From d257e92e54718ae5dfc8b4b0824bb69ac8a31fe1 Mon Sep 17 00:00:00 2001
From: Martin Reinecke <martin@mpa-garching.mpg.de>
Date: Mon, 17 Dec 2018 16:53:28 +0100
Subject: [PATCH] compactification

---
 libsharp/sharp_core.c | 48 +++++++++++++++----------------------------
 1 file changed, 16 insertions(+), 32 deletions(-)

diff --git a/libsharp/sharp_core.c b/libsharp/sharp_core.c
index e611805..2dcee90 100644
--- a/libsharp/sharp_core.c
+++ b/libsharp/sharp_core.c
@@ -559,27 +559,19 @@ NOINLINE static void map2alm_spin_kernel(sxdata_v * restrict d,
       d->l1p[i] = (d->cth[i]-fx11)*fx10*d->l2p[i] - fx12*d->l1p[i];
       d->l1m[i] = (d->cth[i]+fx11)*fx10*d->l2m[i] - fx12*d->l1m[i];
       Tv lw = d->l2p[i] + d->l2m[i];
-      agr1 += d->p1pr[i]*lw;
-      agi1 += d->p1pi[i]*lw;
-      acr1 += d->p1mr[i]*lw;
-      aci1 += d->p1mi[i]*lw;
       Tv lx = d->l2m[i] - d->l2p[i];
-      agr1 -= d->p2mi[i]*lx;
-      agi1 += d->p2mr[i]*lx;
-      acr1 += d->p2pi[i]*lx;
-      aci1 -= d->p2pr[i]*lx;
+      agr1 += d->p1pr[i]*lw - d->p2mi[i]*lx;;
+      agi1 += d->p1pi[i]*lw + d->p2mr[i]*lx;
+      acr1 += d->p1mr[i]*lw + d->p2pi[i]*lx;
+      aci1 += d->p1mi[i]*lw - d->p2pr[i]*lx;
       d->l2p[i] = (d->cth[i]-fx21)*fx20*d->l1p[i] - fx22*d->l2p[i];
       d->l2m[i] = (d->cth[i]+fx21)*fx20*d->l1m[i] - fx22*d->l2m[i];
       lw = d->l1p[i] + d->l1m[i];
-      agr2 += d->p2pr[i]*lw;
-      agi2 += d->p2pi[i]*lw;
-      acr2 += d->p2mr[i]*lw;
-      aci2 += d->p2mi[i]*lw;
       lx = d->l1m[i] - d->l1p[i];
-      agr2 -= d->p1mi[i]*lx;
-      agi2 += d->p1mr[i]*lx;
-      acr2 += d->p1pi[i]*lx;
-      aci2 -= d->p1pr[i]*lx;
+      agr2 += d->p2pr[i]*lw - d->p1mi[i]*lx;
+      agi2 += d->p2pi[i]*lw + d->p1mr[i]*lx;
+      acr2 += d->p2mr[i]*lw + d->p1pi[i]*lx;
+      aci2 += d->p2mi[i]*lw - d->p1pr[i]*lx;
       }
     vhsum_cmplx_special (agr1,agi1,acr1,aci1,&alm[2*l]);
     vhsum_cmplx_special (agr2,agi2,acr2,aci2,&alm[2*l+2]);
@@ -622,27 +614,19 @@ NOINLINE static void calc_map2alm_spin (sharp_job * restrict job,
       d->l1p[i] = (d->cth[i]-fx11)*fx10*d->l2p[i] - fx12*d->l1p[i];
       d->l1m[i] = (d->cth[i]+fx11)*fx10*d->l2m[i] - fx12*d->l1m[i];
       Tv lw = d->l2p[i]*d->cfp[i] + d->l2m[i]*d->cfm[i];
-      agr1 += d->p1pr[i]*lw;
-      agi1 += d->p1pi[i]*lw;
-      acr1 += d->p1mr[i]*lw;
-      aci1 += d->p1mi[i]*lw;
       Tv lx = d->l2m[i]*d->cfm[i] - d->l2p[i]*d->cfp[i];
-      agr1 -= d->p2mi[i]*lx;
-      agi1 += d->p2mr[i]*lx;
-      acr1 += d->p2pi[i]*lx;
-      aci1 -= d->p2pr[i]*lx;
+      agr1 += d->p1pr[i]*lw - d->p2mi[i]*lx;
+      agi1 += d->p1pi[i]*lw + d->p2mr[i]*lx;
+      acr1 += d->p1mr[i]*lw + d->p2pi[i]*lx;
+      aci1 += d->p1mi[i]*lw - d->p2pr[i]*lx;
       d->l2p[i] = (d->cth[i]-fx21)*fx20*d->l1p[i] - fx22*d->l2p[i];
       d->l2m[i] = (d->cth[i]+fx21)*fx20*d->l1m[i] - fx22*d->l2m[i];
       lw = d->l1p[i]*d->cfp[i] + d->l1m[i]*d->cfm[i];
-      agr2 += d->p2pr[i]*lw;
-      agi2 += d->p2pi[i]*lw;
-      acr2 += d->p2mr[i]*lw;
-      aci2 += d->p2mi[i]*lw;
       lx = d->l1m[i]*d->cfm[i] - d->l1p[i]*d->cfp[i];
-      agr2 -= d->p1mi[i]*lx;
-      agi2 += d->p1mr[i]*lx;
-      acr2 += d->p1pi[i]*lx;
-      aci2 -= d->p1pr[i]*lx;
+      agr2 += d->p2pr[i]*lw - d->p1mi[i]*lx;
+      agi2 += d->p2pi[i]*lw + d->p1mr[i]*lx;
+      acr2 += d->p2mr[i]*lw + d->p1pi[i]*lx;
+      aci2 += d->p2mi[i]*lw - d->p1pr[i]*lx;
       if (rescale(&d->l1p[i], &d->l2p[i], &d->scp[i], vload(sharp_ftol)))
         {
         getCorfac(d->scp[i], &d->cfp[i], gen->cf);