expec_totalspin.c

Go to the documentation of this file.

/* HPhi  -  Quantum Lattice Model Simulator */
/* Copyright (C) 2015 The University of Tokyo */

/* This program is free software: you can redistribute it and/or modify */
/* it under the terms of the GNU General Public License as published by */
/* the Free Software Foundation, either version 3 of the License, or */
/* (at your option) any later version. */

/* This program is distributed in the hope that it will be useful, */
/* but WITHOUT ANY WARRANTY; without even the implied warranty of */
/* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the */
/* GNU General Public License for more details. */

/* You should have received a copy of the GNU General Public License */
/* along with this program.  If not, see <http://www.gnu.org/licenses/>. */
#include <bitcalc.h>
#include "mltplyCommon.h"
#include "mltplySpinCore.h"
#include "wrapperMPI.h"
#include "mltplyMPISpin.h"
#include "mltplyMPISpinCore.h"
#include "expec_totalspin.h"

int expec_totalspin
(
 struct BindStruct *X,
 double complex *vec
 )
{
  X->Large.mode = M_TOTALS;
  switch(X->Def.iCalcModel){
  case Spin:
     totalspin_Spin(X,vec);
     X->Phys.Sz=X->Def.Total2SzMPI/2.;
     break;
  case SpinGC:
     totalspin_SpinGC(X,vec);
     break;
   case Hubbard:
   case Kondo:
     totalspin_Hubbard(X,vec);
     break;
  case HubbardGC:
  case KondoGC:
     totalspin_HubbardGC(X,vec);
     break;
  default:
    X->Phys.s2=0.0;
    X->Phys.Sz=0.0;
  }
  return 0;
}

void totalspin_Hubbard(struct BindStruct *X,double complex *vec) {
  long unsigned int j;
  long unsigned int irght, ilft, ihfbit;
  long unsigned int isite1, isite2;
  long unsigned int is1_up, is2_up, is1_down, is2_down;
  long unsigned int iexchg, off;
  int num1_up, num2_up;
  int num1_down, num2_down;
  long unsigned int ibit1_up, ibit2_up, ibit1_down, ibit2_down;
  double complex spn_z, tmp_spn_z;
  double complex spn;
  long unsigned i_max;
  i_max = X->Check.idim_max;

  GetSplitBitByModel(X->Def.Nsite, X->Def.iCalcModel, &irght, &ilft, &ihfbit);
  spn = 0.0;
  spn_z = 0.0;
  for (isite1 = 1; isite1 <= X->Def.NsiteMPI; isite1++) {
    is1_up = X->Def.Tpow[2 * isite1 - 2];
    is1_down = X->Def.Tpow[2 * isite1 - 1];

    for (isite2 = 1; isite2 <= X->Def.NsiteMPI; isite2++) {
      is2_up = X->Def.Tpow[2 * isite2 - 2];
      is2_down = X->Def.Tpow[2 * isite2 - 1];

#pragma omp parallel for reduction(+:spn, spn_z) default(none) firstprivate(i_max, is1_up, is1_down, is2_up, is2_down, irght, ilft, ihfbit, isite1, isite2) private(ibit1_up, num1_up, ibit2_up, num2_up, ibit1_down, num1_down, ibit2_down, num2_down, tmp_spn_z, iexchg, off) shared(vec, list_1, list_2_1, list_2_2)
      for (j = 1; j <= i_max; j++) {

        ibit1_up = list_1[j] & is1_up;
        num1_up = ibit1_up / is1_up;
        ibit2_up = list_1[j] & is2_up;
        num2_up = ibit2_up / is2_up;

        ibit2_down = list_1[j] & is2_down;
        num2_down = ibit2_down / is2_down;
        ibit1_down = list_1[j] & is1_down;
        num1_down = ibit1_down / is1_down;

        tmp_spn_z = (num1_up - num1_down) * (num2_up - num2_down);
        spn += conj(vec[j]) * vec[j] * tmp_spn_z / 4.0;
        if (isite1 == isite2) {
          spn += conj(vec[j]) * vec[j] * (num1_up + num1_down - 2 * num1_up * num1_down) / 2.0;
          spn_z += conj(vec[j]) * vec[j] * (num1_up - num1_down) / 2.0;
        } else {
          if (ibit1_up != 0 && ibit1_down == 0 && ibit2_up == 0 && ibit2_down != 0) {
            iexchg = list_1[j] - (is1_up + is2_down);
            iexchg += (is2_up + is1_down);
            GetOffComp(list_2_1, list_2_2, iexchg, irght, ilft, ihfbit, &off);
            spn += conj(vec[j]) * vec[off] / 2.0;
          } else if (ibit1_up == 0 && ibit1_down != 0 && ibit2_up != 0 && ibit2_down == 0) {
            iexchg = list_1[j] - (is1_down + is2_up);
            iexchg += (is2_down + is1_up);
            GetOffComp(list_2_1, list_2_2, iexchg, irght, ilft, ihfbit, &off);
            spn += conj(vec[j]) * vec[off] / 2.0;
          }
        }
      }
    }
  }
  X->Phys.s2 = creal(spn);
  X->Phys.Sz = creal(spn_z);
}


void totalspin_HubbardGC(struct BindStruct *X,double complex *vec) {
  long unsigned int j;
  long unsigned int isite1, isite2;
  long unsigned int is1_up, is2_up, is1_down, is2_down;
  long unsigned int iexchg, off;
  int num1_up, num2_up;
  int num1_down, num2_down;
  long unsigned int ibit1_up, ibit2_up, ibit1_down, ibit2_down, list_1_j;
  double complex spn_z, tmp_spn_z;
  double complex spn;
  long unsigned int i_max;

  i_max = X->Check.idim_max;

  spn = 0.0;
  spn_z = 0.0;
  for (isite1 = 1; isite1 <= X->Def.NsiteMPI; isite1++) {
    for (isite2 = 1; isite2 <= X->Def.NsiteMPI; isite2++) {
      is1_up = X->Def.Tpow[2 * isite1 - 2];
      is1_down = X->Def.Tpow[2 * isite1 - 1];
      is2_up = X->Def.Tpow[2 * isite2 - 2];
      is2_down = X->Def.Tpow[2 * isite2 - 1];

#pragma omp parallel for reduction(+:spn, spn_z) default(none) firstprivate(i_max, is1_up, is1_down, is2_up, is2_down, isite1, isite2) private(list_1_j, ibit1_up, num1_up, ibit2_up, num2_up, ibit1_down, num1_down, ibit2_down, num2_down, tmp_spn_z, iexchg, off) shared(vec)
      for (j = 1; j <= i_max; j++) {
        list_1_j = j - 1;
        ibit1_up = list_1_j & is1_up;
        num1_up = ibit1_up / is1_up;
        ibit2_up = list_1_j & is2_up;
        num2_up = ibit2_up / is2_up;

        ibit1_down = list_1_j & is1_down;
        num1_down = ibit1_down / is1_down;
        ibit2_down = list_1_j & is2_down;
        num2_down = ibit2_down / is2_down;

        tmp_spn_z = (num1_up - num1_down) * (num2_up - num2_down);
        spn += conj(vec[j]) * vec[j] * tmp_spn_z / 4.0;
        if (isite1 == isite2) {
          spn += conj(vec[j]) * vec[j] * (num1_up + num1_down - 2 * num1_up * num1_down) / 2.0;
          spn_z += conj(vec[j]) * vec[j] * (num1_up - num1_down) / 2.0;
        } else {
          if (ibit1_up != 0 && ibit1_down == 0 && ibit2_up == 0 && ibit2_down != 0) {
            iexchg = list_1_j - (is1_up + is2_down);
            iexchg += (is2_up + is1_down);
            off = iexchg + 1;
            spn += conj(vec[j]) * vec[off] / 2.0;
          } else if (ibit1_up == 0 && ibit1_down != 0 && ibit2_up != 0 && ibit2_down == 0) {
            iexchg = list_1_j - (is1_down + is2_up);
            iexchg += (is2_down + is1_up);
            off = iexchg + 1;
            spn += conj(vec[j]) * vec[off] / 2.0;
          }
        }
      }
    }
  }
  X->Phys.s2 = creal(spn);
  X->Phys.Sz = creal(spn_z);
}

void totalspin_Spin(struct BindStruct *X,double complex *vec) {

  long unsigned int j;
  long unsigned int irght, ilft, ihfbit;
  long unsigned int isite1, isite2;
  long unsigned int tmp_isite1, tmp_isite2;

  long unsigned int is1_up, is2_up;
  long unsigned int iexchg, off, off_2;

  int num1_up, num2_up;
  int num1_down, num2_down;
  int sigma_1, sigma_2;
  long unsigned int ibit1_up, ibit2_up, ibit_tmp, is_up;
  double complex spn_z = 0.0;
  double complex spn_z1, spn_z2, spn_zd;
  double complex spn = 0.0;
  long unsigned int i_max;

  i_max = X->Check.idim_max;
  if (X->Def.iFlgGeneralSpin == FALSE) {
    GetSplitBitByModel(X->Def.Nsite, X->Def.iCalcModel, &irght, &ilft, &ihfbit);
    spn = 0.0;
    spn_z = 0.0;
    spn_zd = 0.0;
    for (isite1 = 1; isite1 <= X->Def.NsiteMPI; isite1++) {
      for (isite2 = 1; isite2 <= X->Def.NsiteMPI; isite2++) {

        if (isite1 > X->Def.Nsite && isite2 > X->Def.Nsite) {
#ifdef MPI
          is1_up = X->Def.Tpow[isite1 - 1];
          is2_up = X->Def.Tpow[isite2 - 1];
          is_up = is1_up + is2_up;
          num1_up = X_SpinGC_CisAis((unsigned long int) myrank + 1, X, is1_up, 1);
          num1_down = 1 - num1_up;
          num2_up = X_SpinGC_CisAis((unsigned long int) myrank + 1, X, is2_up, 1);
          num2_down = 1 - num2_up;
          spn_z = (num1_up - num1_down) * (num2_up - num2_down);

#pragma omp parallel for default(none) reduction (+:spn_zd) shared(vec)  \
  firstprivate(i_max, spn_z) private(j)
          for (j = 1; j <= i_max; j++) {
            spn_zd += conj(vec[j]) * vec[j] * spn_z / 4.0;
          }
          if (isite1 == isite2) {
#pragma omp parallel for default(none) reduction (+:spn_zd) shared(vec)  \
  firstprivate(i_max) private(j)
            for (j = 1; j <= i_max; j++) {
              spn_zd += conj(vec[j]) * vec[j] / 2.0;
            }
          } else {//off diagonal
            spn += X_child_general_int_spin_TotalS_MPIdouble(isite1 - 1, isite2 - 1, X, vec, vec);
          }
#endif
        } else if (isite1 > X->Def.Nsite || isite2 > X->Def.Nsite) {
#ifdef MPI
          if (isite1 < isite2) {
            tmp_isite1 = isite1;
            tmp_isite2 = isite2;
          } else {
            tmp_isite1 = isite2;
            tmp_isite2 = isite1;
          }

          is1_up = X->Def.Tpow[tmp_isite1 - 1];
          is2_up = X->Def.Tpow[tmp_isite2 - 1];
          num2_up = X_SpinGC_CisAis((unsigned long int) myrank + 1, X, is2_up, 1);
          num2_down = 1 - num2_up;

          //diagonal
#pragma omp parallel for reduction(+: spn_zd) default(none) firstprivate(i_max, is1_up, num2_up, num2_down) private(ibit1_up, num1_up, num1_down, spn_z) shared(list_1, vec)
          for (j = 1; j <= i_max; j++) {
            ibit1_up = list_1[j] & is1_up;
            num1_up = ibit1_up / is1_up;
            num1_down = 1 - num1_up;
            spn_z = (num1_up - num1_down) * (num2_up - num2_down);
            spn_zd += conj(vec[j]) * vec[j] * spn_z / 4.0;
          }
          if (isite1 < isite2) {
            spn += X_child_general_int_spin_MPIsingle(isite1 - 1, 0, 1, isite2 - 1, 1, 0, 1.0, X, vec, vec);
          } else {
            spn += conj(X_child_general_int_spin_MPIsingle(isite2 - 1, 1, 0, isite1 - 1, 0, 1, 1.0, X, vec, vec));
          }
#endif
        }//isite1 > Nsite || isite2 > Nsite
        else {
          is1_up = X->Def.Tpow[isite1 - 1];
          is2_up = X->Def.Tpow[isite2 - 1];
          is_up = is1_up + is2_up;

#pragma omp parallel for reduction(+: spn, spn_zd) default(none) firstprivate(i_max, is_up, is1_up, is2_up, irght, ilft, ihfbit, isite1, isite2) private(ibit1_up, num1_up, ibit2_up, num2_up, num1_down, num2_down, spn_z, iexchg, off, ibit_tmp) shared(list_1, list_2_1, list_2_2, vec)
          for (j = 1; j <= i_max; j++) {
            ibit1_up = list_1[j] & is1_up;
            num1_up = ibit1_up / is1_up;
            num1_down = 1 - num1_up;
            ibit2_up = list_1[j] & is2_up;
            num2_up = ibit2_up / is2_up;
            num2_down = 1 - num2_up;

            spn_z = (num1_up - num1_down) * (num2_up - num2_down);
            spn_zd += conj(vec[j]) * vec[j] * spn_z / 4.0;

            if (isite1 == isite2) {
              spn_zd += conj(vec[j]) * vec[j] / 2.0;
            } else {
              ibit_tmp = (num1_up) ^ (num2_up);
              if (ibit_tmp != 0) {
                iexchg = list_1[j] ^ (is_up);
                GetOffComp(list_2_1, list_2_2, iexchg, irght, ilft, ihfbit, &off);
                spn += conj(vec[j]) * vec[off] / 2.0;
              }
            }
          }// j
        }
      }//isite2
    }//isite1
  }//generalspin=FALSE
  else {
    double S1 = 0;
    double S2 = 0;
    for (isite1 = 1; isite1 <= X->Def.NsiteMPI; isite1++) {
      for (isite2 = 1; isite2 <= X->Def.NsiteMPI; isite2++) {
        S1 = 0.5 * (X->Def.SiteToBit[isite1 - 1] - 1);
        S2 = 0.5 * (X->Def.SiteToBit[isite2 - 1] - 1);
        if (isite1 == isite2) {
#pragma omp parallel for reduction(+: spn, spn_z) default(none) firstprivate(i_max, isite1, X, S1) private (spn_z1)shared(vec, list_1)
          for (j = 1; j <= i_max; j++) {
            spn_z1 = 0.5 * GetLocal2Sz(isite1, list_1[j], X->Def.SiteToBit, X->Def.Tpow);
            spn += conj(vec[j]) * vec[j] * S1 * (S1 + 1.0);
            spn_z += conj(vec[j]) * vec[j] * spn_z1;
          }
        } else {
#pragma omp parallel for reduction(+: spn) default(none) firstprivate(i_max, isite1, isite2, X, S1, S2) private(spn_z1, spn_z2, off, off_2, ibit_tmp, sigma_1, sigma_2) shared(vec, list_1)
          for (j = 1; j <= i_max; j++) {
            spn_z1 = 0.5 * GetLocal2Sz(isite1, list_1[j], X->Def.SiteToBit, X->Def.Tpow);
            spn_z2 = 0.5 * GetLocal2Sz(isite2, list_1[j], X->Def.SiteToBit, X->Def.Tpow);
            spn += conj(vec[j]) * vec[j] * spn_z1 * spn_z2;

            sigma_1 = GetBitGeneral(isite1, list_1[j], X->Def.SiteToBit, X->Def.Tpow);
            sigma_2 = GetBitGeneral(isite2, list_1[j], X->Def.SiteToBit, X->Def.Tpow);

            ibit_tmp = GetOffCompGeneralSpin(list_1[j], isite2, sigma_2, sigma_2 + 1, &off, X->Def.SiteToBit,
                                             X->Def.Tpow);
            if (ibit_tmp == TRUE) {
              ibit_tmp = GetOffCompGeneralSpin(off, isite1, sigma_1, sigma_1 - 1, &off_2, X->Def.SiteToBit,
                                               X->Def.Tpow);
              if (ibit_tmp == TRUE) {
                ConvertToList1GeneralSpin(off_2, X->Check.sdim, &off);
                spn += conj(vec[j]) * vec[off] * sqrt(S2 * (S2 + 1) - spn_z2 * (spn_z2 + 1)) *
                       sqrt(S1 * (S1 + 1) - spn_z1 * (spn_z1 - 1)) / 2.0;
              }
            }

            ibit_tmp = GetOffCompGeneralSpin(list_1[j], isite2, sigma_2, sigma_2 - 1, &off, X->Def.SiteToBit,
                                             X->Def.Tpow);
            if (ibit_tmp == TRUE) {
              ibit_tmp = GetOffCompGeneralSpin(off, isite1, sigma_1, sigma_1 + 1, &off_2, X->Def.SiteToBit,
                                               X->Def.Tpow);
              if (ibit_tmp == TRUE) {
                ConvertToList1GeneralSpin(off_2, X->Check.sdim, &off);
                spn += conj(vec[j]) * vec[off] * sqrt(S2 * (S2 + 1) - spn_z2 * (spn_z2 - 1.0)) *
                       sqrt(S1 * (S1 + 1) - spn_z1 * (spn_z1 + 1)) / 2.0;
              }
            }
          }
        }
      }
    }
  }

  spn = SumMPI_dc(spn);
  spn_zd = SumMPI_dc(spn_zd);
  spn_z = SumMPI_dc(spn_z);
  spn += spn_zd;
  X->Phys.s2 = creal(spn);
  X->Phys.Sz = creal(spn_z);
}


void totalspin_SpinGC(struct BindStruct *X,double complex *vec){

  long unsigned int j;
  long unsigned int isite1,isite2, tmp_isite1, tmp_isite2;
  long unsigned int is1_up,is2_up;
  long unsigned int iexchg, off, off_2;
  int num1_up,num2_up;
  int num1_down,num2_down;
  int sigma_1, sigma_2;
  long unsigned int ibit1_up,ibit2_up,ibit_tmp,is_up;
  double complex spn_z;
  double complex spn_z1, spn_z2;
  double complex spn, spn_d;
  long unsigned int list_1_j;
  long unsigned int i_max;
  i_max=X->Check.idim_max;
  X->Large.mode = M_TOTALS;
  spn=0.0;
  spn_d=0.0;
  spn_z=0.0;
  if(X->Def.iFlgGeneralSpin==FALSE){
    for(isite1=1;isite1<=X->Def.NsiteMPI;isite1++){
      if(isite1 > X->Def.Nsite){
  is1_up      = X->Def.Tpow[isite1-1];
  ibit1_up = myrank&is1_up;
  num1_up = ibit1_up/is1_up;
  num1_down =1-num1_up;
#pragma omp parallel for reduction(+: spn_z) default(none) firstprivate(i_max, is1_up,  num1_up, num1_down) shared(vec)
  for(j=1;j<=i_max;j++){
    spn_z  +=  conj(vec[j])*vec[j]*(num1_up-num1_down)/2.0;
  }
      }
      else{
  is1_up      = X->Def.Tpow[isite1-1];
#pragma omp parallel for reduction(+: spn_z) default(none) firstprivate(i_max, is1_up) private(list_1_j, ibit1_up, num1_up, num1_down) shared(vec)
  for(j=1;j<=i_max;j++){
    list_1_j=j-1;
    ibit1_up  = list_1_j&is1_up;
    num1_up   = ibit1_up/is1_up;
    num1_down = 1-num1_up;
    spn_z  +=  conj(vec[j])*vec[j]*(num1_up-num1_down)/2.0;
  }

      }
      for(isite2=1;isite2<=X->Def.NsiteMPI;isite2++){

  if(isite1 > X->Def.Nsite && isite2 > X->Def.Nsite){
    is1_up      = X->Def.Tpow[isite1-1];
    is2_up      = X->Def.Tpow[isite2-1];
    num1_up = X_SpinGC_CisAis((unsigned long int)myrank + 1, X, is1_up, 1);
    num1_down = 1-num1_up;
    num2_up = X_SpinGC_CisAis((unsigned long int)myrank + 1, X, is2_up, 1);
    num2_down = 1-num2_up;
    spn_z2  = (num1_up-num1_down)*(num2_up-num2_down)/4.0;
#pragma omp parallel for default(none) reduction (+:spn_d) shared(vec)  \
  firstprivate(i_max, spn_z2) private(j)
    for (j = 1; j <= i_max; j++) {
      spn_d   += conj(vec[j])*vec[j]*spn_z2;
    }
    if(isite1 == isite2){
#pragma omp parallel for default(none) reduction (+:spn_d) shared(vec)  \
  firstprivate(i_max) private(j)
      for (j = 1; j <= i_max; j++) {
        spn_d      += conj(vec[j])*vec[j]/2.0;
      }
    }//isite1 = isite2
    else{//off diagonal
      spn += X_GC_child_CisAitCiuAiv_spin_MPIdouble(isite1-1, 0, 1, isite2-1, 1, 0, 1.0, X, vec, vec)/2.0;
    }
  }
  else if(isite1 > X->Def.Nsite || isite2 > X->Def.Nsite){
    if(isite1 < isite2){
      tmp_isite1=isite1;
      tmp_isite2=isite2;
    }
    else{
      tmp_isite1 = isite2;
      tmp_isite2 = isite1;
    }
    is1_up = X->Def.Tpow[tmp_isite1 - 1];
    is2_up = X->Def.Tpow[tmp_isite2 - 1];
    num2_up = X_SpinGC_CisAis((unsigned long int)myrank + 1, X, is2_up, 1);
    num2_down =1-num2_up;
    //diagonal
#pragma omp parallel for reduction(+: spn_d) default(none) firstprivate(i_max, is1_up, num2_up, num2_down) private(ibit1_up, num1_up, num1_down, spn_z2, list_1_j) shared(vec)
    for(j=1;j<=i_max;j++){
      list_1_j=j-1;
      ibit1_up  = list_1_j&is1_up;
      num1_up   = ibit1_up/is1_up;
      num1_down = 1-num1_up;
      spn_z2  = (num1_up-num1_down)*(num2_up-num2_down);
      spn_d   += conj(vec[j])*vec[j]*spn_z2/4.0;
    }
    if(isite1 < isite2){
      spn += X_GC_child_CisAitCiuAiv_spin_MPIsingle(isite1-1, 0, 1, isite2-1, 1, 0, 1.0, X, vec, vec)/2.0;
    }
    else{
      spn += conj(X_GC_child_CisAitCiuAiv_spin_MPIsingle(isite2-1, 1, 0, isite1-1, 0, 1, 1.0, X, vec, vec))/2.0;
    }
  }
  else{
    is2_up      = X->Def.Tpow[isite2-1];
    is_up       = is1_up+is2_up;
#pragma omp parallel for reduction(+: spn, spn_d) default(none) firstprivate(i_max, is_up, is1_up, is2_up, isite1, isite2) private(list_1_j, ibit1_up, num1_up, ibit2_up, num2_up, num1_down, num2_down, spn_z2, iexchg, off, ibit_tmp) shared(vec)
    for(j=1;j<=i_max;j++){
      list_1_j=j-1;
      ibit1_up  = list_1_j&is1_up;
      num1_up   = ibit1_up/is1_up;
      num1_down = 1-num1_up;
      ibit2_up  = list_1_j&is2_up;
      num2_up   = ibit2_up/is2_up;
      num2_down = 1-num2_up;

      spn_z2  = (num1_up-num1_down)*(num2_up-num2_down);
      spn_d   += conj(vec[j])*vec[j]*spn_z2/4.0;

      if(isite1==isite2){
        spn_d      += conj(vec[j])*vec[j]/2.0;
      }else{
        ibit_tmp  = (num1_up) ^ (num2_up);
        if(ibit_tmp!=0){
    iexchg  = list_1_j ^ (is_up);
    off    = iexchg+1;
    spn    += conj(vec[j])*vec[off]/2.0;
        }

      }
    }//j
  }//else
      }
    }
  }
  else{//general spin
    double S1=0;
    double S2=0;
    spn =0.0;
    spn_z=0.0;
    for(isite1=1;isite1<=X->Def.NsiteMPI;isite1++){
      S1=0.5*(X->Def.SiteToBit[isite1-1]-1);
      if(isite1 > X->Def.Nsite){
  spn_z1  = 0.5*GetLocal2Sz(isite1, (unsigned long int) myrank, X->Def.SiteToBit, X->Def.Tpow);
#pragma omp parallel for reduction(+: spn, spn_z) default(none) firstprivate(S1, spn_z1,i_max) shared(vec)
  for(j=1;j<=i_max;j++){
    spn   += conj(vec[j])*vec[j]*S1*(S1+1.0);
    spn_z += conj(vec[j])*vec[j]*spn_z1;
  }
      }
      else{
#pragma omp parallel for reduction(+: spn, spn_z) default(none) firstprivate(i_max, isite1, X, S1) private(spn_z1) shared(vec)
  for(j=1;j<=i_max;j++){
    spn_z1  = 0.5*GetLocal2Sz(isite1, j-1, X->Def.SiteToBit, X->Def.Tpow);
    spn    += conj(vec[j])*vec[j]*S1*(S1+1.0);
    spn_z += conj(vec[j])*vec[j]*spn_z1;
  }
      }
      for(isite2=1;isite2<=X->Def.NsiteMPI;isite2++){
  if(isite1==isite2) continue;
  S2=0.5*(X->Def.SiteToBit[isite2-1]-1);

  if(isite1 > X->Def.Nsite && isite2 > X->Def.Nsite){
    /*
    spn_z1  = 0.5*GetLocal2Sz(isite1, (unsigned long int) myrank, X->Def.SiteToBit, X->Def.Tpow);
    spn_z2  = 0.5*GetLocal2Sz(isite2, (unsigned long int) myrank, X->Def.SiteToBit, X->Def.Tpow);
#pragma omp parallel for reduction(+: spn, spn_z) default(none) firstprivate(spn_z1, spn_z2, i_max) shared(vec)
    for(j=1;j<=i_max;j++){
      spn   += conj(vec[j])*vec[j]*spn_z1*spn_z2;
    }
    tmp_V= sqrt(S2*(S2+1) - spn_z2*(spn_z2+1))*sqrt(S1*(S1+1) - spn_z1*(spn_z1-1))/2.0;
    spn += X_GC_child_CisAitCjuAjv_GeneralSpin_MPIdouble(isite1-1, sigma_1-1, sigma_1, isite2-1, sigma_2+1, sigma_2, tmp_V, X,vec, vec);
    tmp_V= sqrt(S2*(S2+1) - spn_z2*(spn_z2-1))*sqrt(S1*(S1+1) - spn_z1*(spn_z1+1))/2.0;
    spn += X_GC_child_CisAitCjuAjv_GeneralSpin_MPIdouble(isite1-1, sigma_1+1, sigma_1, isite2-1, sigma_2-1, sigma_2, tmp_V, X,vec, vec);
    */
  }
  else if(isite1 > X->Def.Nsite || isite2 > X->Def.Nsite){
    /*
   if(isite1 < isite2){
      tmp_isite1=isite1;
      tmp_isite2=isite2;
    }
    else{
      tmp_isite1 = isite2;
      tmp_isite2 = isite1;
    }
   spn_z2  = 0.5*GetLocal2Sz(tmp_isite2, (unsigned long int) myrank, X->Def.SiteToBit, X->Def.Tpow);
   sigma_2 = GetBitGeneral(tmp_isite2, (unsigned long int) myrank, X->Def.SiteToBit, X->Def.Tpow);
    */
  }
  else{ //inner-process
#pragma omp parallel for reduction(+: spn) default(none) firstprivate(i_max, isite1, isite2, X, S1, S2) private(spn_z1, spn_z2, off, off_2, ibit_tmp, sigma_1, sigma_2) shared(vec)
    for(j=1;j<=i_max;j++){
      spn_z1  = 0.5*GetLocal2Sz(isite1, j-1, X->Def.SiteToBit, X->Def.Tpow);
      spn_z2  = 0.5*GetLocal2Sz(isite2, j-1, X->Def.SiteToBit, X->Def.Tpow);
      spn    += conj(vec[j])*vec[j]*spn_z1*spn_z2;

      sigma_1=GetBitGeneral(isite1, j-1, X->Def.SiteToBit, X->Def.Tpow);
      sigma_2=GetBitGeneral(isite2, j-1, X->Def.SiteToBit, X->Def.Tpow);

      ibit_tmp = GetOffCompGeneralSpin(j-1, isite2, sigma_2, sigma_2+1, &off, X->Def.SiteToBit, X->Def.Tpow);
      if(ibit_tmp!=0){
        ibit_tmp = GetOffCompGeneralSpin(off, isite1, sigma_1, sigma_1-1,&off_2, X->Def.SiteToBit, X->Def.Tpow);
        if(ibit_tmp!=0){
    spn += conj(vec[j])*vec[off_2+1]*sqrt(S2*(S2+1) - spn_z2*(spn_z2+1))*sqrt(S1*(S1+1) - spn_z1*(spn_z1-1))/2.0;
        }
      }

      ibit_tmp = GetOffCompGeneralSpin(j-1, isite2, sigma_2, sigma_2-1, &off, X->Def.SiteToBit, X->Def.Tpow);
      if(ibit_tmp !=0){
        ibit_tmp = GetOffCompGeneralSpin(off, isite1, sigma_1, sigma_1+1, &off_2, X->Def.SiteToBit, X->Def.Tpow);
        if(ibit_tmp!=0){
    spn += conj(vec[j])*vec[off_2+1]*sqrt(S2*(S2+1) - spn_z2*(spn_z2-1.0))*sqrt(S1*(S1+1)- spn_z1*(spn_z1+1))/2.0;
        }
      }
    }//j
  }//inner-process
      }//isite2
    }//isite1
  }

  spn = SumMPI_dc(spn);
  spn_d = SumMPI_dc(spn_d);
  spn_z = SumMPI_dc(spn_z);
  X->Phys.s2=creal(spn+spn_d);
  X->Phys.Sz=creal(spn_z);
}

int expec_totalSz(
 struct BindStruct *X,
 double complex *vec
) {
  X->Large.mode = M_TOTALS;
  switch (X->Def.iCalcModel) {
    case Spin:
      X->Phys.Sz = X->Def.Total2SzMPI / 2.;
      break;
    case SpinGC:
      totalSz_SpinGC(X, vec);
      break;
    case Hubbard:
    case Kondo:
      X->Phys.Sz = X->Def.Total2SzMPI / 2.;

      break;
    case HubbardGC:
    case KondoGC:
      totalSz_HubbardGC(X, vec);
      break;
    default:
      X->Phys.Sz = 0.0;
  }

  return 0;
}


void totalSz_HubbardGC
(
 struct BindStruct *X,
 double complex *vec
 ) {

  long unsigned int j;
  long unsigned int isite1;
  long unsigned int is1_up, is1_down;
  int num1_up, num1_down, num1_sz;
  long unsigned int ibit1_up, ibit1_down, list_1_j;
  double complex spn_z;
  long unsigned int i_max;

  i_max = X->Check.idim_max;
  spn_z = 0.0;
  for (isite1 = 1; isite1 <= X->Def.NsiteMPI; isite1++) {
    if (isite1 > X->Def.Nsite) {
#ifdef MPI
      is1_up = X->Def.Tpow[2 * isite1 - 2];
      is1_down = X->Def.Tpow[2 * isite1 - 1];
      ibit1_up = (unsigned long int) myrank & is1_up;
      num1_up = ibit1_up / is1_up;
      ibit1_down = (unsigned long int) myrank & is1_down;
      num1_down = ibit1_down / is1_down;
      num1_sz = num1_up - num1_down;
#pragma omp parallel for reduction(+:spn_z) default(none) firstprivate(i_max) private(j) shared(num1_sz,vec)
      for (j = 1; j <= i_max; j++) {
        spn_z += (num1_sz) / 2.0 * conj(vec[j]) * vec[j];
      }
#endif
    } else {//isite1 > X->Def.Nsite
      is1_up = X->Def.Tpow[2 * isite1 - 2];
      is1_down = X->Def.Tpow[2 * isite1 - 1];
#pragma omp parallel for reduction(+:spn_z) default(none) firstprivate(i_max, is1_up, is1_down, isite1) \
  private(list_1_j, ibit1_up, num1_up, ibit1_down, num1_down) shared(vec)
      for (j = 1; j <= i_max; j++) {
        list_1_j = j - 1;
        ibit1_up = list_1_j & is1_up;
        num1_up = ibit1_up / is1_up;
        ibit1_down = list_1_j & is1_down;
        num1_down = ibit1_down / is1_down;
        spn_z += conj(vec[j]) * vec[j] * (num1_up - num1_down) / 2.0;
      }
    }
  }
  spn_z = SumMPI_dc(spn_z);
  X->Phys.Sz = creal(spn_z);
}

void totalSz_SpinGC
(
 struct BindStruct *X,
 double complex *vec
 ) {
  long unsigned int j, list_1_j;
  long unsigned int isite1;
  long unsigned int is1_up;
  int num1_up;
  int num1_down;
  long unsigned int ibit1_up;
  double complex spn_z, spn_z1;
  long unsigned int i_max;
  i_max = X->Check.idim_max;
  X->Large.mode = M_TOTALS;
  spn_z = 0.0;
  if (X->Def.iFlgGeneralSpin == FALSE) {
    for (isite1 = 1; isite1 <= X->Def.NsiteMPI; isite1++) {
      if (isite1 > X->Def.Nsite) {
#ifdef MPI
        is1_up = X->Def.Tpow[isite1 - 1];
        ibit1_up = myrank & is1_up;
        num1_up = ibit1_up / is1_up;
        num1_down = 1 - num1_up;
#pragma omp parallel for reduction(+: spn_z) default(none) firstprivate(i_max, is1_up, num1_up, num1_down) shared(vec)
        for (j = 1; j <= i_max; j++) {
          spn_z += conj(vec[j]) * vec[j] * (num1_up - num1_down) / 2.0;
        }
#endif
      } else {
        is1_up = X->Def.Tpow[isite1 - 1];
#pragma omp parallel for reduction(+: spn_z) default(none) firstprivate(i_max, is1_up) private(list_1_j, ibit1_up, num1_up, num1_down) shared(vec)
        for (j = 1; j <= i_max; j++) {
          list_1_j = j - 1;
          ibit1_up = list_1_j & is1_up;
          num1_up = ibit1_up / is1_up;
          num1_down = 1 - num1_up;
          spn_z += conj(vec[j]) * vec[j] * (num1_up - num1_down) / 2.0;
        }
      }//else
    }
  } else {//general spin
    spn_z = 0.0;
    for (isite1 = 1; isite1 <= X->Def.NsiteMPI; isite1++) {
      if (isite1 > X->Def.Nsite) {
        spn_z1 = 0.5 * GetLocal2Sz(isite1, (unsigned long int) myrank, X->Def.SiteToBit, X->Def.Tpow);
#pragma omp parallel for reduction(+: spn_z) default(none) firstprivate(spn_z1, i_max) shared(vec)
        for (j = 1; j <= i_max; j++) {
          spn_z += conj(vec[j]) * vec[j] * spn_z1;
        }
      } else {
#pragma omp parallel for reduction(+:  spn_z) default(none) firstprivate(i_max, isite1, X) private(spn_z1) shared(vec)
        for (j = 1; j <= i_max; j++) {
          spn_z1 = 0.5 * GetLocal2Sz(isite1, j - 1, X->Def.SiteToBit, X->Def.Tpow);
          spn_z += conj(vec[j]) * vec[j] * spn_z1;
        }
      }
    }//isite1
  }
  spn_z = SumMPI_dc(spn_z);
  X->Phys.Sz = creal(spn_z);
}