diff options
Diffstat (limited to 'ML_BSSN/src/ML_BSSN_convertFromADMBaseGamma.cc')
| -rw-r--r-- | ML_BSSN/src/ML_BSSN_convertFromADMBaseGamma.cc | 111 |
1 files changed, 38 insertions, 73 deletions
diff --git a/ML_BSSN/src/ML_BSSN_convertFromADMBaseGamma.cc b/ML_BSSN/src/ML_BSSN_convertFromADMBaseGamma.cc index b32e95b..37a355b 100644 --- a/ML_BSSN/src/ML_BSSN_convertFromADMBaseGamma.cc +++ b/ML_BSSN/src/ML_BSSN_convertFromADMBaseGamma.cc @@ -47,8 +47,6 @@ static void ML_BSSN_convertFromADMBaseGamma_Body(cGH const * restrict const cctk DECLARE_CCTK_PARAMETERS; - /* Declare finite differencing variables */ - /* Include user-supplied include files */ /* Initialise finite differencing variables */ @@ -85,9 +83,9 @@ static void ML_BSSN_convertFromADMBaseGamma_Body(cGH const * restrict const cctk CCTK_REAL_VEC const p1o12dx = kmul(INV(dx),ToReal(0.0833333333333333333333333333333)); CCTK_REAL_VEC const p1o12dy = kmul(INV(dy),ToReal(0.0833333333333333333333333333333)); CCTK_REAL_VEC const p1o12dz = kmul(INV(dz),ToReal(0.0833333333333333333333333333333)); - CCTK_REAL_VEC const p1o144dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(0.00694444444444444444444444444444))); - CCTK_REAL_VEC const p1o144dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(0.00694444444444444444444444444444))); - CCTK_REAL_VEC const p1o144dydz = kmul(INV(dy),kmul(INV(dz),ToReal(0.00694444444444444444444444444444))); + CCTK_REAL_VEC const p1o144dxdy = kmul(INV(kmul(dx,dy)),ToReal(0.00694444444444444444444444444444)); + CCTK_REAL_VEC const p1o144dxdz = kmul(INV(kmul(dx,dz)),ToReal(0.00694444444444444444444444444444)); + CCTK_REAL_VEC const p1o144dydz = kmul(INV(kmul(dy,dz)),ToReal(0.00694444444444444444444444444444)); CCTK_REAL_VEC const p1o1680dx = kmul(INV(dx),ToReal(0.000595238095238095238095238095238)); CCTK_REAL_VEC const p1o1680dy = kmul(INV(dy),ToReal(0.000595238095238095238095238095238)); CCTK_REAL_VEC const p1o1680dz = kmul(INV(dz),ToReal(0.000595238095238095238095238095238)); @@ -106,14 +104,14 @@ static void ML_BSSN_convertFromADMBaseGamma_Body(cGH const * restrict const cctk CCTK_REAL_VEC const p1o2dx = kmul(INV(dx),ToReal(0.5)); CCTK_REAL_VEC const p1o2dy = kmul(INV(dy),ToReal(0.5)); CCTK_REAL_VEC const p1o2dz = kmul(INV(dz),ToReal(0.5)); - CCTK_REAL_VEC const p1o3600dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(0.000277777777777777777777777777778))); - CCTK_REAL_VEC const p1o3600dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(0.000277777777777777777777777777778))); - CCTK_REAL_VEC const p1o3600dydz = kmul(INV(dy),kmul(INV(dz),ToReal(0.000277777777777777777777777777778))); + CCTK_REAL_VEC const p1o3600dxdy = kmul(INV(kmul(dx,dy)),ToReal(0.000277777777777777777777777777778)); + CCTK_REAL_VEC const p1o3600dxdz = kmul(INV(kmul(dx,dz)),ToReal(0.000277777777777777777777777777778)); + CCTK_REAL_VEC const p1o3600dydz = kmul(INV(kmul(dy,dz)),ToReal(0.000277777777777777777777777777778)); CCTK_REAL_VEC const p1o4dx = kmul(INV(dx),ToReal(0.25)); - CCTK_REAL_VEC const p1o4dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(0.25))); - CCTK_REAL_VEC const p1o4dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(0.25))); + CCTK_REAL_VEC const p1o4dxdy = kmul(INV(kmul(dx,dy)),ToReal(0.25)); + CCTK_REAL_VEC const p1o4dxdz = kmul(INV(kmul(dx,dz)),ToReal(0.25)); CCTK_REAL_VEC const p1o4dy = kmul(INV(dy),ToReal(0.25)); - CCTK_REAL_VEC const p1o4dydz = kmul(INV(dy),kmul(INV(dz),ToReal(0.25))); + CCTK_REAL_VEC const p1o4dydz = kmul(INV(kmul(dy,dz)),ToReal(0.25)); CCTK_REAL_VEC const p1o4dz = kmul(INV(dz),ToReal(0.25)); CCTK_REAL_VEC const p1o5040dx2 = kmul(INV(SQR(dx)),ToReal(0.000198412698412698412698412698413)); CCTK_REAL_VEC const p1o5040dy2 = kmul(INV(SQR(dy)),ToReal(0.000198412698412698412698412698413)); @@ -127,9 +125,9 @@ static void ML_BSSN_convertFromADMBaseGamma_Body(cGH const * restrict const cctk CCTK_REAL_VEC const p1o64dx = kmul(INV(dx),ToReal(0.015625)); CCTK_REAL_VEC const p1o64dy = kmul(INV(dy),ToReal(0.015625)); CCTK_REAL_VEC const p1o64dz = kmul(INV(dz),ToReal(0.015625)); - CCTK_REAL_VEC const p1o705600dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(1.41723356009070294784580498866e-6))); - CCTK_REAL_VEC const p1o705600dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(1.41723356009070294784580498866e-6))); - CCTK_REAL_VEC const p1o705600dydz = kmul(INV(dy),kmul(INV(dz),ToReal(1.41723356009070294784580498866e-6))); + CCTK_REAL_VEC const p1o705600dxdy = kmul(INV(kmul(dx,dy)),ToReal(1.41723356009070294784580498866e-6)); + CCTK_REAL_VEC const p1o705600dxdz = kmul(INV(kmul(dx,dz)),ToReal(1.41723356009070294784580498866e-6)); + CCTK_REAL_VEC const p1o705600dydz = kmul(INV(kmul(dy,dz)),ToReal(1.41723356009070294784580498866e-6)); CCTK_REAL_VEC const p1o840dx = kmul(INV(dx),ToReal(0.00119047619047619047619047619048)); CCTK_REAL_VEC const p1o840dy = kmul(INV(dy),ToReal(0.00119047619047619047619047619048)); CCTK_REAL_VEC const p1o840dz = kmul(INV(dz),ToReal(0.00119047619047619047619047619048)); @@ -210,7 +208,7 @@ static void ML_BSSN_convertFromADMBaseGamma_Body(cGH const * restrict const cctk /* Loop over the grid points */ #pragma omp parallel - LC_LOOP3VEC (ML_BSSN_convertFromADMBaseGamma, + LC_LOOP3VEC(ML_BSSN_convertFromADMBaseGamma, i,j,k, imin[0],imin[1],imin[2], imax[0],imax[1],imax[2], cctk_lsh[0],cctk_lsh[1],cctk_lsh[2], CCTK_REAL_VEC_SIZE) @@ -841,13 +839,13 @@ static void ML_BSSN_convertFromADMBaseGamma_Body(cGH const * restrict const cctk if (ShiftBCoeff*ShiftGammaCoeff != 0) { B1L = - kmul(INV(theta),kmul(INV(ToReal(ShiftGammaCoeff)),knmsub(kmadd(beta1L,JacPDupwindNthAnti1beta1,kmadd(beta2L,JacPDupwindNthAnti2beta1,kmadd(beta3L,JacPDupwindNthAnti3beta1,kmadd(JacPDupwindNthSymm1beta1,kfabs(beta1L),kmadd(JacPDupwindNthSymm2beta1,kfabs(beta2L),kmul(JacPDupwindNthSymm3beta1,kfabs(beta3L))))))),ToReal(ShiftAdvectionCoeff),dtbetaxL))); + kmul(INV(kmul(theta,ToReal(ShiftGammaCoeff))),knmsub(kmadd(beta1L,JacPDupwindNthAnti1beta1,kmadd(beta2L,JacPDupwindNthAnti2beta1,kmadd(beta3L,JacPDupwindNthAnti3beta1,kmadd(JacPDupwindNthSymm1beta1,kfabs(beta1L),kmadd(JacPDupwindNthSymm2beta1,kfabs(beta2L),kmul(JacPDupwindNthSymm3beta1,kfabs(beta3L))))))),ToReal(ShiftAdvectionCoeff),dtbetaxL)); B2L = - kmul(INV(theta),kmul(INV(ToReal(ShiftGammaCoeff)),knmsub(kmadd(beta1L,JacPDupwindNthAnti1beta2,kmadd(beta2L,JacPDupwindNthAnti2beta2,kmadd(beta3L,JacPDupwindNthAnti3beta2,kmadd(JacPDupwindNthSymm1beta2,kfabs(beta1L),kmadd(JacPDupwindNthSymm2beta2,kfabs(beta2L),kmul(JacPDupwindNthSymm3beta2,kfabs(beta3L))))))),ToReal(ShiftAdvectionCoeff),dtbetayL))); + kmul(INV(kmul(theta,ToReal(ShiftGammaCoeff))),knmsub(kmadd(beta1L,JacPDupwindNthAnti1beta2,kmadd(beta2L,JacPDupwindNthAnti2beta2,kmadd(beta3L,JacPDupwindNthAnti3beta2,kmadd(JacPDupwindNthSymm1beta2,kfabs(beta1L),kmadd(JacPDupwindNthSymm2beta2,kfabs(beta2L),kmul(JacPDupwindNthSymm3beta2,kfabs(beta3L))))))),ToReal(ShiftAdvectionCoeff),dtbetayL)); B3L = - kmul(INV(theta),kmul(INV(ToReal(ShiftGammaCoeff)),knmsub(kmadd(beta1L,JacPDupwindNthAnti1beta3,kmadd(beta2L,JacPDupwindNthAnti2beta3,kmadd(beta3L,JacPDupwindNthAnti3beta3,kmadd(JacPDupwindNthSymm1beta3,kfabs(beta1L),kmadd(JacPDupwindNthSymm2beta3,kfabs(beta2L),kmul(JacPDupwindNthSymm3beta3,kfabs(beta3L))))))),ToReal(ShiftAdvectionCoeff),dtbetazL))); + kmul(INV(kmul(theta,ToReal(ShiftGammaCoeff))),knmsub(kmadd(beta1L,JacPDupwindNthAnti1beta3,kmadd(beta2L,JacPDupwindNthAnti2beta3,kmadd(beta3L,JacPDupwindNthAnti3beta3,kmadd(JacPDupwindNthSymm1beta3,kfabs(beta1L),kmadd(JacPDupwindNthSymm2beta3,kfabs(beta2L),kmul(JacPDupwindNthSymm3beta3,kfabs(beta3L))))))),ToReal(ShiftAdvectionCoeff),dtbetazL)); } else { @@ -858,60 +856,17 @@ static void ML_BSSN_convertFromADMBaseGamma_Body(cGH const * restrict const cctk B3L = ToReal(0); } - /* If necessary, store only partial vectors after the first iteration */ - - if (CCTK_REAL_VEC_SIZE > 2 && CCTK_BUILTIN_EXPECT(i < lc_imin && i+CCTK_REAL_VEC_SIZE > lc_imax, 0)) - { - ptrdiff_t const elt_count_lo = lc_imin-i; - ptrdiff_t const elt_count_hi = lc_imax-i; - vec_store_nta_partial_mid(A[index],AL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(B1[index],B1L,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(B2[index],B2L,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(B3[index],B3L,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(Xt1[index],Xt1L,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(Xt2[index],Xt2L,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(Xt3[index],Xt3L,elt_count_lo,elt_count_hi); - break; - } - - /* If necessary, store only partial vectors after the first iteration */ - - if (CCTK_REAL_VEC_SIZE > 1 && CCTK_BUILTIN_EXPECT(i < lc_imin, 0)) - { - ptrdiff_t const elt_count = lc_imin-i; - vec_store_nta_partial_hi(A[index],AL,elt_count); - vec_store_nta_partial_hi(B1[index],B1L,elt_count); - vec_store_nta_partial_hi(B2[index],B2L,elt_count); - vec_store_nta_partial_hi(B3[index],B3L,elt_count); - vec_store_nta_partial_hi(Xt1[index],Xt1L,elt_count); - vec_store_nta_partial_hi(Xt2[index],Xt2L,elt_count); - vec_store_nta_partial_hi(Xt3[index],Xt3L,elt_count); - continue; - } - - /* If necessary, store only partial vectors after the last iteration */ - - if (CCTK_REAL_VEC_SIZE > 1 && CCTK_BUILTIN_EXPECT(i+CCTK_REAL_VEC_SIZE > lc_imax, 0)) - { - ptrdiff_t const elt_count = lc_imax-i; - vec_store_nta_partial_lo(A[index],AL,elt_count); - vec_store_nta_partial_lo(B1[index],B1L,elt_count); - vec_store_nta_partial_lo(B2[index],B2L,elt_count); - vec_store_nta_partial_lo(B3[index],B3L,elt_count); - vec_store_nta_partial_lo(Xt1[index],Xt1L,elt_count); - vec_store_nta_partial_lo(Xt2[index],Xt2L,elt_count); - vec_store_nta_partial_lo(Xt3[index],Xt3L,elt_count); - break; - } - vec_store_nta(A[index],AL); - vec_store_nta(B1[index],B1L); - vec_store_nta(B2[index],B2L); - vec_store_nta(B3[index],B3L); - vec_store_nta(Xt1[index],Xt1L); - vec_store_nta(Xt2[index],Xt2L); - vec_store_nta(Xt3[index],Xt3L); + /* Copy local copies back to grid functions */ + vec_store_partial_prepare(i,lc_imin,lc_imax); + vec_store_nta_partial(A[index],AL); + vec_store_nta_partial(B1[index],B1L); + vec_store_nta_partial(B2[index],B2L); + vec_store_nta_partial(B3[index],B3L); + vec_store_nta_partial(Xt1[index],Xt1L); + vec_store_nta_partial(Xt2[index],Xt2L); + vec_store_nta_partial(Xt3[index],Xt3L); } - LC_ENDLOOP3VEC (ML_BSSN_convertFromADMBaseGamma); + LC_ENDLOOP3VEC(ML_BSSN_convertFromADMBaseGamma); } extern "C" void ML_BSSN_convertFromADMBaseGamma(CCTK_ARGUMENTS) @@ -930,7 +885,17 @@ extern "C" void ML_BSSN_convertFromADMBaseGamma(CCTK_ARGUMENTS) return; } - const char *groups[] = {"ADMBase::dtlapse","ADMBase::dtshift","grid::coordinates","Grid::coordinates","ML_BSSN::ML_dtlapse","ML_BSSN::ML_dtshift","ML_BSSN::ML_Gamma","ML_BSSN::ML_lapse","ML_BSSN::ML_metric","ML_BSSN::ML_shift"}; + const char *const groups[] = { + "ADMBase::dtlapse", + "ADMBase::dtshift", + "grid::coordinates", + "Grid::coordinates", + "ML_BSSN::ML_dtlapse", + "ML_BSSN::ML_dtshift", + "ML_BSSN::ML_Gamma", + "ML_BSSN::ML_lapse", + "ML_BSSN::ML_metric", + "ML_BSSN::ML_shift"}; GenericFD_AssertGroupStorage(cctkGH, "ML_BSSN_convertFromADMBaseGamma", 10, groups); switch(fdOrder) @@ -952,7 +917,7 @@ extern "C" void ML_BSSN_convertFromADMBaseGamma(CCTK_ARGUMENTS) break; } - GenericFD_LoopOverInterior(cctkGH, &ML_BSSN_convertFromADMBaseGamma_Body); + GenericFD_LoopOverInterior(cctkGH, ML_BSSN_convertFromADMBaseGamma_Body); if (verbose > 1) { |