMax);
+    root = buildTree(sortingHelper, activeCells, 0, absoluteMin, absoluteMax);    
    std::cout << "   done." << std::endl;
  }
  template<int N, typename ValType, typename CType, typename CellSplitter>
-  uint64_t KDTree<N,ValType,CType,CellSplitter>::getIntersection(const coords& x, CoordType r,
+  uint32_t KDTree<N,ValType,CType,CellSplitter>::getIntersection(const coords& x, CoordType r, 
 						    KDTree<N,ValType,CType,CellSplitter>::Cell **cells,
-						    uint64_t numCells)
+						    uint32_t numCells)
  {
    RecursionInfoCells<N,ValType,CType> info;
@ -112,10 +112,10 @@ namespace CosmoTool {
  }
  template<int N, typename ValType, typename CType, typename CellSplitter>
-  uint64_t KDTree<N,ValType,CType,CellSplitter>::getIntersection(const coords& x, CoordType r,
+  uint32_t KDTree<N,ValType,CType,CellSplitter>::getIntersection(const coords& x, CoordType r, 
 						    Cell **cells,
 						    CoordType *distances,
-						    uint64_t numCells)
+						    uint32_t numCells)
  {
    RecursionInfoCells<N,ValType> info;
@ -144,7 +144,7 @@ namespace CosmoTool {
  }
  template<int N, typename ValType, typename CType, typename CellSplitter>
-  uint64_t KDTree<N,ValType,CType,CellSplitter>::countCells(const coords& x, CoordType r)
+  uint32_t KDTree<N,ValType,CType,CellSplitter>::countCells(const coords& x, CoordType r)
  {
    RecursionInfoCells<N,ValType> info;
@ -175,7 +175,7 @@ namespace CosmoTool {
  template<int N, typename ValType, typename CType, typename CellSplitter>
  template<bool justCount>
-  void KDTree<N,ValType,CType,CellSplitter>::recursiveIntersectionCells(RecursionInfoCells<N,ValType,CType>& info,
+  void KDTree<N,ValType,CType,CellSplitter>::recursiveIntersectionCells(RecursionInfoCells<N,ValType,CType>& info, 
 							   Node *node,
 							   int level)
  {
@ -183,7 +183,7 @@ namespace CosmoTool {
    CoordType d2 = 0;
 #if __KD_TREE_ACTIVE_CELLS == 1
-    if (node->value->active)
+    if (node->value->active) 
 #endif
      {
 	for (int j = 0; j < 3; j++)
@ -250,9 +250,9 @@ namespace CosmoTool {
  }
  template<int N, typename ValType, typename CType>
-  void KD_default_cell_splitter<N,ValType,CType>::operator()(KDCell<N,ValType,CType> **cells, NodeIntType Ncells,
+  void KD_default_cell_splitter<N,ValType,CType>::operator()(KDCell<N,ValType,CType> **cells, NodeIntType Ncells, 
                                                             NodeIntType& split_index, int axis,
-                                                             typename KDDef<N,CType>::KDCoordinates minBound,
+                                                             typename KDDef<N,CType>::KDCoordinates minBound, 
                                                             typename KDDef<N,CType>::KDCoordinates maxBound)
  {
    CellCompare<N,ValType,CType> compare(axis);
@ -279,9 +279,9 @@ namespace CosmoTool {
 //#pragma omp atomic capture
    nodeId = (this->lastNode)++;
-
+    
    node = &nodes[nodeId];
-
+    
    // Isolate the environment
    splitter(cell0, Ncells, mid, axis, minBound, maxBound);
@ -297,12 +297,12 @@ namespace CosmoTool {
    {
      node->children[0] = buildTree(cell0, mid, depth, minBound, tmpBound);
    }
-
+    
    memcpy(tmpBound, minBound, sizeof(coords));
    tmpBound[axis] = node->value->coord[axis];
 #pragma omp task private(tmpBound)
    {
-      node->children[1] = buildTree(cell0+mid+1, Ncells-mid-1, depth,
+      node->children[1] = buildTree(cell0+mid+1, Ncells-mid-1, depth, 
 				  tmpBound, maxBound);
    }
@ -391,17 +391,17 @@ namespace CosmoTool {
 	  }
 	return;
      }
-
+    
    // Check if current node is not the nearest
-    CoordType thisR2 =
+    CoordType thisR2 = 
      computeDistance(node->value, x);
-
+    
    if (thisR2 < R2)
      {
 	R2 = thisR2;
 	best = node->value;
-      }
+      }    
-
+    
    // Now we found the best. We check whether the hypersphere
    // intersect the hyperplane of the other branch
@ -435,11 +435,11 @@ namespace CosmoTool {
         {
           coords x_new;
           r.getPosition(x_new);
-           recursiveNearest(root, 0, x_new, R2, best);
+           recursiveNearest(root, 0, x_new, R2, best); 
         }
       while (r.next());
     }
-
+   
   return best;
 }
@ -474,15 +474,15 @@ namespace CosmoTool {
      {
 	recursiveMultipleNearest(info, go, level+1);
      }
-
+    
    // Check if current node is not the nearest
-    CoordType thisR2 =
+    CoordType thisR2 = 
      computeDistance(node->value, info.x);
    info.queue.push(node->value, thisR2);
    info.traversed++;
    //    if (go == 0)
    //      return;
-
+    
    // Now we found the best. We check whether the hypersphere
    // intersect the hyperplane of the other branch
@ -497,15 +497,15 @@ namespace CosmoTool {
 	  {
 	    recursiveMultipleNearest(info, other, level+1);
 	  }
-      }
+      }    
  }
 template<int N, typename ValType, typename CType, typename CellSplitter>
- void KDTree<N,ValType,CType,CellSplitter>::getNearestNeighbours(const coords& x, uint64_t N2,
+ void KDTree<N,ValType,CType,CellSplitter>::getNearestNeighbours(const coords& x, uint32_t N2,
 						    Cell **cells)
 {
   RecursionMultipleInfo<N,ValType> info(x, cells, N2);
-
+   
   for (int i = 0; i < N2; i++)
     cells[i] = 0;
@ -527,12 +527,12 @@ namespace CosmoTool {
 }
 template<int N, typename ValType, typename CType, typename CellSplitter>
- void KDTree<N,ValType,CType,CellSplitter>::getNearestNeighbours(const coords& x, uint64_t N2,
+ void KDTree<N,ValType,CType,CellSplitter>::getNearestNeighbours(const coords& x, uint32_t N2,
 						    Cell **cells,
 						    CoordType *distances)
 {
   RecursionMultipleInfo<N,ValType> info(x, cells, N2);
-
+   
   for (int i = 0; i < N2; i++)
     cells[i] = 0;
@ -555,14 +555,14 @@ namespace CosmoTool {
 #ifdef __KD_TREE_SAVE_ON_DISK
 #define KDTREE_DISK_SIGNATURE "KDTREE"
 #define KDTREE_DISK_SIGNATURE_LEN 7
-
+  
  template<int N, typename CType>
  struct KDTreeOnDisk
  {
    long cell_id;
    long children_node[2];
    typename KDDef<N, CType>::KDCoordinates minBound, maxBound;
-  };
+  };  
  struct KDTreeHeader
  {
@ -619,7 +619,7 @@ namespace CosmoTool {
      {
         std::cerr << "KDTree Signature invalid" << std::endl;
         throw InvalidOnDiskKDTree();
-      }
+      } 
    if (h.numCells != Ncells || h.nodesUsed < 0) {
      std::cerr << "The number of cells has changed (" << h.numCells << " != " << Ncells << ") or nodesUsed=" << h.nodesUsed << std::endl;
@ -643,8 +643,8 @@ namespace CosmoTool {
           throw InvalidOnDiskKDTree();
        }
        if (node_on_disk.cell_id > numNodes || node_on_disk.cell_id < 0 ||
-	    (node_on_disk.children_node[0] >= 0 && node_on_disk.children_node[0] > lastNode) || node_on_disk.children_node[0] < -1 ||
+	    node_on_disk.children_node[0] > lastNode || node_on_disk.children_node[0] < -1 ||
-	    (node_on_disk.children_node[1] >= 0 && node_on_disk.children_node[1] > lastNode) || node_on_disk.children_node[1] < -1)
+	    node_on_disk.children_node[1] > lastNode || node_on_disk.children_node[1] < -1)
 	  {
 	    delete[] nodes;
            std::cerr << "Invalid cell id or children node id invalid" << std::endl;
@ -683,10 +683,10 @@ namespace CosmoTool {
      }
    root = &nodes[h.rootId];
-
+    
    sortingHelper = new Cell *[Ncells];
    for (NodeIntType i = 0; i < Ncells; i++)
-      sortingHelper[i] = &cells[i];
+      sortingHelper[i] = &cells[i];    
  }
 #endif
--- a/src/numpy_adaptors.hpp
+++ b/src/numpy_adaptors.hpp
@ -1,30 +0,0 @@
 #ifndef __COSMOTOOL_NUMPY_ADAPTOR_HPP
 #define __COSMOTOOL_NUMPY_ADAPTOR_HPP
 namespace CosmoTool {
  template<typename T, typename IT>
  void parallel_ufunc_dd_d(char **args, IT* dimensions, IT* steps, void *func) {
    IT i;
    IT n = dimensions[0];
    char *in = args[0], *in2 = args[1], *out = args[2];
    IT in_step = steps[0], in2_step = steps[1], out_step = steps[2];
    double tmp;
    typedef double (*F_t)(double,double);
    F_t f = (F_t)func;
 #pragma omp parallel for schedule(static)
    for (i = 0; i < n; i++) {
        T *out_t = (T *)(out + i * out_step);
        T *in_t = (T *)(in + i * in_step);
        T *in2_t = (T *)(in2 + i * in2_step);
        *out_t = f(*in_t, *in2_t);
    }
  }
 }
 #endif
--- a/src/powerSpectrum.cpp
+++ b/src/powerSpectrum.cpp
@ -58,7 +58,6 @@ using namespace std;
 #define POWER_TEST 8
 #define POWER_SPECTRUM POWER_EFSTATHIOU
 #define SAMPLE_WIGGLES 9
 namespace Cosmology {
@ -207,31 +206,6 @@ double powG(double y)
 */
 double powerSpectrum(double k, double normPower)
 {
 #if POWER_SPECTRUM == SAMPLE_WIGGLES
  // BAO wiggle parameterization for reconstruction
  // Babic et al. 2022, https://arxiv.org/abs/2203.06177
  // No-wiggle transfer function
  double s = 44.5 * log(9.83 / OmegaEff) / (sqrt(1 + 10 * pow(OMEGA_B * h * h, 0.75)));
  double alpha_Gamma = 1 - 0.328 * log(431 * OmegaEff) * OMEGA_B / OMEGA_0 + 0.38 * log(22.3 * OmegaEff) * pow(OMEGA_B / OMEGA_0, 2);
  double GammaEff = OMEGA_0 * h * (alpha_Gamma + (1 - alpha_Gamma)/(1 + pow(0.43 * k * s, 4)));
  double q = k/(h*GammaEff) * pow(Theta_27, 2);
  double L_0 = log(2 * M_E + 1.8 * q);
  double C_0 = 14.2 + 731 / (1 + 62.5 * q);
  double T0 = L_0 / (L_0 + C_0 * q * q);
  // Wiggle parameterization
  double A = 0;
  double r_s = 10;
  double k_D = 2 * M_PI / 100;
  double param = 1 + A * sin(k * r_s) * exp(- k / k_D);
  return normPower * pow(k, n) * T0 * T0 * param;
 #endif
 #if POWER_SPECTRUM == POWER_EFSTATHIOU
  double a = 6.4/Gamma0;
  double b = 3/Gamma0;
--- a/src/sphSmooth.hpp
+++ b/src/sphSmooth.hpp
@ -1,5 +1,5 @@
 /*+
-This is CosmoTool (./src/sphSmooth.hpp) -- Copyright (C) Guilhem Lavaux (2007-2022)
+This is CosmoTool (./src/sphSmooth.hpp) -- Copyright (C) Guilhem Lavaux (2007-2014)
 guilhem.lavaux@gmail.com
@ -7,16 +7,16 @@ This software is a computer program whose purpose is to provide a toolbox for co
 data analysis (e.g. filters, generalized Fourier transforms, power spectra, ...)
 This software is governed by the CeCILL license under French law and
-abiding by the rules of distribution of free software.  You can  use,
+abiding by the rules of distribution of free software.  You can  use, 
 modify and/ or redistribute the software under the terms of the CeCILL
 license as circulated by CEA, CNRS and INRIA at the following URL
-"http://www.cecill.info".
+"http://www.cecill.info". 
 As a counterpart to the access to the source code and  rights to copy,
 modify and redistribute granted by the license, users are provided only
 with a limited warranty  and the software's author,  the holder of the
 economic rights,  and the successive licensors  have only  limited
-liability.
+liability. 
 In this respect, the user's attention is drawn to the risks associated
 with loading,  using,  modifying and/or developing or reproducing the
@ -25,9 +25,9 @@ that may mean  that it is complicated to manipulate,  and  that  also
 therefore means  that it is reserved for developers  and  experienced
 professionals having in-depth computer knowledge. Users are therefore
 encouraged to load and test the software's suitability as regards their
-requirements in conditions enabling the security of their systems and/or
+requirements in conditions enabling the security of their systems and/or 
-data to be ensured and,  more generally, to use and operate it in the
+data to be ensured and,  more generally, to use and operate it in the 
-same conditions as regards security.
+same conditions as regards security. 
 The fact that you are presently reading this means that you have had
 knowledge of the CeCILL license and that you accept its terms.
@ -39,89 +39,80 @@ knowledge of the CeCILL license and that you accept its terms.
 #include "config.hpp"
 #include "mykdtree.hpp"
-namespace CosmoTool {
+namespace CosmoTool
 {
  template <typename ValType, int Ndims = NUMDIMS>
-  class SPHSmooth {
+  class SPHSmooth
  {
  public:
-    typedef struct {
+    typedef struct
    {
      ComputePrecision weight;
      ValType pValue;
    } FullType;
-    typedef KDTree<Ndims, FullType> SPHTree;
+    typedef KDTree<Ndims,FullType> SPHTree;
-    typedef KDTreeNode<Ndims, FullType> SPHNode;
+    typedef KDTreeNode<Ndims,FullType> SPHNode;
-    typedef KDCell<Ndims, FullType> SPHCell;
+    typedef KDCell<Ndims,FullType>  SPHCell;
-    typedef typename KDTree<Ndims, FullType>::CoordType CoordType;
+    typedef typename KDTree<Ndims,FullType>::CoordType CoordType;
-    typedef ComputePrecision (*computeParticleValue)(const ValType &t);
+    typedef ComputePrecision (*computeParticleValue)(const ValType& t);
-    typedef void (*runParticleValue)(ValType &t);
+    typedef void (*runParticleValue)(ValType& t);
  public:
    typedef SPHCell *P_SPHCell;
-    struct SPHState {
+    struct SPHState
    {
      boost::shared_ptr<P_SPHCell[]> ngb;
      boost::shared_ptr<CoordType[]> distances;
      typename SPHTree::coords currentCenter;
      int currentNgb;
      ComputePrecision smoothRadius;
    };
-
+    
    SPHSmooth(SPHTree *tree, uint32_t Nsph);
-    virtual ~SPHSmooth();
+    virtual ~SPHSmooth();   
-    void
+    void fetchNeighbours(const typename SPHTree::coords& c, SPHState *state = 0);
-    fetchNeighbours(const typename SPHTree::coords &c, SPHState *state = 0);
+    
-
+    void fetchNeighbours(const typename SPHTree::coords& c, uint32_t newNsph);
-    void fetchNeighbours(
+    void fetchNeighboursOnVolume(const typename SPHTree::coords& c, ComputePrecision radius);
-        const typename SPHTree::coords &c, uint32_t newNsph,
+    const typename SPHTree::coords& getCurrentCenter() const
-        SPHState *state = 0);
+    {
    void fetchNeighboursOnVolume(
        const typename SPHTree::coords &c, ComputePrecision radius);
    const typename SPHTree::coords &getCurrentCenter() const {
      return internal.currentCenter;
    }
-    /** This is the pure SPH smoothing function. It does not reweight by the
+    template<typename FuncT>
-     * value computed at each grid site.
+    ComputePrecision computeSmoothedValue(const typename SPHTree::coords& c,
-     */
+					  FuncT fun, SPHState *state = 0);    
-    template <typename FuncT>
+	
-    ComputePrecision computeSmoothedValue(
+	template<typename FuncT>
-        const typename SPHTree::coords &c, FuncT fun, SPHState *state = 0);
+    ComputePrecision computeInterpolatedValue(const typename SPHTree::coords& c,
 					      FuncT fun, SPHState *state = 0);
    ComputePrecision getMaxDistance(const typename SPHTree::coords& c, 
 				    SPHNode *node) const;
-    /** This is the weighted SPH smoothing function. It does reweight by the
+    ComputePrecision getSmoothingLen() const
-     * value computed at each grid site. This ensures the total sum of the interpolated
+    {
-     * quantity is preserved by interpolating to the target mesh.
+      return internal.smoothRadius;
-     */
+    }
    template <typename FuncT>
    ComputePrecision computeInterpolatedValue(
        const typename SPHTree::coords &c, FuncT fun, SPHState *state = 0);
    /** This is the adjoint gradient of computeInterpolatedValue w.r.t. to the value
     * array. FuncT is expected to have the following prototype:
     *    void((CellValue defined by the user), ComputePrecision weighted_ag_value)
     */
    template <typename FuncT>
    void computeAdjointGradientSmoothedValue(
        const typename SPHTree::coords &c, ComputePrecision ag_value, FuncT fun,
        SPHState *state = 0);
    ComputePrecision
    getMaxDistance(const typename SPHTree::coords &c, SPHNode *node) const;
    ComputePrecision getSmoothingLen() const { return internal.smoothRadius; }
    // TO USE WITH EXTREME CARE !
-    void setSmoothingLen(ComputePrecision len) { internal.smoothRadius = len; }
+    void setSmoothingLen(ComputePrecision len)
    {
      internal.smoothRadius = len;
    }
    // END
-    template <typename FuncT>
+    template<typename FuncT>
    void runForEachNeighbour(FuncT fun, SPHState *state = 0);
-    void addGridSite(const typename SPHTree::coords &c, SPHState *state);
+    void addGridSite(const typename SPHTree::coords& c);
    void addGridSite(const typename SPHTree::coords &c);
    bool hasNeighbours() const;
-    virtual ComputePrecision getKernel(ComputePrecision d) const;
+    virtual ComputePrecision getKernel(ComputePrecision d) const;    
    SPHTree *getTree() { return tree; }
@ -134,29 +125,29 @@ namespace CosmoTool {
    uint32_t getCurrent() const { return internal.currentNgb; }
    uint32_t getNgb() const { return maxNgb; }
-
+ 
  protected:
    SPHState internal;
    uint32_t Nsph;
    uint32_t deltaNsph;
    uint32_t maxNgb;
    SPHTree *tree;
-
+    
-    template <typename FuncT>
+    template<typename FuncT>
-    ComputePrecision computeWValue(
+    ComputePrecision computeWValue(const typename SPHTree::coords & c,
-        const typename SPHTree::coords &c, SPHCell &cell, CoordType d,
+				   SPHCell& cell,
-        FuncT fun, SPHState *state);
+				   CoordType d,
-
+				   FuncT fun, SPHState *state);
-    template <typename FuncT>
+    
-    void runUnrollNode(SPHNode *node, FuncT fun);
+    template<typename FuncT>
    void runUnrollNode(SPHNode *node,
 		       FuncT fun);
  };
-  template <class ValType1, class ValType2, int Ndims>
+  template<class ValType1, class ValType2, int Ndims>
-  bool operator<(
+  bool operator<(const SPHSmooth<ValType1, Ndims>& s1, const SPHSmooth<ValType2, Ndims>& s2);
      const SPHSmooth<ValType1, Ndims> &s1,
      const SPHSmooth<ValType2, Ndims> &s2);
-}; // namespace CosmoTool
+};
 #include "sphSmooth.tcc"
--- a/src/sphSmooth.tcc
+++ b/src/sphSmooth.tcc
@ -3,248 +3,225 @@
 namespace CosmoTool {
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  SPHSmooth<ValType, Ndims>::SPHSmooth(SPHTree *tree, uint32_t Nsph) {
+SPHSmooth<ValType,Ndims>::SPHSmooth(SPHTree *tree, uint32_t Nsph)
-    this->Nsph = Nsph;
+{
-    this->tree = tree;
+  this->Nsph = Nsph;
-    internal.currentNgb = 0;
+  this->tree = tree;
  internal.currentNgb = 0;
-    this->maxNgb = Nsph;
+  this->maxNgb = Nsph;
-    internal.ngb = boost::shared_ptr<SPHCell *[]>(new SPHCell *[maxNgb]);
+  internal.ngb = boost::shared_ptr<SPHCell *[]>(new SPHCell *[maxNgb]);  
-    internal.distances = boost::shared_ptr<CoordType[]>(new CoordType[maxNgb]);
+  internal.distances = boost::shared_ptr<CoordType[]>(new CoordType[maxNgb]);
-  }
+}
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  SPHSmooth<ValType, Ndims>::~SPHSmooth() {}
+SPHSmooth<ValType,Ndims>::~SPHSmooth()
 {
 }
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  template <typename FuncT>
+template<typename FuncT>
-  ComputePrecision SPHSmooth<ValType, Ndims>::computeWValue(
+ComputePrecision SPHSmooth<ValType,Ndims>::computeWValue(const typename SPHTree::coords& c,
-      const typename SPHTree::coords &c, SPHCell &cell, CoordType d, FuncT fun,
+							 SPHCell& cell,
-      SPHState *state) {
+							 CoordType d,
-    CoordType weight;
+							 FuncT fun, SPHState *state)
 {
  CoordType weight;
-    d /= state->smoothRadius;
+  d /= state->smoothRadius;
-    weight = getKernel(d);
+  weight = getKernel(d);
-    if (cell.val.weight != 0)
+  if (cell.val.weight != 0)
-      return weight * fun(cell.val.pValue) / cell.val.weight;
+    return weight * fun(cell.val.pValue) / cell.val.weight;
-    else
+  else
-      return 0;
+    return 0;
-  }
+}
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  void SPHSmooth<ValType, Ndims>::fetchNeighbours(
+void
-      const typename SPHTree::coords &c, uint32_t newNngb, SPHState *state) {
+SPHSmooth<ValType,Ndims>::fetchNeighbours(const typename SPHTree::coords& c, uint32_t newNngb)
-    ComputePrecision d2, max_dist = 0;
+{
-    uint32_t requested = newNngb;
+  ComputePrecision d2, max_dist = 0;
  uint32_t requested = newNngb;
-    if (state != 0) {
+  if (requested > maxNgb)
-      state->distances = boost::shared_ptr<CoordType[]>(new CoordType[newNngb]);
+    {
-      state->ngb = boost::shared_ptr<SPHCell *[]>(new SPHCell *[newNngb]);
+      maxNgb = requested;
-    } else {
+      internal.ngb = boost::shared_ptr<P_SPHCell[]>(new P_SPHCell[maxNgb]);
-      state = &internal;
+      internal.distances = boost::shared_ptr<CoordType[]>(new CoordType[maxNgb]);
      if (requested > maxNgb) {
        maxNgb = requested;
        internal.ngb = boost::shared_ptr<P_SPHCell[]>(new P_SPHCell[maxNgb]);
        internal.distances =
            boost::shared_ptr<CoordType[]>(new CoordType[maxNgb]);
      }
    }
-    memcpy(state->currentCenter, c, sizeof(c));
+  memcpy(internal.currentCenter, c, sizeof(c));
-    tree->getNearestNeighbours(
+  tree->getNearestNeighbours(c, requested, (SPHCell **)internal.ngb.get(), (CoordType*)internal.distances.get());
        c, requested, (SPHCell **)state->ngb.get(),
        (CoordType *)state->distances.get());
-    state->currentNgb = 0;
+  internal.currentNgb = 0;
-    for (uint32_t i = 0; i < requested && (state->ngb)[i] != 0;
+  for (uint32_t i = 0; i < requested && (internal.ngb)[i] != 0; i++,internal.currentNgb++)
-         i++, state->currentNgb++) {
+    {
-      state->distances[i] = sqrt(state->distances[i]);
+      internal.distances[i] = sqrt(internal.distances[i]);
-      d2 = state->distances[i];
+      d2 = internal.distances[i];
      if (d2 > max_dist)
        max_dist = d2;
    }
-    state->smoothRadius = max_dist / 2;
+  internal.smoothRadius = max_dist / 2;  
-  }
+}
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  void SPHSmooth<ValType, Ndims>::fetchNeighbours(
+void SPHSmooth<ValType,Ndims>::fetchNeighbours(const typename SPHTree::coords& c, SPHState *state)
-      const typename SPHTree::coords &c, SPHState *state) {
+{
-    ComputePrecision d2, max_dist = 0;
+  ComputePrecision d2, max_dist = 0;
-    uint32_t requested = Nsph;
+  uint32_t requested = Nsph;
-    if (state != 0) {
+  if (state != 0) {
-      state->distances = boost::shared_ptr<CoordType[]>(new CoordType[Nsph]);
+    state->distances = boost::shared_ptr<CoordType[]>(new CoordType[Nsph]);
-      state->ngb = boost::shared_ptr<SPHCell *[]>(new SPHCell *[Nsph]);
+    state->ngb = boost::shared_ptr<SPHCell *[]>(new SPHCell *[Nsph]);
-    } else
+  } else
-      state = &internal;
+    state = &internal;
  memcpy(state->currentCenter, c, sizeof(c));
  tree->getNearestNeighbours(c, requested, state->ngb.get(), state->distances.get());
-    memcpy(state->currentCenter, c, sizeof(c));
+  state->currentNgb = 0;
-
+  for (uint32_t i = 0; i < requested && state->ngb[i] != 0; i++,state->currentNgb++)
-    tree->getNearestNeighbours(
+    {
        c, requested, state->ngb.get(), state->distances.get());
    state->currentNgb = 0;
    for (uint32_t i = 0; i < requested && state->ngb[i] != 0;
         i++, state->currentNgb++) {
      d2 = state->distances[i] = sqrt(state->distances[i]);
      if (d2 > max_dist)
        max_dist = d2;
    }
    state->smoothRadius = max_dist / 2;
  }
  template <typename ValType, int Ndims>
  void SPHSmooth<ValType, Ndims>::fetchNeighboursOnVolume(
      const typename SPHTree::coords &c, ComputePrecision radius) {
    uint32_t numPart;
    ComputePrecision d2, max_dist = 0;
    memcpy(internal.currentCenter, c, sizeof(c));
    internal.currentNgb = tree->getIntersection(
        c, radius, internal.ngb, internal.distances, maxNgb);
    for (uint32_t i = 0; i < internal.currentNgb; i++) {
      d2 = internal.distances[i] = sqrt(internal.distances[i]);
      if (d2 > max_dist)
        max_dist = d2;
    }
    internal.smoothRadius = max_dist / 2;
  }
-  template <typename ValType, int Ndims>
+  state->smoothRadius = max_dist / 2;  
-  template <typename FuncT>
+}
  ComputePrecision SPHSmooth<ValType, Ndims>::computeSmoothedValue(
      const typename SPHTree::coords &c, FuncT fun, SPHState *state) {
    if (state == 0)
      state = &internal;
    ComputePrecision outputValue = 0;
    ComputePrecision max_dist = 0;
    ComputePrecision r3 = cube(state->smoothRadius);
-    for (uint32_t i = 0; i < state->currentNgb; i++) {
+template<typename ValType, int Ndims>
-      outputValue +=
+void
-          computeWValue(c, *state->ngb[i], state->distances[i], fun, state);
+SPHSmooth<ValType,Ndims>::fetchNeighboursOnVolume(const typename SPHTree::coords& c,
                                                  ComputePrecision radius)
 {
  uint32_t numPart;
  ComputePrecision d2, max_dist = 0;
  memcpy(internal.currentCenter, c, sizeof(c));
  internal.currentNgb = tree->getIntersection(c, radius, internal.ngb, internal.distances,
 				  maxNgb);
  for (uint32_t i = 0; i < internal.currentNgb; i++)
    {
      d2 = internal.distances[i] = sqrt(internal.distances[i]);
      if (d2 > max_dist)
        max_dist = d2;
    }
  internal.smoothRadius = max_dist / 2;  
 }				
 template<typename ValType, int Ndims>
 template<typename FuncT>
 ComputePrecision 
 SPHSmooth<ValType,Ndims>::computeSmoothedValue(const typename SPHTree::coords& c,
                                               FuncT fun, SPHState *state)
 {
  if (state == 0)
    state = &internal;
  ComputePrecision outputValue = 0;
  ComputePrecision max_dist = 0;
  ComputePrecision r3 = cube(state->smoothRadius);
  for (uint32_t i = 0; i < state->currentNgb; i++)
    {
      outputValue += computeWValue(c, *state->ngb[i], state->distances[i], fun, state);
    }
-    return outputValue / r3;
+  return outputValue / r3;
-  }
+}
-  template <typename ValType>
+template<typename ValType>
-  ComputePrecision interpolateOne(const ValType &t) {
+ComputePrecision interpolateOne(const ValType& t)
-    return 1.0;
+{
-  }
+  return 1.0;
 }
-  template <typename ValType, int Ndims>
+// WARNING ! Cell's weight must be 1 !!!
-  template <typename FuncT>
+template<typename ValType, int Ndims>
-  void SPHSmooth<ValType, Ndims>::computeAdjointGradientSmoothedValue(
+template<typename FuncT>
-      const typename SPHTree::coords &c, ComputePrecision ag_value, FuncT fun,
+ComputePrecision SPHSmooth<ValType,Ndims>::computeInterpolatedValue(const typename SPHTree::coords& c,
-      SPHState *state) {
+								    FuncT fun, SPHState *state)
-    if (state == 0)
+{
-      state = &internal;
+  if (state == 0)
    state = &internal;
  ComputePrecision outputValue = 0;
  ComputePrecision max_dist = 0;
  ComputePrecision weight = 0;
-    ComputePrecision outputValue = 0;
+  for (uint32_t i = 0; i < state->currentNgb; i++)
-    ComputePrecision max_dist = 0;
+    {
    ComputePrecision weight = 0;
    for (uint32_t i = 0; i < state->currentNgb; i++) {
      weight +=
          computeWValue(c, *state->ngb[i], state->distances[i], interpolateOne);
    }
    for (uint32_t i = 0; i < state->currentNgb; i++) {
      auto &cell = *state->ngb[i];
      double partial_ag =
          computeWValue(
              c, cell, state->distances[i],
              [ag_value](ComputePrecision) { return ag_value; }) /
          weight;
      fun(cell.val.pValue, ag_value);
    }
  }
  // WARNING ! Cell's weight must be 1 !!!
  template <typename ValType, int Ndims>
  template <typename FuncT>
  ComputePrecision SPHSmooth<ValType, Ndims>::computeInterpolatedValue(
      const typename SPHTree::coords &c, FuncT fun, SPHState *state) {
    if (state == 0)
      state = &internal;
    ComputePrecision outputValue = 0;
    ComputePrecision max_dist = 0;
    ComputePrecision weight = 0;
    for (uint32_t i = 0; i < state->currentNgb; i++) {
      outputValue += computeWValue(c, *state->ngb[i], state->distances[i], fun);
-      weight +=
+      weight += computeWValue(c, *state->ngb[i], state->distances[i], interpolateOne);
          computeWValue(c, *state->ngb[i], state->distances[i], interpolateOne);
    }
-    return (outputValue == 0) ? 0 : (outputValue / weight);
+  return (outputValue == 0) ? 0 : (outputValue / weight);  
-  }
+}
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  template <typename FuncT>
+template<typename FuncT>
-  void
+void SPHSmooth<ValType,Ndims>::runForEachNeighbour(FuncT fun, SPHState *state)
-  SPHSmooth<ValType, Ndims>::runForEachNeighbour(FuncT fun, SPHState *state) {
+{
-    if (state == 0)
+  if (state == 0)
-      state = &internal;
+    state = &internal;
-
+    
-    for (uint32_t i = 0; i < state->currentNgb; i++) {
+  for (uint32_t i = 0; i < state->currentNgb; i++)
    {
      fun(state->ngb[i]);
    }
-  }
+}
  template <typename ValType, int Ndims>
  void SPHSmooth<ValType, Ndims>::addGridSite(
      const typename SPHTree::coords &c, SPHState *state) {
    ComputePrecision outputValue = 0;
    ComputePrecision max_dist = 0;
    ComputePrecision r3 = cube(state->smoothRadius);
-    for (uint32_t i = 0; i < state->currentNgb; i++) {
+template<typename ValType, int Ndims>
-      ComputePrecision d = state->distances[i];
+void SPHSmooth<ValType,Ndims>::addGridSite(const typename SPHTree::coords& c)
-      SPHCell &cell = *(state->ngb[i]);
+{
-      double kernel_value = getKernel(d / state->smoothRadius) / r3;
+  ComputePrecision outputValue = 0;
-#pragma omp atomic update
+  ComputePrecision max_dist = 0;
-      cell.val.weight += kernel_value;
+  
  ComputePrecision r3 = cube(internal.smoothRadius);
  for (uint32_t i = 0; i < internal.currentNgb; i++)
    {
      ComputePrecision d = internal.distances[i];
      SPHCell& cell = *(internal.ngb[i]);
      cell.val.weight += getKernel(d/internal.smoothRadius) / r3;
    }
-  }
+}
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  void
+ComputePrecision 
-  SPHSmooth<ValType, Ndims>::addGridSite(const typename SPHTree::coords &c) {
+SPHSmooth<ValType,Ndims>::getKernel(ComputePrecision x) const
-    addGridSite(c, &internal);
+{
-  }
+  // WARNING !!! This is an unnormalized version of the kernel.
-
+  if (x < 1)
-  template <typename ValType, int Ndims>
+    return 1 - 1.5 * x * x + 0.75 * x * x * x;
-  ComputePrecision
+  else if (x < 2)
-  SPHSmooth<ValType, Ndims>::getKernel(ComputePrecision x) const {
+    {
    // WARNING !!! This is an unnormalized version of the kernel.
    if (x < 1)
      return 1 - 1.5 * x * x + 0.75 * x * x * x;
    else if (x < 2) {
      ComputePrecision d = 2 - x;
      return 0.25 * d * d * d;
-    } else
+    }
-      return 0;
+  else
-  }
+    return 0;
 }
-  template <typename ValType, int Ndims>
+template<typename ValType, int Ndims>
-  bool SPHSmooth<ValType, Ndims>::hasNeighbours() const {
+bool SPHSmooth<ValType,Ndims>::hasNeighbours() const
-    return (internal.currentNgb != 0);
+{
-  }
+  return (internal.currentNgb != 0);
 }
-  template <class ValType1, class ValType2, int Ndims>
+template<class ValType1, class ValType2, int Ndims>
-  bool operator<(
+bool operator<(const SPHSmooth<ValType1, Ndims>& s1, const SPHSmooth<ValType2, Ndims>& s2)
-      const SPHSmooth<ValType1, Ndims> &s1,
+{
-      const SPHSmooth<ValType2, Ndims> &s2) {
+  return (s1.getSmoothingLen() < s2.getSmoothingLen());
-    return (s1.getSmoothingLen() < s2.getSmoothingLen());
+}
-  }
+
-}; // namespace CosmoTool
+};
--- a/src/yorick.hpp
+++ b/src/yorick.hpp
@ -7,16 +7,16 @@ This software is a computer program whose purpose is to provide a toolbox for co
 data analysis (e.g. filters, generalized Fourier transforms, power spectra, ...)
 This software is governed by the CeCILL license under French law and
-abiding by the rules of distribution of free software.  You can  use,
+abiding by the rules of distribution of free software.  You can  use, 
 modify and/ or redistribute the software under the terms of the CeCILL
 license as circulated by CEA, CNRS and INRIA at the following URL
-"http://www.cecill.info".
+"http://www.cecill.info". 
 As a counterpart to the access to the source code and  rights to copy,
 modify and redistribute granted by the license, users are provided only
 with a limited warranty  and the software's author,  the holder of the
 economic rights,  and the successive licensors  have only  limited
-liability.
+liability. 
 In this respect, the user's attention is drawn to the risks associated
 with loading,  using,  modifying and/or developing or reproducing the
@ -25,9 +25,9 @@ that may mean  that it is complicated to manipulate,  and  that  also
 therefore means  that it is reserved for developers  and  experienced
 professionals having in-depth computer knowledge. Users are therefore
 encouraged to load and test the software's suitability as regards their
-requirements in conditions enabling the security of their systems and/or
+requirements in conditions enabling the security of their systems and/or 
-data to be ensured and,  more generally, to use and operate it in the
+data to be ensured and,  more generally, to use and operate it in the 
-same conditions as regards security.
+same conditions as regards security. 
 The fact that you are presently reading this means that you have had
 knowledge of the CeCILL license and that you accept its terms.
@ -42,7 +42,7 @@ knowledge of the CeCILL license and that you accept its terms.
 #include <string>
-namespace CosmoTool
+namespace CosmoTool 
 {
  class ProgressiveDoubleOutputImpl
@ -75,7 +75,7 @@ namespace CosmoTool
    bool initialized;
    int *ref;
    ProgressiveDoubleOutputImpl *impl;
-
+    
    friend ProgressiveDoubleOutput saveDoubleArrayProgressive(const char *fname, uint32_t *dimList, uint32_t rank);
    void decRef();
  public:
@ -86,7 +86,7 @@ namespace CosmoTool
    virtual void addDouble(double a);
-    const ProgressiveDoubleOutput& operator=(const ProgressiveDoubleOutput& b);
+    const ProgressiveDoubleOutput& operator=(const ProgressiveDoubleOutput& b); 
  };
  template<class T>
@ -95,12 +95,12 @@ namespace CosmoTool
  private:
    int *ref;
    ProgressiveInputImpl<T> *impl;
-
+     
    void decRef()
    {
      if (ref == 0)
 	return;
-
+      
      (*ref)--;
      if (*ref == 0)
 	{
@ -112,17 +112,17 @@ namespace CosmoTool
    }
  public:
-    static ProgressiveInput<T>
+    static ProgressiveInput<T> 
-    loadArrayProgressive(const std::string& fname, uint32_t *&dimList,
+    loadArrayProgressive(const std::string& fname, uint32_t *&dimList, 
 			 uint32_t& rank);
    ProgressiveInput() {
-      impl = 0;
+      impl = 0; 
-      ref = 0;
+      ref = 0; 
    }
    ProgressiveInput(ProgressiveInputImpl<T> *i) {
-      impl = i;
+      impl = i; 
-      ref = new int;
+      ref = new int; 
      *ref = 1;
    }
    ProgressiveInput(const ProgressiveInput<T>& o) {
@ -161,12 +161,12 @@ namespace CosmoTool
  private:
    int *ref;
    ProgressiveOutputImpl<T> *impl;
-
+     
    void decRef()
    {
      if (ref == 0)
 	return;
-
+      
      (*ref)--;
      if (*ref == 0)
 	{
@ -178,17 +178,17 @@ namespace CosmoTool
    }
  public:
-    static ProgressiveOutput<T>
+    static ProgressiveOutput<T> 
-    saveArrayProgressive(const std::string& fname, uint32_t *dimList,
+    saveArrayProgressive(const std::string& fname, uint32_t *dimList, 
 			 uint32_t rank);
    ProgressiveOutput() {
-      impl = 0;
+      impl = 0; 
-      ref = 0;
+      ref = 0; 
    }
    ProgressiveOutput(ProgressiveOutputImpl<T> *i) {
-      impl = i;
+      impl = i; 
-      ref = new int;
+      ref = new int; 
      *ref = 1;
    }
    ProgressiveOutput(const ProgressiveOutput<T>& o) {
@ -222,9 +222,10 @@ namespace CosmoTool
  template<typename T>
  void loadArray(const std::string& fname,
-		 T*& array, uint32_t *& dimList, uint32_t& rank);
+		 T*& array, uint32_t *& dimList, uint32_t& rank)
    throw (NoSuchFileException);
-  ProgressiveDoubleOutput saveDoubleArrayProgressive(const char *fname, uint32_t *dimList, uint32_t rank);
+  ProgressiveDoubleOutput saveDoubleArrayProgressive(const char *fname, uint32_t *dimList, uint32_t rank);  
 };
 #endif
--- a/src/yorick_nc3.cpp
+++ b/src/yorick_nc3.cpp
@ -260,6 +260,7 @@ namespace CosmoTool {
  template<typename T>
  void loadArray(const std::string& fname,
 		 T*&array, uint32_t *&dimList, uint32_t& rank)
 	throw (NoSuchFileException)
  {
    NcFile f(fname.c_str(), NcFile::ReadOnly);
--- a/src/yorick_nc4.cpp
+++ b/src/yorick_nc4.cpp
@ -7,16 +7,16 @@ This software is a computer program whose purpose is to provide a toolbox for co
 data analysis (e.g. filters, generalized Fourier transforms, power spectra, ...)
 This software is governed by the CeCILL license under French law and
-abiding by the rules of distribution of free software.  You can  use,
+abiding by the rules of distribution of free software.  You can  use, 
 modify and/ or redistribute the software under the terms of the CeCILL
 license as circulated by CEA, CNRS and INRIA at the following URL
-"http://www.cecill.info".
+"http://www.cecill.info". 
 As a counterpart to the access to the source code and  rights to copy,
 modify and redistribute granted by the license, users are provided only
 with a limited warranty  and the software's author,  the holder of the
 economic rights,  and the successive licensors  have only  limited
-liability.
+liability. 
 In this respect, the user's attention is drawn to the risks associated
 with loading,  using,  modifying and/or developing or reproducing the
@ -25,9 +25,9 @@ that may mean  that it is complicated to manipulate,  and  that  also
 therefore means  that it is reserved for developers  and  experienced
 professionals having in-depth computer knowledge. Users are therefore
 encouraged to load and test the software's suitability as regards their
-requirements in conditions enabling the security of their systems and/or
+requirements in conditions enabling the security of their systems and/or 
-data to be ensured and,  more generally, to use and operate it in the
+data to be ensured and,  more generally, to use and operate it in the 
-same conditions as regards security.
+same conditions as regards security. 
 The fact that you are presently reading this means that you have had
 knowledge of the CeCILL license and that you accept its terms.
@ -53,7 +53,7 @@ public:
  vector<size_t> counts;
  vector<NcDim> dimList;
  uint32_t rank;
-
+  
  NetCDF_handle(NcFile *f, NcVar v, vector<NcDim>& dimList, uint32_t rank);
  virtual ~NetCDF_handle();
 };
@ -86,14 +86,14 @@ public:
  InputGenCDF(NcFile *f, NcVar v, vector<NcDim>& dimList, uint32_t rank)
    : NetCDF_handle(f,v,dimList,rank)
  {}
-  virtual ~InputGenCDF() {}
+  virtual ~InputGenCDF() {} 
  virtual T read()
  {
    T a;
    curVar.getVar(curPos, counts, &a);
-
+    
    curPos[rank-1]++;
    for (long i = rank-1; i >= 1; i--)
      {
@ -102,7 +102,7 @@ public:
 	    curPos[i-1]++;
 	  curPos[i] = 0;
 	}
-      }
+      }    
    return a;
  }
@ -120,12 +120,12 @@ public:
  OutputGenCDF(NcFile *f, NcVar v, vector<NcDim>& dimList, uint32_t rank)
    : NetCDF_handle(f,v,dimList,rank)
  {}
-  virtual ~OutputGenCDF() {}
+  virtual ~OutputGenCDF() {} 
  virtual void put(T a)
  {
    curVar.putVar(curPos, counts, &a);
-
+    
    curPos[rank-1]++;
    for (long i = rank-1; i >= 1; i--)
      {
@ -134,7 +134,7 @@ public:
            curPos[i-1]++;
            curPos[i] = 0;
 	  }
-      }
+      }    
  }
 };
@ -159,17 +159,17 @@ namespace CosmoTool {
 					     uint32_t rank)
  {
    NcFile *f = new NcFile(fname, NcFile::replace);
-
+    
    vector<NcDim> dimArray;
    for (uint32_t i = 0; i < rank; i++)
      {
 	char dimName[255];
-
+	
 	sprintf(dimName, "dim%d", i);
 	dimArray.push_back(f->addDim(dimName, dimList[rank-1-i]));
      }
-
+    
-    NcVar v = f->addVar("array", get_NetCDF_type<T>(), dimArray);
+    NcVar v = f->addVar("array", get_NetCDF_type<T>(), dimArray);  
    vector<NcDim> ldimList;
@ -177,7 +177,7 @@ namespace CosmoTool {
      ldimList.push_back(dimArray[i]);
    OutputGenCDF<T> *impl = new OutputGenCDF<T>(f, v, ldimList, rank);
-    return ProgressiveOutput<T>(impl);
+    return ProgressiveOutput<T>(impl);   
  }
  template<typename T>
@ -206,31 +206,32 @@ namespace CosmoTool {
 		 const T *array, uint32_t *dimList, uint32_t rank)
  {
    NcFile f(fname.c_str(), NcFile::replace);
-
+    
    vector<NcDim> dimArray;
    for (uint32_t i = 0; i < rank; i++)
      {
 	char dimName[255];
-
+	
 	sprintf(dimName, "dim%d", i);
 	dimArray.push_back(f.addDim(dimName, dimList[i]));
      }
    NcVar v = f.addVar("array", get_NetCDF_type<T>(), dimArray);
-
+    
    v.putVar(array);
  }
  template<typename T>
  void loadArray(const std::string& fname,
 		 T*&array, uint32_t *&dimList, uint32_t& rank)
 	throw (NoSuchFileException)
  {
    NcFile f(fname.c_str(), NcFile::read);
-
+    
    //if (!f.is_valid())
    //  throw NoSuchFileException(fname);
-
+    
-    NcVar v = f.getVar("array");
+    NcVar v = f.getVar("array"); 
    vector<NcDim> dims = v.getDims();
    rank = v.getDimCount();
    uint32_t fullSize = 1;
@ -267,5 +268,5 @@ namespace CosmoTool {
 				 const float *array, uint32_t *dimList, uint32_t rank);
  template void saveArray<double>(const std::string& fname,
 				  const double *array, uint32_t *dimList, uint32_t rank);
-
+  
 }