Fix OEP full - spin polarized case

Description

The OEP full was not taking the right occupations for the spin-polarized case.

News snippet

The OEP full was not taking the right occupations for the spin-polarized case.

Checklist

• I have checked that my code follows the Octopus coding standards
• I have added tests for all the new features added in this request.

share/opencl/projector.cl

@@ -42,7 +42,7 @@
 #endif
 
 
-__device__ inline double dwarpReduce(double val)
+__device__ inline rtype X(warpReduce)(rtype val)
 {
 #pragma unroll
   for (int offset = warpSize/2; offset > 0; offset /= 2){
@@ -51,18 +51,6 @@ __device__ inline double dwarpReduce(double val)
   return val;
 }
 
-__device__ inline double2 zwarpReduce(double2 val)
-{
-#pragma unroll
-  for (int offset = warpSize/2; offset > 0; offset /= 2){
-    val.x += warpShflDown(val.x, offset);
-    val.y += warpShflDown(val.y, offset);
-  }
-  return val;
-}
-
-#endif
-
 
 __kernel void X(projector_bra)(const int nmat,
           __global int const * restrict offsets,
@@ -79,8 +67,8 @@ __kernel void X(projector_bra)(const int nmat,
   const int my_warp_size = 1;
 #endif
 
-  const int ist = get_global_id(0)/my_warp_size;
-  const int ipj = get_global_id(1);
+  const int ist  = get_global_id(0) / my_warp_size;
+  const int ipj  = get_global_id(1);
   const int imat = get_global_id(2);
 
   const int npoints       = offsets[OFFSET_SIZE*imat + 0];
@@ -91,27 +79,27 @@ __kernel void X(projector_bra)(const int nmat,
 
   if(ipj >= nprojs) return;
 
-  const int nppj = npoints*ipj;
+  const int nppj = npoints * ipj;
 
 #ifdef CUDA
-  const int slice = npoints%my_warp_size==0 ? npoints/my_warp_size : npoints/my_warp_size+1;
-  const int start = slice * ( get_local_id(0)%my_warp_size ) ;
-  const int end   = min( start + slice , npoints );
+  const int slice = npoints%my_warp_size == 0 ? npoints/my_warp_size : npoints/my_warp_size+1;
+  const int start = slice * (get_local_id(0)%my_warp_size) ;
+  const int end   = min(start + slice, npoints);
   const int step  = 1;
 #else
   const int start = 0;
-  const int end = npoints;
-  const int step = 1;
+  const int end   = npoints;
+  const int step  = 1;
 #endif
 
   rtype aa = 0.0;
-  for(int ip = start; ip < end; ip+=step){
+  for(int ip = start; ip < end; ip += step){
     aa += MUL(CONJ(matrix[matrix_offset + ip + nppj]), psi[((map[map_offset + ip] - 1)<<ldpsi) + ist]);
   }
 
 #ifdef CUDA
   aa = X(warpReduce)(aa);
-  if(get_local_id(0)%my_warp_size==0)
+  if(get_local_id(0)%my_warp_size == 0)
 #endif
     projection[ist + ((scal_offset + ipj)<<ldprojection)] = MUL(scal[scal_offset + ipj], aa);
 
@@ -674,9 +662,9 @@ __kernel void zprojector_mix(const int nmat,
   const int ipj  = get_global_id(1);
   const int imat = get_global_id(2);
 
-  const int nprojs        = offsets[OFFSET_SIZE*imat + 1];
-  const int scal_offset   = offsets[OFFSET_SIZE*imat + 4];
-  const int mix_offset_0  = offsets[OFFSET_SIZE*imat + 5];
+  const int nprojs       = offsets[OFFSET_SIZE*imat + 1];
+  const int scal_offset  = offsets[OFFSET_SIZE*imat + 4];
+  const int mix_offset_0 = offsets[OFFSET_SIZE*imat + 5];
   const int mix_offset_1 = mix_offset_0 + nprojs*nprojs;
   const int mix_offset_2 = mix_offset_1 + nprojs*nprojs;
   const int mix_offset_3 = mix_offset_2 + nprojs*nprojs;