[cig-commits] [commit] rajesh-petsc-schur: replaced all occurences of the caps_per_proc iteration variable by CPPR in Topo_gravity.c (6dec4de)

cig_noreply at geodynamics.org cig_noreply at geodynamics.org
Wed Nov 5 19:06:21 PST 2014


Repository : https://github.com/geodynamics/citcoms

On branch  : rajesh-petsc-schur
Link       : https://github.com/geodynamics/citcoms/compare/464e1b32299b15819f93efd98d969cddb84dfe51...f97ae655a50bdbd6dac1923a3471ee4dae178fbd

>---------------------------------------------------------------

commit 6dec4de733bb6d29753f6a8ccae9a62c97768175
Author: Rajesh Kommu <rajesh.kommu at gmail.com>
Date:   Tue Sep 16 11:57:33 2014 -0700

    replaced all occurences of the caps_per_proc iteration variable by CPPR in Topo_gravity.c


>---------------------------------------------------------------

6dec4de733bb6d29753f6a8ccae9a62c97768175
 lib/Topo_gravity.c | 270 ++++++++++++++++++++++++++---------------------------
 1 file changed, 135 insertions(+), 135 deletions(-)

diff --git a/lib/Topo_gravity.c b/lib/Topo_gravity.c
index 5481db6..02daf61 100644
--- a/lib/Topo_gravity.c
+++ b/lib/Topo_gravity.c
@@ -93,15 +93,15 @@ void get_STD_topo(E,tpg,tpgb,divg,vort,ii)
 
    for(m=1;m<=E->sphere.caps_per_proc;m++)
      for(snode=1;snode<=E->lmesh.nsf;snode++)   {
-        node = E->surf_node[m][snode];
-        tpg[m][snode]  = -2*SZZ[m][node]               + SZZ[m][node-1];
-        tpgb[m][snode] =  2*SZZ[m][node-E->lmesh.noz+1]- SZZ[m][node-E->lmesh.noz+2];
+        node = E->surf_node[CPPR][snode];
+        tpg[CPPR][snode]  = -2*SZZ[CPPR][node]               + SZZ[CPPR][node-1];
+        tpgb[CPPR][snode] =  2*SZZ[CPPR][node-E->lmesh.noz+1]- SZZ[CPPR][node-E->lmesh.noz+2];
 
-        tpg[m][snode]  =  tpg[m][snode] *topo_scaling1;
-        tpgb[m][snode]  = tpgb[m][snode]*topo_scaling2;
+        tpg[CPPR][snode]  =  tpg[CPPR][snode] *topo_scaling1;
+        tpgb[CPPR][snode]  = tpgb[CPPR][snode]*topo_scaling2;
 
-        divg[m][snode] = 2*divv[m][node]-divv[m][node-1];
-        vort[m][snode] = 2*vorv[m][node]-vorv[m][node-1];
+        divg[CPPR][snode] = 2*divv[CPPR][node]-divv[CPPR][node-1];
+        vort[CPPR][snode] = 2*vorv[CPPR][node]-vorv[CPPR][node-1];
      }
 
    free_STD_mem(E, SXX, SYY, SZZ, SXY, SXZ, SZY, divv, vorv);
@@ -116,9 +116,9 @@ void get_STD_freesurf(struct All_variables *E,float **freesurf)
         if (E->parallel.me_loc[3]==E->parallel.nprocz-1)
                 for(m=1;m<=E->sphere.caps_per_proc;m++)
                         for(snode=1;snode<=E->lmesh.nsf;snode++) {
-                                node = E->surf_node[m][snode];
+                                node = E->surf_node[CPPR][snode];
                                 /*freesurf[m][snode] += 0.5*(E->sphere.cap[m].V[3][node]+E->sphere.cap[m].Vprev[3][node])*E->advection.timestep;*/
-                                freesurf[m][snode] += E->sphere.cap[m].V[3][node]*E->advection.timestep;
+                                freesurf[CPPR][snode] += E->sphere.cap[CPPR].V[3][node]*E->advection.timestep;
                         }
         return;
 }
@@ -132,26 +132,26 @@ void allocate_STD_mem(struct All_variables *E,
   int m, i;
 
   for(m=1;m<=E->sphere.caps_per_proc;m++) {
-    SXX[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
-    SYY[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
-    SXY[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
-    SXZ[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
-    SZY[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
-    SZZ[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
-    divv[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
-    vorv[m] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    SXX[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    SYY[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    SXY[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    SXZ[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    SZY[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    SZZ[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    divv[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
+    vorv[CPPR] = (float *)malloc((E->lmesh.nno+1)*sizeof(float));
   }
 
   for(m=1;m<=E->sphere.caps_per_proc;m++) {
     for(i=1;i<=E->lmesh.nno;i++) {
-      SZZ[m][i] = 0.0;
-      SXX[m][i] = 0.0;
-      SYY[m][i] = 0.0;
-      SXY[m][i] = 0.0;
-      SXZ[m][i] = 0.0;
-      SZY[m][i] = 0.0;
-      divv[m][i] = 0.0;
-      vorv[m][i] = 0.0;
+      SZZ[CPPR][i] = 0.0;
+      SXX[CPPR][i] = 0.0;
+      SYY[CPPR][i] = 0.0;
+      SXY[CPPR][i] = 0.0;
+      SXZ[CPPR][i] = 0.0;
+      SZY[CPPR][i] = 0.0;
+      divv[CPPR][i] = 0.0;
+      vorv[CPPR][i] = 0.0;
     }
   }
   return;
@@ -165,14 +165,14 @@ void free_STD_mem(struct All_variables *E,
 {
   int m;
   for(m=1;m<=E->sphere.caps_per_proc;m++)        {
-    free((void *)SXX[m]);
-    free((void *)SYY[m]);
-    free((void *)SXY[m]);
-    free((void *)SXZ[m]);
-    free((void *)SZY[m]);
-    free((void *)SZZ[m]);
-    free((void *)divv[m]);
-    free((void *)vorv[m]);
+    free((void *)SXX[CPPR]);
+    free((void *)SYY[CPPR]);
+    free((void *)SXY[CPPR]);
+    free((void *)SXZ[CPPR]);
+    free((void *)SZY[CPPR]);
+    free((void *)SZZ[CPPR]);
+    free((void *)divv[CPPR]);
+    free((void *)vorv[CPPR]);
     }
 }
 
@@ -255,13 +255,13 @@ void compute_nodal_stress(struct All_variables *E,
       div = 0.0;
       vor = 0.0;
 
-      get_rtf_at_vpts(E, m, lev, e, rtf);// gets r,theta,phi coordinates at the integration points
-      velo_from_element(E,VV,m,e,sphere_key); /* assign node-global
+      get_rtf_at_vpts(E, CPPR, lev, e, rtf);// gets r,theta,phi coordinates at the integration points
+      velo_from_element(E,VV,CPPR,e,sphere_key); /* assign node-global
 						 velocities to nodes
 						 local to the
 						 element */
-      dOmega = &(E->gDA[m][e]);	/* Jacobian at integration points */
-      GNx = &(E->gNX[m][e]);	/* derivatives of shape functions at
+      dOmega = &(E->gDA[CPPR][e]);	/* Jacobian at integration points */
+      GNx = &(E->gNX[CPPR][e]);	/* derivatives of shape functions at
 				   integration points */
 
       /* Vxyz is the strain rate vector, whose relationship with
@@ -276,7 +276,7 @@ void compute_nodal_stress(struct All_variables *E,
        */
       for(j=1;j <= vpts;j++)  {	/* loop through velocity Gauss points  */
 	/* E->EVi[j] = E->EVI[E->mesh.levmax][j]; */
-        pre[j] =  E->EVi[m][(e-1)*vpts+j]*dOmega->vpt[j];
+        pre[j] =  E->EVi[CPPR][(e-1)*vpts+j]*dOmega->vpt[j];
         dilation[j] = 0.0;
         Vxyz[1][j] = 0.0;
         Vxyz[2][j] = 0.0;
@@ -302,7 +302,7 @@ void compute_nodal_stress(struct All_variables *E,
 	*/
 
 	if ((e-1)%E->lmesh.elz==0) {
-	  construct_c3x3matrix_el(E,e,&E->element_Cc,&E->element_Ccx,lev,m,0);
+	  construct_c3x3matrix_el(E,e,&E->element_Cc,&E->element_Ccx,lev,CPPR,0);
 	}
 	/* get B at velocity Gauss points */
 	get_ba(&(E->N), GNx, &E->element_Cc, &E->element_Ccx,rtf, dims, ba);
@@ -375,11 +375,11 @@ void compute_nodal_stress(struct All_variables *E,
 
 	*/
 	get_constitutive(D,rtf[1][i],rtf[2][i],TRUE,
-			 E->EVIn1[E->mesh.levmax][m][l1], 
-			 E->EVIn2[E->mesh.levmax][m][l1], 
-			 E->EVIn3[E->mesh.levmax][m][l1],
-			 E->EVI2[E->mesh.levmax][m][l1],
-			 E->avmode[E->mesh.levmax][m][l1],
+			 E->EVIn1[E->mesh.levmax][CPPR][l1], 
+			 E->EVIn2[E->mesh.levmax][CPPR][l1], 
+			 E->EVIn3[E->mesh.levmax][CPPR][l1],
+			 E->EVI2[E->mesh.levmax][CPPR][l1],
+			 E->avmode[E->mesh.levmax][CPPR][l1],
 			 E);
 	
 	/* deviatoric stress, pressure will be added later */
@@ -422,30 +422,30 @@ void compute_nodal_stress(struct All_variables *E,
     }
 #endif
       /* normalize by volume */
-      Sxx /= E->eco[m][e].area;
-      Syy /= E->eco[m][e].area;
-      Szz /= E->eco[m][e].area;
-      Sxy /= E->eco[m][e].area;
-      Sxz /= E->eco[m][e].area;
-      Szy /= E->eco[m][e].area;
-      div /= E->eco[m][e].area;
-      vor /= E->eco[m][e].area;
+      Sxx /= E->eco[CPPR][e].area;
+      Syy /= E->eco[CPPR][e].area;
+      Szz /= E->eco[CPPR][e].area;
+      Sxy /= E->eco[CPPR][e].area;
+      Sxz /= E->eco[CPPR][e].area;
+      Szy /= E->eco[CPPR][e].area;
+      div /= E->eco[CPPR][e].area;
+      vor /= E->eco[CPPR][e].area;
 
       /* add the pressure term */
-      Szz -= E->P[m][e-1]; /* E->P is now 0-based. Change e-1 to e */
-      Sxx -= E->P[m][e-1]; /* when the loop variable e goes from */
-      Syy -= E->P[m][e-1]; /* 0 to nel-1, instead of the current 1 to nel */
+      Szz -= E->P[CPPR][e-1]; /* E->P is now 0-based. Change e-1 to e */
+      Sxx -= E->P[CPPR][e-1]; /* when the loop variable e goes from */
+      Syy -= E->P[CPPR][e-1]; /* 0 to nel-1, instead of the current 1 to nel */
 
       for(i=1;i<=ends;i++) {
-        node = E->ien[m][e].node[i]; /* assign to global nodes */
-        SZZ[m][node] += tww[i] * Szz;
-        SXX[m][node] += tww[i] * Sxx;
-        SYY[m][node] += tww[i] * Syy;
-        SXY[m][node] += tww[i] * Sxy;
-        SXZ[m][node] += tww[i] * Sxz;
-        SZY[m][node] += tww[i] * Szy;
-        divv[m][node]+= tww[i] * div;
-        vorv[m][node]+= tww[i] * vor;
+        node = E->ien[CPPR][e].node[i]; /* assign to global nodes */
+        SZZ[CPPR][node] += tww[i] * Szz;
+        SXX[CPPR][node] += tww[i] * Sxx;
+        SYY[CPPR][node] += tww[i] * Syy;
+        SXY[CPPR][node] += tww[i] * Sxy;
+        SXZ[CPPR][node] += tww[i] * Sxz;
+        SZY[CPPR][node] += tww[i] * Szy;
+        divv[CPPR][node]+= tww[i] * div;
+        vorv[CPPR][node]+= tww[i] * vor;
       }
 
     }    /* end for el */
@@ -464,28 +464,28 @@ void compute_nodal_stress(struct All_variables *E,
 
   for(m=1;m<=E->sphere.caps_per_proc;m++)
     for(node=1;node<=E->lmesh.nno;node++)   {
-      mass_fac = E->Mass[m][node]*stress_scaling;
-      SZZ[m][node] *= mass_fac;
-      SXX[m][node] *= mass_fac;
-      SYY[m][node] *= mass_fac;
-      SXY[m][node] *= mass_fac;
-      SXZ[m][node] *= mass_fac;
-      SZY[m][node] *= mass_fac;
-      
-      mass_fac = E->Mass[m][node]*velo_scaling;
-      vorv[m][node] *= mass_fac;
-      divv[m][node] *= mass_fac;
+      mass_fac = E->Mass[CPPR][node]*stress_scaling;
+      SZZ[CPPR][node] *= mass_fac;
+      SXX[CPPR][node] *= mass_fac;
+      SYY[CPPR][node] *= mass_fac;
+      SXY[CPPR][node] *= mass_fac;
+      SXZ[CPPR][node] *= mass_fac;
+      SZY[CPPR][node] *= mass_fac;
+      
+      mass_fac = E->Mass[CPPR][node]*velo_scaling;
+      vorv[CPPR][node] *= mass_fac;
+      divv[CPPR][node] *= mass_fac;
     }
 
   /* assign stress to all the nodes */
   for(m=1;m<=E->sphere.caps_per_proc;m++)
     for (node=1;node<=E->lmesh.nno;node++) {
-      E->gstress[m][(node-1)*6+1] = SXX[m][node];
-      E->gstress[m][(node-1)*6+2] = SYY[m][node];
-      E->gstress[m][(node-1)*6+3] = SZZ[m][node];
-      E->gstress[m][(node-1)*6+4] = SXY[m][node];
-      E->gstress[m][(node-1)*6+5] = SXZ[m][node];
-      E->gstress[m][(node-1)*6+6] = SZY[m][node];
+      E->gstress[CPPR][(node-1)*6+1] = SXX[CPPR][node];
+      E->gstress[CPPR][(node-1)*6+2] = SYY[CPPR][node];
+      E->gstress[CPPR][(node-1)*6+3] = SZZ[CPPR][node];
+      E->gstress[CPPR][(node-1)*6+4] = SXY[CPPR][node];
+      E->gstress[CPPR][(node-1)*6+5] = SXZ[CPPR][node];
+      E->gstress[CPPR][(node-1)*6+6] = SZY[CPPR][node];
     }
 
   /* replace boundary stresses with boundary conditions (if specified) */
@@ -524,19 +524,19 @@ void stress_conform_bcs(struct All_variables *E)
 
             for(d=1; d<=E->mesh.nsd; d++)
 
-              if(E->node[m][n] & sbc_flag[d]) {
+              if(E->node[CPPR][n] & sbc_flag[d]) {
                 if(i==1)
-                  E->gstress[m][(n-1)*6+stress_index[d][2]] = E->sbc.SB[m][SIDE_WEST][d][ E->sbc.node[m][n] ];
+                  E->gstress[CPPR][(n-1)*6+stress_index[d][2]] = E->sbc.SB[CPPR][SIDE_WEST][d][ E->sbc.node[CPPR][n] ];
                 if(i==E->lmesh.noy)
-                  E->gstress[m][(n-1)*6+stress_index[d][2]] = E->sbc.SB[m][SIDE_EAST][d][ E->sbc.node[m][n] ];
+                  E->gstress[CPPR][(n-1)*6+stress_index[d][2]] = E->sbc.SB[CPPR][SIDE_EAST][d][ E->sbc.node[CPPR][n] ];
                 if(j==1)
-                  E->gstress[m][(n-1)*6+stress_index[d][1]] = E->sbc.SB[m][SIDE_NORTH][d][ E->sbc.node[m][n] ];
+                  E->gstress[CPPR][(n-1)*6+stress_index[d][1]] = E->sbc.SB[CPPR][SIDE_NORTH][d][ E->sbc.node[CPPR][n] ];
                 if(j==E->lmesh.nox)
-                  E->gstress[m][(n-1)*6+stress_index[d][1]] = E->sbc.SB[m][SIDE_SOUTH][d][ E->sbc.node[m][n] ];
+                  E->gstress[CPPR][(n-1)*6+stress_index[d][1]] = E->sbc.SB[CPPR][SIDE_SOUTH][d][ E->sbc.node[CPPR][n] ];
                 if(k==1)
-                  E->gstress[m][(n-1)*6+stress_index[d][3]] = E->sbc.SB[m][SIDE_BOTTOM][d][ E->sbc.node[m][n] ];
+                  E->gstress[CPPR][(n-1)*6+stress_index[d][3]] = E->sbc.SB[CPPR][SIDE_BOTTOM][d][ E->sbc.node[CPPR][n] ];
                 if(k==E->lmesh.noz)
-                  E->gstress[m][(n-1)*6+stress_index[d][3]] = E->sbc.SB[m][SIDE_TOP][d][ E->sbc.node[m][n] ];
+                  E->gstress[CPPR][(n-1)*6+stress_index[d][3]] = E->sbc.SB[CPPR][SIDE_TOP][d][ E->sbc.node[CPPR][n] ];
               }
           }
 
@@ -553,9 +553,9 @@ void stress_conform_bcs(struct All_variables *E)
 	    for(k=1; k<=E->lmesh.noz; k++) {
 	      n = k+(j-1)*E->lmesh.noz+(i-1)*noxnoz;
 	      for(d=1; d<=E->mesh.nsd; d++)
-		if(E->node[m][n] & sbc_flag[d]) {
+		if(E->node[CPPR][n] & sbc_flag[d]) {
 		  /* apply internal traction vector on horizontal surface */
-		  E->gstress[m][(n-1)*6+stress_index[d][3]] = E->sphere.cap[m].VB[d][n];
+		  E->gstress[CPPR][(n-1)*6+stress_index[d][3]] = E->sphere.cap[CPPR].VB[d][n];
 		}
 	    }
     }else{
@@ -566,13 +566,13 @@ void stress_conform_bcs(struct All_variables *E)
 	    for(k=1; k<=E->lmesh.noz; k++) {
 	      n = k+(j-1)*E->lmesh.noz+(i-1)*noxnoz;
 	      for(d=1; d<=E->mesh.nsd; d++)
-		if(E->node[m][n] & sbc_flag[d]) {
+		if(E->node[CPPR][n] & sbc_flag[d]) {
 		  if(i==1 || i==E->lmesh.noy)
-		    E->gstress[m][(n-1)*6+stress_index[d][2]] = E->sphere.cap[m].VB[d][n];
+		    E->gstress[CPPR][(n-1)*6+stress_index[d][2]] = E->sphere.cap[CPPR].VB[d][n];
 		  if(j==1 || j==E->lmesh.nox)
-		    E->gstress[m][(n-1)*6+stress_index[d][1]] = E->sphere.cap[m].VB[d][n];
+		    E->gstress[CPPR][(n-1)*6+stress_index[d][1]] = E->sphere.cap[CPPR].VB[d][n];
 		  if(k==1 || k==E->lmesh.noz)
-		    E->gstress[m][(n-1)*6+stress_index[d][3]] = E->sphere.cap[m].VB[d][n];
+		    E->gstress[CPPR][(n-1)*6+stress_index[d][3]] = E->sphere.cap[CPPR].VB[d][n];
 		}
 	    }
     }
@@ -612,7 +612,7 @@ static void geoid_from_buoyancy(struct All_variables *E,
 
     /* density of one layer */
     for(m=1;m<=E->sphere.caps_per_proc;m++)
-        TT[m] = (float *) malloc ((E->lmesh.nsf+1)*sizeof(float));
+        TT[CPPR] = (float *) malloc ((E->lmesh.nsf+1)*sizeof(float));
 
     /* cos coeff */
     geoid[0] = (float*)malloc(E->sphere.hindice*sizeof(float));
@@ -639,7 +639,7 @@ static void geoid_from_buoyancy(struct All_variables *E,
                     p = j + (i-1)*E->lmesh.nox;
                     /* convert non-dimensional buoyancy to */
                     /* dimensional density */
-                    TT[m][p] = (E->buoyancy[m][node]+E->buoyancy[m][node+1])
+                    TT[CPPR][p] = (E->buoyancy[CPPR][node]+E->buoyancy[CPPR][node+1])
                         * 0.5 * buoy2rho;
                 }
 
@@ -647,10 +647,10 @@ static void geoid_from_buoyancy(struct All_variables *E,
         sphere_expansion(E,TT,geoid[0],geoid[1]);
 
         /* thickness of the layer */
-        dlayer = (E->sx[1][3][k+1]-E->sx[1][3][k])*radius_m;
+        dlayer = (E->sx[CPPR][3][k+1]-E->sx[CPPR][3][k])*radius_m;
 
         /* mean radius of the layer */
-        radius = (E->sx[1][3][k+1]+E->sx[1][3][k])*0.5;
+        radius = (E->sx[CPPR][3][k+1]+E->sx[CPPR][3][k])*0.5;
 
         /* geoid contribution of density at this layer, ignore degree-0 term */
         for (ll=1;ll<=E->output.llmax;ll++) {
@@ -677,7 +677,7 @@ static void geoid_from_buoyancy(struct All_variables *E,
     sum_across_depth_sph1(E, harm_geoidb[0], harm_geoidb[1]);
 
     for(m=1;m<=E->sphere.caps_per_proc;m++)
-        free ((void *)TT[m]);
+        free ((void *)TT[CPPR]);
 
     free ((void *)geoid[0]);
     free ((void *)geoid[1]);
@@ -1034,24 +1034,24 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
     eltTL = (float *)malloc((1+Tsize)*sizeof(float));
 
   for(j=1;j<=E->sphere.caps_per_proc;j++)          {
-    SU[j] = (float *)malloc((1+lnsf)*sizeof(float));
-    SL[j] = (float *)malloc((1+lnsf)*sizeof(float));
-    RU[j] = (float *)malloc((1+lnsf)*sizeof(float));
-    RL[j] = (float *)malloc((1+lnsf)*sizeof(float));
+    SU[CPPR] = (float *)malloc((1+lnsf)*sizeof(float));
+    SL[CPPR] = (float *)malloc((1+lnsf)*sizeof(float));
+    RU[CPPR] = (float *)malloc((1+lnsf)*sizeof(float));
+    RL[CPPR] = (float *)malloc((1+lnsf)*sizeof(float));
     }
 
   for(j=1;j<=E->sphere.caps_per_proc;j++)          {
 
     for(i=0;i<=lnsf;i++)
-      RU[j][i] = RL[j][i] = SU[j][i] = SL[j][i] = 0.0;
+      RU[CPPR][i] = RL[CPPR][i] = SU[CPPR][i] = SL[CPPR][i] = 0.0;
 
     /* calculate the element residuals */
 
     for(els=1;els<=E->lmesh.snel;els++) {
-      el = E->surf_element[j][els];
+      el = E->surf_element[CPPR][els];
       elb = el - elz+1;
 
-      velo_from_element(E,VV,j,elb,sphere_key);
+      velo_from_element(E,VV,CPPR,elb,sphere_key);
 
       for(m=0;m<ends;m++) {
          eub [m*dims  ] = VV[1][m+1];
@@ -1059,7 +1059,7 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
          eub [m*dims+2] = VV[3][m+1];
          }
 
-      velo_from_element(E,VV,j,el,sphere_key);
+      velo_from_element(E,VV,CPPR,el,sphere_key);
 
       for(m=0;m<ends;m++) {
          eu [m*dims  ] = VV[1][m+1];
@@ -1070,22 +1070,22 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
       /* The statement order is important:
          elb must be executed before el when calling get_elt_f().
          Otherwise, construct_c3x3matrix_el() would be skipped incorrectly. */
-      get_elt_f(E,elb,eltfb,1,j);
-      get_elt_f(E,el,eltf,1,j);
+      get_elt_f(E,elb,eltfb,1,CPPR);
+      get_elt_f(E,el,eltf,1,CPPR);
 
-      get_elt_k(E,elb,eltkb,lev,j,1);
-      get_elt_k(E,el,eltk,lev,j,1);
+      get_elt_k(E,elb,eltkb,lev,CPPR,1);
+      get_elt_k(E,el,eltk,lev,CPPR,1);
 
       if (E->control.augmented_Lagr) {
-          get_aug_k(E,elb,eltkb,lev,1);
-          get_aug_k(E,el,eltk,lev,1);
+          get_aug_k(E,elb,eltkb,lev,CPPR);
+          get_aug_k(E,el,eltk,lev,CPPR);
       }
 //      get_elt_g(E,elb,eltgb,lev,j);
 //      get_elt_g(E,el,eltg,lev,j);
 
       for(m=0;m<dims*ends;m++) {
-           res[m]  = eltf[m]  - E->elt_del[lev][j][el].g[m][0]  * E->P[j][el-1];
-           resb[m] = eltfb[m] - E->elt_del[lev][j][elb].g[m][0]* E->P[j][elb-1];
+           res[m]  = eltf[m]  - E->elt_del[lev][CPPR][el].g[m][0]  * E->P[CPPR][el-1];
+           resb[m] = eltfb[m] - E->elt_del[lev][CPPR][elb].g[m][0]* E->P[CPPR][elb-1];
 //           res[m]  = eltf[m] - eltg[m][0]  * E->P[j][el];
 //           resb[m] = eltfb[m] - eltgb[m][0]* E->P[j][elb];
             }
@@ -1102,36 +1102,36 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
         if (m<=4)  {
           switch (m) {
              case 1:
-                nodes = E->sien[j][els].node[1];
+                nodes = E->sien[CPPR][els].node[1];
 		break;
              case 2:
-                nodes = E->sien[j][els].node[2];
+                nodes = E->sien[CPPR][els].node[2];
 		break;
              case 3:
-                nodes = E->sien[j][els].node[3];
+                nodes = E->sien[CPPR][els].node[3];
 		break;
              case 4:
-                nodes = E->sien[j][els].node[4];
+                nodes = E->sien[CPPR][els].node[4];
 		break;
 	     }
-	     RL[j][nodes] += resb[(m-1)*dims+2];
+	     RL[CPPR][nodes] += resb[(m-1)*dims+2];
 	  }
         else   {
            switch (m) {
              case 5:
-                nodes = E->sien[j][els].node[1];
+                nodes = E->sien[CPPR][els].node[1];
                 break;
              case 6:
-                nodes = E->sien[j][els].node[2];
+                nodes = E->sien[CPPR][els].node[2];
                 break;
              case 7:
-                nodes = E->sien[j][els].node[3];
+                nodes = E->sien[CPPR][els].node[3];
                 break;
              case 8:
-                nodes = E->sien[j][els].node[4];
+                nodes = E->sien[CPPR][els].node[4];
                 break;
              }
-             RU[j][nodes] += res[(m-1)*dims+2];
+             RU[CPPR][nodes] += res[(m-1)*dims+2];
           }
         }      /* end for m */
       }
@@ -1143,8 +1143,8 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
        el = E->surf_element[j][els];
        elb = el - elz+1;
 
-       get_global_1d_shape_fn_L(E,el,&GM,&dGammax,1,j);
-       get_global_1d_shape_fn_L(E,elb,&GMb,&dGammabx,0,j);
+       get_global_1d_shape_fn_L(E,el,&GM,&dGammax,1,CPPR);
+       get_global_1d_shape_fn_L(E,elb,&GMb,&dGammabx,0,CPPR);
 
        for(m=1;m<=onedv;m++)        {
           eltTU[m-1] = 0.0;
@@ -1160,10 +1160,10 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
           }
 
         for (m=1;m<=onedv;m++)     /* for bottom */
-            SL[j][E->sien[j][els].node[m]] += eltTL[m-1];
+            SL[CPPR][E->sien[CPPR][els].node[m]] += eltTL[m-1];
 
         for (m=1;m<=onedv;m++)
-            SU[j][E->sien[j][els].node[m]] += eltTU[m-1];
+            SU[CPPR][E->sien[CPPR][els].node[m]] += eltTU[m-1];
 
         }
 
@@ -1178,7 +1178,7 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
 
   for (j=1;j<=E->sphere.caps_per_proc;j++)
       for(i=1;i<=E->lmesh.nsf;i++)
-          HB[j][i] = RL[j][i]/SL[j][i];
+          HB[CPPR][i] = RL[CPPR][i]/SL[CPPR][i];
   }
   /* for top topo */
   if(E->parallel.me_loc[3] == E->parallel.nprocz-1) {
@@ -1189,15 +1189,15 @@ void get_CBF_topo(E,H,HB)       /* call this only for top and bottom processors*
 
   for (j=1;j<=E->sphere.caps_per_proc;j++)
       for(i=1;i<=E->lmesh.nsf;i++)
-          H[j][i] = RU[j][i]/SU[j][i];
+          H[CPPR][i] = RU[CPPR][i]/SU[CPPR][i];
   }
     free((void *)eltTU);
     free((void *)eltTL);
     for (j=1;j<=E->sphere.caps_per_proc;j++)   {
-      free((void *)SU[j]);
-      free((void *)SL[j]);
-      free((void *)RU[j]);
-      free((void *)RL[j]);
+      free((void *)SU[CPPR]);
+      free((void *)SL[CPPR]);
+      free((void *)RU[CPPR]);
+      free((void *)RL[CPPR]);
       }
     return;
 }



More information about the CIG-COMMITS mailing list