subroutine WHAM_CALC(islice,*) ! Weighed Histogram Analysis Method (WHAM) code ! Written by A. Liwo based on the work of Kumar et al., ! J.Comput.Chem., 13, 1011 (1992) ! ! 2/1/05 Multiple temperatures allowed. ! 2/2/05 Free energies calculated directly from data points ! acc. to Eq. (21) of Kumar et al.; final histograms also ! constructed based on this equation. ! 2/12/05 Multiple parameter sets included ! ! 2/2/05 Parallel version implicit none include "DIMENSIONS" include "DIMENSIONS.ZSCOPT" include "DIMENSIONS.FREE" integer nGridT parameter (NGridT=400) integer MaxBinRms,MaxBinRgy parameter (MaxBinRms=100,MaxBinRgy=100) integer MaxHdim c parameter (MaxHdim=200000) parameter (MaxHdim=200) integer maxinde parameter (maxinde=200) #ifdef MPI include "mpif.h" include "COMMON.MPI" integer ierror,errcode,status(MPI_STATUS_SIZE) #endif include "COMMON.CONTROL" include "COMMON.IOUNITS" include "COMMON.FREE" include "COMMON.ENERGIES" include "COMMON.FFIELD" include "COMMON.SBRIDGE" include "COMMON.PROT" include "COMMON.ENEPS" integer MaxPoint,MaxPointProc parameter (MaxPoint=MaxStr, & MaxPointProc=MaxStr_Proc) double precision finorm_max,potfac,entmin,entmax,expfac,vf parameter (finorm_max=1.0d0) integer islice integer i,ii,j,jj,k,kk,l,m,ind,iter,t,tmax,ient,ientmax,iln integer start,end,iharm,ib,iib,nbin1,nbin,nbin_rms,nbin_rgy, & nbin_rmsrgy,liczba,iparm,nFi,indrgy,indrms integer htot(0:MaxHdim),histent(0:2000) double precision v(MaxPointProc,MaxR,MaxT_h,Max_Parm) double precision energia(0:max_ene) #ifdef MPI integer tmax_t,upindE_p double precision fi_p(MaxR,MaxT_h,Max_Parm) double precision sumW_p(0:nGridT,Max_Parm), & sumE_p(0:nGridT,Max_Parm),sumEsq_p(0:nGridT,Max_Parm), & sumQ_p(MaxQ1,0:nGridT,Max_Parm), & sumQsq_p(MaxQ1,0:nGridT,Max_Parm), & sumEQ_p(MaxQ1,0:nGridT,Max_Parm), & sumEprim_p(MaxQ1,0:nGridT,Max_Parm), & sumEbis_p(0:nGridT,Max_Parm) double precision hfin_p(0:MaxHdim,maxT_h), & hfin_ent_p(0:MaxHdim),histE_p(0:maxindE),sumH, & hrmsrgy_p(0:MaxBinRgy,0:MaxBinRms,maxT_h) double precision rgymin_t,rmsmin_t,rgymax_t,rmsmax_t double precision potEmin_t,entmin_p,entmax_p integer histent_p(0:2000) logical lprint /.true./ #endif double precision delta_T /1.0d0/ double precision rgymin,rmsmin,rgymax,rmsmax double precision sumW(0:NGridT,Max_Parm),sumE(0:NGridT,Max_Parm), & sumEsq(0:NGridT,Max_Parm),sumQ(MaxQ1,0:NGridT,Max_Parm), & sumQsq(MaxQ1,0:NGridT,Max_Parm),sumEQ(MaxQ1,0:NGridT,Max_Parm), & sumEprim(0:NGridT,Max_Parm),sumEbis(0:NGridT,Max_Parm),betaT, & weight,econstr double precision fi(MaxR,maxT_h,Max_Parm), & dd,dd1,dd2,hh,dmin,denom,finorm,avefi,pom, & hfin(0:MaxHdim,maxT_h),histE(0:maxindE), & hrmsrgy(0:MaxBinRgy,0:MaxBinRms,maxT_h), & potEmin,ent, & hfin_ent(0:MaxHdim),vmax,aux double precision fT(6),fTprim(6),fTbis(6),quot,quotl1,quotl,kfacl, & eprim,ebis,temper,kfac/2.4d0/,T0/300.0d0/,startGridT/200.0d0/, & eplus,eminus,logfac,tanhT,tt double precision etot,evdw,evdw_t,evdw2,ees,evdw1,ebe,etors, & escloc,ehpb,ecorr,ecorr5,ecorr6,eello_turn4,eello_turn3, & eturn6,eel_loc,edihcnstr,etors_d,estr,evdw2_14,esccor integer ind_point(maxpoint),upindE,indE character*16 plik character*1 licz1 character*2 licz2 character*3 licz3 character*128 nazwa integer ilen external ilen write(licz2,'(bz,i2.2)') islice nbin1 = 1.0d0/delta write (iout,'(//80(1h-)/"Solving WHAM equations for slice", & i2/80(1h-)//)') islice write (iout,*) "delta",delta," nbin1",nbin1 write (iout,*) "MaxN",MaxN," MaxQ",MaxQ," MaHdim",MaxHdim call flush(iout) dmin=0.0d0 tmax=0 potEmin=1.0d10 rgymin=1.0d10 rmsmin=1.0d10 rgymax=0.0d0 rmsmax=0.0d0 do t=0,MaxN htot(t)=0 enddo #ifdef MPI do i=1,scount(me1) #else do i=1,ntot(islice) #endif do j=1,nParmSet if (potE(i,j).le.potEmin) potEmin=potE(i,j) enddo if (q(nQ+1,i).lt.rmsmin) rmsmin=q(nQ+1,i) if (q(nQ+1,i).gt.rmsmax) rmsmax=q(nQ+1,i) if (q(nQ+2,i).lt.rgymin) rgymin=q(nQ+2,i) if (q(nQ+2,i).gt.rgymax) rgymax=q(nQ+2,i) ind_point(i)=0 do j=nQ,1,-1 ind=(q(j,i)-dmin+1.0d-8)/delta if (j.eq.1) then ind_point(i)=ind_point(i)+ind else ind_point(i)=ind_point(i)+nbin1**(j-1)*ind endif c write (iout,*) "i",i," j",j," q",q(j,i)," ind_point", c & ind_point(i) call flush(iout) if (ind_point(i).lt.0 .or. ind_point(i).gt.MaxHdim) then write (iout,*) "Error - index exceeds range for point",i, & " q=",q(j,i)," ind",ind_point(i) #ifdef MPI write (iout,*) "Processor",me1 call flush(iout) call MPI_Abort(MPI_COMM_WORLD, Ierror, Errcode ) #endif stop endif enddo ! j if (ind_point(i).gt.tmax) tmax=ind_point(i) htot(ind_point(i))=htot(ind_point(i))+1 #ifdef DEBUG write (iout,*) "i",i,"q",(q(j,i),j=1,nQ)," ind",ind_point(i), & " htot",htot(ind_point(i)) call flush(iout) #endif enddo ! i call flush(iout) nbin=nbin1**nQ-1 write (iout,'(a)') "Numbers of counts in Q bins" do t=0,tmax if (htot(t).gt.0) then write (iout,'(i15,$)') t liczba=t do j=1,nQ jj = mod(liczba,nbin1) liczba=liczba/nbin1 write (iout,'(i5,$)') jj enddo write (iout,'(i8)') htot(t) endif enddo do iparm=1,nParmSet write (iout,'(a,i3)') "Number of data points for parameter set", & iparm write (iout,'(i7,$)') ((snk(m,ib,iparm,islice),m=1,nr(ib,iparm)), & ib=1,nT_h(iparm)) write (iout,'(i8)') stot(islice) write (iout,'(a)') enddo call flush(iout) #ifdef MPI call MPI_AllReduce(tmax,tmax_t,1,MPI_INTEGER,MPI_MAX, & WHAM_COMM,IERROR) tmax=tmax_t call MPI_AllReduce(potEmin,potEmin_t,1,MPI_DOUBLE_PRECISION, & MPI_MIN,WHAM_COMM,IERROR) call MPI_AllReduce(rmsmin,rmsmin_t,1,MPI_DOUBLE_PRECISION, & MPI_MIN,WHAM_COMM,IERROR) call MPI_AllReduce(rmsmax,rmsmax_t,1,MPI_DOUBLE_PRECISION, & MPI_MAX,WHAM_COMM,IERROR) call MPI_AllReduce(rgymin,rgymin_t,1,MPI_DOUBLE_PRECISION, & MPI_MIN,WHAM_COMM,IERROR) call MPI_AllReduce(rgymax,rgymax_t,1,MPI_DOUBLE_PRECISION, & MPI_MAX,WHAM_COMM,IERROR) potEmin=potEmin_t/2 rgymin=rgymin_t rgymax=rgymax_t rmsmin=rmsmin_t rmsmax=rmsmax_t write (iout,*) "potEmin",potEmin #endif rmsmin=deltrms*dint(rmsmin/deltrms) rmsmax=deltrms*dint(rmsmax/deltrms) rgymin=deltrms*dint(rgymin/deltrgy) rgymax=deltrms*dint(rgymax/deltrgy) nbin_rms=(rmsmax-rmsmin)/deltrms nbin_rgy=(rgymax-rgymin)/deltrgy write (iout,*) "rmsmin",rmsmin," rmsmax",rmsmax," rgymin",rgymin, & " rgymax",rgymax," nbin_rms",nbin_rms," nbin_rgy",nbin_rgy nFi=0 do i=1,nParmSet do j=1,nT_h(i) nFi=nFi+nR(j,i) enddo enddo write (iout,*) "nFi",nFi ! Compute the Boltzmann factor corresponing to restrain potentials in different ! simulations. #ifdef MPI do i=1,scount(me1) #else do i=1,ntot(islice) #endif c write (9,'(3i5,f10.5)') i,(iparm,potE(i,iparm),iparm=1,nParmSet) do iparm=1,nParmSet #ifdef DEBUG write (iout,'(2i5,21f8.2)') i,iparm, & (enetb(k,i,iparm),k=1,21) #endif call restore_parm(iparm) #ifdef DEBUG write (iout,*) wsc,wscp,welec,wvdwpp,wang,wtor,wscloc, & wcorr,wcorr5,wcorr6,wturn4,wturn3,wturn6,wel_loc, & wtor_d,wsccor,wbond #endif do ib=1,nT_h(iparm) if (rescale_mode.eq.1) then quot=1.0d0/(beta_h(ib,iparm)*1.987D-3*T0) quotl=1.0d0 kfacl=1.0d0 do l=1,5 quotl1=quotl quotl=quotl*quot kfacl=kfacl*kfac fT(l)=kfacl/(kfacl-1.0d0+quotl) enddo #if defined(FUNCTH) tt = 1.0d0/(beta_h(ib,iparm)*1.987D-3) ft(6)=(320.0d0+80.0d0*dtanh((tt-320.0d0)/80.0d0))/320.0d0 #elif defined(FUNCT) ft(6)=1.0d0/(beta_h(ib,iparm)*1.987D-3*T0) #else ft(6)=1.0d0 #endif else if (rescale_mode.eq.2) then quot=1.0d0/(T0*beta_h(ib,iparm)*1.987D-3) quotl=1.0d0 do l=1,5 quotl=quotl*quot fT(l)=1.12692801104297249644d0/ & dlog(dexp(quotl)+dexp(-quotl)) enddo #if defined(FUNCTH) tt = 1.0d0/(beta_h(ib,iparm)*1.987D-3) ft(6)=(320.0d0+80.0d0*dtanh((tt-320.0d0)/80.0d0))/320.0d0 #elif defined(FUNCT) ft(6)=1.0d0/(beta_h(ib,iparm)*1.987D-3*T0) #else ft(6)=1.0d0 #endif c write (iout,*) 1.0d0/(beta_h(ib,iparm)*1.987D-3),ft else if (rescale_mode.eq.0) then do l=1,6 fT(l)=1.0d0 enddo else write (iout,*) "Error in WHAM_CALC: wrong RESCALE_MODE", & rescale_mode call flush(iout) return1 endif evdw=enetb(1,i,iparm) evdw_t=enetb(21,i,iparm) #ifdef SCP14 evdw2_14=enetb(17,i,iparm) evdw2=enetb(2,i,iparm)+evdw2_14 #else evdw2=enetb(2,i,iparm) evdw2_14=0.0d0 #endif #ifdef SPLITELE ees=enetb(3,i,iparm) evdw1=enetb(16,i,iparm) #else ees=enetb(3,i,iparm) evdw1=0.0d0 #endif ecorr=enetb(4,i,iparm) ecorr5=enetb(5,i,iparm) ecorr6=enetb(6,i,iparm) eel_loc=enetb(7,i,iparm) eello_turn3=enetb(8,i,iparm) eello_turn4=enetb(9,i,iparm) eturn6=enetb(10,i,iparm) ebe=enetb(11,i,iparm) escloc=enetb(12,i,iparm) etors=enetb(13,i,iparm) etors_d=enetb(14,i,iparm) ehpb=enetb(15,i,iparm) estr=enetb(18,i,iparm) esccor=enetb(19,i,iparm) edihcnstr=enetb(20,i,iparm) #ifdef DEBUG write (iout,'(3i5,6f5.2,14f12.3)') i,ib,iparm,(ft(l),l=1,6), & evdw+evdw_t,evdw2,ees,evdw1,ecorr,eel_loc,estr,ebe,escloc, & etors,etors_d,eello_turn3,eello_turn4,esccor #endif #ifdef SPLITELE etot=wsc*(evdw+ft(6)*evdw_t)+wscp*evdw2+ft(1)*welec*ees & +wvdwpp*evdw1 & +wang*ebe+ft(1)*wtor*etors+wscloc*escloc & +wstrain*ehpb+nss*ebr+ft(3)*wcorr*ecorr+ft(4)*wcorr5*ecorr5 & +ft(5)*wcorr6*ecorr6+ft(3)*wturn4*eello_turn4 & +ft(2)*wturn3*eello_turn3 & +ft(5)*wturn6*eturn6+ft(2)*wel_loc*eel_loc & +edihcnstr+ft(2)*wtor_d*etors_d+ft(1)*wsccor*esccor & +wbond*estr #else etot=wsc*(evdw+ft(6)*evdw_t)+wscp*evdw2 & +ft(1)*welec*(ees+evdw1) & +wang*ebe+ft(1)*wtor*etors+wscloc*escloc & +wstrain*ehpb+nss*ebr+ft(3)*wcorr*ecorr+ft(4)*wcorr5*ecorr5 & +ft(5)*wcorr6*ecorr6+ft(3)*wturn4*eello_turn4 & +ft(2)*wturn3*eello_turn3 & +ft(5)*wturn6*eturn6+ft(2)*wel_loc*eel_loc+edihcnstr & +ft(2)*wtor_d*etors_d+ft(1)*wsccor*esccor & +wbond*estr #endif #ifdef DEBUG write (iout,*) i,iparm,1.0d0/(beta_h(ib,iparm)*1.987D-3), & etot,potEmin #endif #ifdef DEBUG if (iparm.eq.1 .and. ib.eq.1) then write (iout,*)"Conformation",i energia(0)=etot do k=1,max_ene energia(k)=enetb(k,i,iparm) enddo call enerprint(energia(0),fT) endif #endif do kk=1,nR(ib,iparm) Econstr=0.0d0 do j=1,nQ dd = q(j,i) Econstr=Econstr+Kh(j,kk,ib,iparm) & *(dd-q0(j,kk,ib,iparm))**2 enddo v(i,kk,ib,iparm)= & -beta_h(ib,iparm)*(etot-potEmin+Econstr) #ifdef DEBUG write (iout,'(4i5,4e15.5)') i,kk,ib,iparm, & etot,potEmin,etot-potEmin,v(i,kk,ib,iparm) #endif enddo ! kk enddo ! ib enddo ! iparm enddo ! i ! Simple iteration to calculate free energies corresponding to all simulation ! runs. do iter=1,maxit ! Compute new free-energy values corresponding to the righ-hand side of the ! equation and their derivatives. write (iout,*) "------------------------fi" #ifdef MPI do t=1,scount(me1) #else do t=1,ntot(islice) #endif vmax=-1.0d+20 do i=1,nParmSet do k=1,nT_h(i) do l=1,nR(k,i) vf=v(t,l,k,i)+f(l,k,i) if (vf.gt.vmax) vmax=vf enddo enddo enddo denom=0.0d0 do i=1,nParmSet do k=1,nT_h(i) do l=1,nR(k,i) aux=f(l,k,i)+v(t,l,k,i)-vmax if (aux.gt.-200.0d0) & denom=denom+snk(l,k,i,islice)*dexp(aux) enddo enddo enddo entfac(t)=-dlog(denom)-vmax #ifdef DEBUG write (iout,*) t,"vmax",vmax," denom",denom,"entfac",entfac(t) #endif enddo do iparm=1,nParmSet do iib=1,nT_h(iparm) do ii=1,nR(iib,iparm) #ifdef MPI fi_p(ii,iib,iparm)=0.0d0 do t=1,scount(me) fi_p(ii,iib,iparm)=fi_p(ii,iib,iparm) & +dexp(v(t,ii,iib,iparm)+entfac(t)) #ifdef DEBUG write (iout,'(4i5,3e15.5)') t,ii,iib,iparm, & v(t,ii,iib,iparm),entfac(t),fi_p(ii,iib,iparm) #endif enddo #else fi(ii,iib,iparm)=0.0d0 do t=1,ntot(islice) fi(ii,iib,iparm)=fi(ii,iib,iparm) & +dexp(v(t,ii,iib,iparm)+entfac(t)) enddo #endif enddo ! ii enddo ! iib enddo ! iparm #ifdef MPI #ifdef DEBUG write (iout,*) "fi before MPI_Reduce me",me,' master',master do iparm=1,nParmSet do ib=1,nT_h(nparmset) write (iout,*) "iparm",iparm," ib",ib write (iout,*) "beta=",beta_h(ib,iparm) write (iout,'(8e15.5)') (fi_p(i,ib,iparm),i=1,nR(ib,iparm)) enddo enddo #endif write (iout,*) "REDUCE size",maxR,MaxT_h,nParmSet, & maxR*MaxT_h*nParmSet write (iout,*) "MPI_COMM_WORLD",MPI_COMM_WORLD, & " WHAM_COMM",WHAM_COMM call MPI_Reduce(fi_p(1,1,1),fi(1,1,1),maxR*MaxT_h*nParmSet, & MPI_DOUBLE_PRECISION, & MPI_SUM,Master,WHAM_COMM,IERROR) #ifdef DEBUG write (iout,*) "fi after MPI_Reduce nparmset",nparmset do iparm=1,nParmSet write (iout,*) "iparm",iparm do ib=1,nT_h(iparm) write (iout,*) "beta=",beta_h(ib,iparm) write (iout,'(8e15.5)') (fi(i,ib,iparm),i=1,nR(ib,iparm)) enddo enddo #endif if (me1.eq.Master) then #endif avefi=0.0d0 do iparm=1,nParmSet do ib=1,nT_h(iparm) do i=1,nR(ib,iparm) fi(i,ib,iparm)=-dlog(fi(i,ib,iparm)) avefi=avefi+fi(i,ib,iparm) enddo enddo enddo avefi=avefi/nFi do iparm=1,nParmSet write (iout,*) "Parameter set",iparm do ib =1,nT_h(iparm) write (iout,*) "beta=",beta_h(ib,iparm) do i=1,nR(ib,iparm) fi(i,ib,iparm)=fi(i,ib,iparm)-avefi enddo write (iout,'(8f10.5)') (fi(i,ib,iparm),i=1,nR(ib,iparm)) write (iout,'(8f10.5)') (f(i,ib,iparm),i=1,nR(ib,iparm)) enddo enddo ! Compute the norm of free-energy increments. finorm=0.0d0 do iparm=1,nParmSet do ib=1,nT_h(iparm) do i=1,nR(ib,iparm) finorm=finorm+dabs(fi(i,ib,iparm)-f(i,ib,iparm)) f(i,ib,iparm)=fi(i,ib,iparm) enddo enddo enddo write (iout,*) 'Iteration',iter,' finorm',finorm #ifdef MPI endif call MPI_Bcast(f(1,1,1),MaxR*MaxT_h*nParmSet, & MPI_DOUBLE_PRECISION,Master, & WHAM_COMM,IERROR) call MPI_Bcast(finorm,1,MPI_DOUBLE_PRECISION,Master, & WHAM_COMM,IERROR) #endif ! Exit, if the increment norm is smaller than pre-assigned tolerance. if (finorm.lt.fimin) then write (iout,*) 'Iteration converged' goto 20 endif enddo ! iter 20 continue ! Now, put together the histograms from all simulations, in order to get the ! unbiased total histogram. #ifdef MPI do t=0,tmax hfin_ent_p(t)=0.0d0 enddo #else do t=0,tmax hfin_ent(t)=0.0d0 enddo #endif write (iout,*) "--------------hist" #ifdef MPI do iparm=1,nParmSet do i=0,nGridT sumW_p(i,iparm)=0.0d0 sumE_p(i,iparm)=0.0d0 sumEbis_p(i,iparm)=0.0d0 sumEsq_p(i,iparm)=0.0d0 do j=1,nQ+2 sumQ_p(j,i,iparm)=0.0d0 sumQsq_p(j,i,iparm)=0.0d0 sumEQ_p(j,i,iparm)=0.0d0 enddo enddo enddo upindE_p=0 #else do iparm=1,nParmSet do i=0,nGridT sumW(i,iparm)=0.0d0 sumE(i,iparm)=0.0d0 sumEbis(i,iparm)=0.0d0 sumEsq(i,iparm)=0.0d0 do j=1,nQ+2 sumQ(j,i,iparm)=0.0d0 sumQsq(j,i,iparm)=0.0d0 sumEQ(j,i,iparm)=0.0d0 enddo enddo enddo upindE=0 #endif c 8/26/05 entropy distribution #ifdef MPI entmin_p=1.0d10 entmax_p=-1.0d10 do t=1,scount(me1) c ent=-dlog(entfac(t)) ent=entfac(t) if (ent.lt.entmin_p) entmin_p=ent if (ent.gt.entmax_p) entmax_p=ent enddo write (iout,*) "entmin",entmin_p," entmax",entmax_p call flush(iout) call MPI_Allreduce(entmin_p,entmin,1,MPI_DOUBLE_PRECISION,MPI_MIN, & WHAM_COMM,IERROR) call MPI_Allreduce(entmax_p,entmax,1,MPI_DOUBLE_PRECISION,MPI_MAX, & WHAM_COMM,IERROR) ientmax=entmax-entmin if (ientmax.gt.2000) ientmax=2000 write (iout,*) "entmin",entmin," entmax",entmax," ientmax",ientmax call flush(iout) do t=1,scount(me1) c ient=-dlog(entfac(t))-entmin ient=entfac(t)-entmin if (ient.le.2000) histent_p(ient)=histent_p(ient)+1 enddo call MPI_Allreduce(histent_p(0),histent(0),ientmax+1,MPI_INTEGER, & MPI_SUM,WHAM_COMM,IERROR) if (me1.eq.Master) then write (iout,*) "Entropy histogram" do i=0,ientmax write(iout,'(f15.4,i10)') entmin+i,histent(i) enddo endif #else entmin=1.0d10 entmax=-1.0d10 do t=1,ntot(islice) ent=entfac(t) if (ent.lt.entmin) entmin=ent if (ent.gt.entmax) entmax=ent enddo ientmax=-dlog(entmax)-entmin if (ientmax.gt.2000) ientmax=2000 do t=1,ntot(islice) ient=entfac(t)-entmin if (ient.le.2000) histent(ient)=histent(ient)+1 enddo write (iout,*) "Entropy histogram" do i=0,ientmax write(iout,'(2f15.4)') entmin+i,histent(i) enddo #endif #ifdef MPI c write (iout,*) "me1",me1," scount",scount(me1) do iparm=1,nParmSet #ifdef MPI do ib=1,nT_h(iparm) do t=0,tmax hfin_p(t,ib)=0.0d0 enddo enddo do i=1,maxindE histE_p(i)=0.0d0 enddo #else do ib=1,nT_h(iparm) do t=0,tmax hfin(t,ib)=0.0d0 enddo enddo do i=1,maxindE histE(i)=0.0d0 enddo #endif do ib=1,nT_h(iparm) do i=0,MaxBinRms do j=0,MaxBinRgy hrmsrgy(j,i,ib)=0.0d0 #ifdef MPI hrmsrgy_p(j,i,ib)=0.0d0 #endif enddo enddo enddo do t=1,scount(me1) #else do t=1,ntot(islice) #endif ind = ind_point(t) #ifdef MPI hfin_ent_p(ind)=hfin_ent_p(ind)+dexp(entfac(t)) #else hfin_ent(ind)=hfin_ent(ind)+dexp(entfac(t)) #endif c write (iout,'(2i5,20f8.2)') t,t,(enetb(k,t,iparm),k=1,18) call restore_parm(iparm) evdw=enetb(21,t,iparm) evdw_t=enetb(1,t,iparm) #ifdef SCP14 evdw2_14=enetb(17,t,iparm) evdw2=enetb(2,t,iparm)+evdw2_14 #else evdw2=enetb(2,t,iparm) evdw2_14=0.0d0 #endif #ifdef SPLITELE ees=enetb(3,t,iparm) evdw1=enetb(16,t,iparm) #else ees=enetb(3,t,iparm) evdw1=0.0d0 #endif ecorr=enetb(4,t,iparm) ecorr5=enetb(5,t,iparm) ecorr6=enetb(6,t,iparm) eel_loc=enetb(7,t,iparm) eello_turn3=enetb(8,t,iparm) eello_turn4=enetb(9,t,iparm) eturn6=enetb(10,t,iparm) ebe=enetb(11,t,iparm) escloc=enetb(12,t,iparm) etors=enetb(13,t,iparm) etors_d=enetb(14,t,iparm) ehpb=enetb(15,t,iparm) estr=enetb(18,t,iparm) esccor=enetb(19,t,iparm) edihcnstr=enetb(20,t,iparm) edihcnstr=0.0d0 do k=0,nGridT betaT=startGridT+k*delta_T temper=betaT c fT=T0/betaT c ft=2*T0/(T0+betaT) if (rescale_mode.eq.1) then quot=betaT/T0 quotl=1.0d0 kfacl=1.0d0 do l=1,5 quotl1=quotl quotl=quotl*quot kfacl=kfacl*kfac denom=kfacl-1.0d0+quotl fT(l)=kfacl/denom ftprim(l)=-l*ft(l)*quotl1/(T0*denom) ftbis(l)=l*kfacl*quotl1* & (2*l*quotl-(l-1)*denom)/(quot*t0*t0*denom**3) enddo #if defined(FUNCTH) ft(6)=(320.0d0+80.0d0*dtanh((betaT-320.0d0)/80.0d0))/ & 320.0d0 ftprim(6)=1.0d0/(320.0d0*dcosh((betaT-320.0d0)/80.0d0)**2) ftbis(6)=-2.0d0*dtanh((betaT-320.0d0)/80.0d0) & /(320.0d0*80.0d0*dcosh((betaT-320.0d0)/80.0d0)**3) #elif defined(FUNCT) fT(6)=betaT/T0 ftprim(6)=1.0d0/T0 ftbis(6)=0.0d0 #else fT(6)=1.0d0 ftprim(6)=0.0d0 ftbis(6)=0.0d0 #endif else if (rescale_mode.eq.2) then quot=betaT/T0 quotl=1.0d0 do l=1,5 quotl1=quotl quotl=quotl*quot eplus=dexp(quotl) eminus=dexp(-quotl) logfac=1.0d0/dlog(eplus+eminus) tanhT=(eplus-eminus)/(eplus+eminus) fT(l)=1.12692801104297249644d0*logfac ftprim(l)=-l*quotl1*ft(l)*tanhT*logfac/T0 ftbis(l)=(l-1)*ftprim(l)/(quot*T0)- & 2*l*quotl1/T0*logfac* & (2*l*quotl1*ft(l)/(T0*(eplus+eminus)**2) & +ftprim(l)*tanhT) enddo #if defined(FUNCTH) ft(6)=(320.0d0+80.0d0*dtanh((betaT-320.0d0)/80.0d0))/ & 320.0d0 ftprim(6)=1.0d0/(320.0d0*dcosh((betaT-320.0d0)/80.0d0)**2) ftbis(6)=-2.0d0*dtanh((betaT-320.0d0)/80.0d0) & /(320.0d0*80.0d0*dcosh((betaT-320.0d0)/80.0d0)**3) #elif defined(FUNCT) fT(6)=betaT/T0 ftprim(6)=1.0d0/T0 ftbis(6)=0.0d0 #else fT(6)=1.0d0 ftprim(6)=0.0d0 ftbis(6)=0.0d0 #endif else if (rescale_mode.eq.0) then do l=1,5 fT(l)=1.0d0 ftprim(l)=0.0d0 enddo else write (iout,*) "Error in WHAM_CALC: wrong RESCALE_MODE", & rescale_mode call flush(iout) return1 endif c write (iout,*) "ftprim",ftprim c write (iout,*) "ftbis",ftbis betaT=1.0d0/(1.987D-3*betaT) #ifdef SPLITELE etot=wsc*(evdw+ft(6)*evdw_t)+wscp*evdw2+ft(1)*welec*ees & +wvdwpp*evdw1 & +wang*ebe+ft(1)*wtor*etors+wscloc*escloc & +wstrain*ehpb+nss*ebr+ft(3)*wcorr*ecorr+ft(4)*wcorr5*ecorr5 & +ft(5)*wcorr6*ecorr6+ft(3)*wturn4*eello_turn4 & +ft(2)*wturn3*eello_turn3 & +ft(5)*wturn6*eturn6+ft(2)*wel_loc*eel_loc & +edihcnstr+ft(2)*wtor_d*etors_d+ft(1)*wsccor*esccor & +wbond*estr eprim=ftprim(6)*evdw_t+ftprim(1)*welec*ees & +ftprim(1)*wtor*etors+ & ftprim(3)*wcorr*ecorr+ftprim(4)*wcorr5*ecorr5+ & ftprim(5)*wcorr6*ecorr6+ftprim(3)*wturn4*eello_turn4+ & ftprim(2)*wturn3*eello_turn3+ftprim(5)*wturn6*eturn6+ & ftprim(2)*wel_loc*eel_loc+ftprim(2)*wtor_d*etors_d+ & ftprim(1)*wsccor*esccor ebis=ftbis(1)*welec*ees+ftbis(1)*wtor*etors+ & ftbis(3)*wcorr*ecorr+ftbis(4)*wcorr5*ecorr5+ & ftbis(5)*wcorr6*ecorr6+ftbis(3)*wturn4*eello_turn4+ & ftbis(2)*wturn3*eello_turn3+ftbis(5)*wturn6*eturn6+ & ftbis(2)*wel_loc*eel_loc+ftbis(2)*wtor_d*etors_d+ & ftbis(1)*wsccor*esccor #else etot=wsc*(evdw+ft(6)*evdw_t)+wscp*evdw2 & +ft(1)*welec*(ees+evdw1) & +wang*ebe+ft(1)*wtor*etors+wscloc*escloc & +wstrain*ehpb+nss*ebr+ft(3)*wcorr*ecorr+ft(4)*wcorr5*ecorr5 & +ft(5)*wcorr6*ecorr6+ft(3)*wturn4*eello_turn4 & +ft(2)*wturn3*eello_turn3 & +ft(5)*wturn6*eturn6+ft(2)*wel_loc*eel_loc+edihcnstr & +ft(2)*wtor_d*etors_d+ft(1)*wsccor*esccor & +wbond*estr eprim=ftprim(6)*evdw_t+ftprim(1)*welec*(ees+evdw1) & +ftprim(1)*wtor*etors+ & ftprim(3)*wcorr*ecorr+ftprim(4)*wcorr5*ecorr5+ & ftprim(5)*wcorr6*ecorr6+ftprim(3)*wturn4*eello_turn4+ & ftprim(2)*wturn3*eello_turn3+ftprim(5)*wturn6*eturn6+ & ftprim(2)*wel_loc*eel_loc+ftprim(2)*wtor_d*etors_d+ & ftprim(1)*wsccor*esccor ebis=ftbis(1)*welec*(ees+evdw1)+ftbis(1)*wtor*etors+ & ftbis(3)*wcorr*ecorr+ftbis(4)*wcorr5*ecorr5+ & ftbis(5)*wcorr6*ecorr6+ftbis(3)*wturn4*eello_turn4+ & ftbis(2)*wturn3*eello_turn3+ftbis(5)*wturn6*eturn6+ & ftbis(2)*wel_loc*eel_loc+ftbis(2)*wtor_d*etors_d+ & ftprim(1)*wsccor*esccor #endif weight=dexp(-betaT*(etot-potEmin)+entfac(t)) #ifdef DEBUG write (iout,*) "iparm",iparm," t",t," betaT",betaT, & " etot",etot," entfac",entfac(t), & " weight",weight," ebis",ebis #endif etot=etot-temper*eprim #ifdef MPI sumW_p(k,iparm)=sumW_p(k,iparm)+weight sumE_p(k,iparm)=sumE_p(k,iparm)+etot*weight sumEbis_p(k,iparm)=sumEbis_p(k,iparm)+ebis*weight sumEsq_p(k,iparm)=sumEsq_p(k,iparm)+etot**2*weight do j=1,nQ+2 sumQ_p(j,k,iparm)=sumQ_p(j,k,iparm)+q(j,t)*weight sumQsq_p(j,k,iparm)=sumQsq_p(j,k,iparm)+q(j,t)**2*weight sumEQ_p(j,k,iparm)=sumEQ_p(j,k,iparm) & +etot*q(j,t)*weight enddo #else sumW(k,iparm)=sumW(k,iparm)+weight sumE(k,iparm)=sumE(k,iparm)+etot*weight sumEbis(k,iparm)=sumEbis(k,iparm)+ebis*weight sumEsq(k,iparm)=sumEsq(k,iparm)+etot**2*weight do j=1,nQ+2 sumQ(j,k,iparm)=sumQ(j,k,iparm)+q(j,t)*weight sumQsq(j,k,iparm)=sumQsq(j,k,iparm)+q(j,t)**2*weight sumEQ(j,k,iparm)=sumEQ(j,k,iparm) & +etot*q(j,t)*weight enddo #endif enddo indE = aint(potE(t,iparm)-aint(potEmin)) if (indE.ge.0 .and. indE.le.maxinde) then if (indE.gt.upindE_p) upindE_p=indE histE_p(indE)=histE_p(indE)+dexp(-entfac(t)) endif #ifdef MPI do ib=1,nT_h(iparm) expfac=dexp(-beta_h(ib,iparm)*(etot-potEmin)+entfac(t)) hfin_p(ind,ib)=hfin_p(ind,ib)+ & dexp(-beta_h(ib,iparm)*(etot-potEmin)+entfac(t)) if (rmsrgymap) then indrgy=dint((q(nQ+2,t)-rgymin)/deltrgy) indrms=dint((q(nQ+1,t)-rmsmin)/deltrms) hrmsrgy_p(indrgy,indrms,ib)= & hrmsrgy_p(indrgy,indrms,ib)+expfac endif enddo #else do ib=1,nT_h(iparm) expfac=dexp(-beta_h(ib,iparm)*(etot-potEmin)+entfac(t)) hfin(ind,ib)=hfin(ind,ib)+ & dexp(-beta_h(ib,iparm)*(etot-potEmin)+entfac(t)) if (rmsrgymap) then indrgy=dint((q(nQ+2,t)-rgymin)/deltrgy) indrms=dint((q(nQ+1,t)-rmsmin)/deltrms) hrmsrgy(indrgy,indrms,ib)= & hrmsrgy(indrgy,indrms,ib)+expfac endif enddo #endif enddo ! t do ib=1,nT_h(iparm) if (histout) call MPI_Reduce(hfin_p(0,ib),hfin(0,ib),nbin, & MPI_DOUBLE_PRECISION,MPI_SUM,Master,WHAM_COMM,IERROR) if (rmsrgymap) then call MPI_Reduce(hrmsrgy_p(0,0,ib),hrmsrgy(0,0,ib), & (MaxBinRgy+1)*(nbin_rms+1),MPI_DOUBLE_PRECISION,MPI_SUM,Master, & WHAM_COMM,IERROR) endif enddo call MPI_Reduce(upindE_p,upindE,1, & MPI_INTEGER,MPI_MAX,Master,WHAM_COMM,IERROR) call MPI_Reduce(histE_p(0),histE(0),maxindE, & MPI_DOUBLE_PRECISION,MPI_SUM,Master,WHAM_COMM,IERROR) if (me1.eq.master) then if (histout) then write (iout,'(6x,$)') write (iout,'(f20.2,$)') (1.0d0/(1.987D-3*beta_h(ib,iparm)), & ib=1,nT_h(iparm)) write (iout,*) write (iout,'(/a)') 'Final histograms' if (histfile) then if (nslice.eq.1) then if (separate_parset) then write(licz3,"(bz,i3.3)") myparm histname=prefix(:ilen(prefix))//'_par'//licz3//'.hist' else histname=prefix(:ilen(prefix))//'.hist' endif else if (separate_parset) then write(licz3,"(bz,i3.3)") myparm histname=prefix(:ilen(prefix))//'_par'//licz3// & '_slice_'//licz2//'.hist' else histname=prefix(:ilen(prefix))//'_slice_'//licz2//'.hist' endif endif #if defined(AIX) || defined(PGI) open (ihist,file=histname,position='append') #else open (ihist,file=histname,access='append') #endif endif do t=0,tmax liczba=t sumH=0.0d0 do ib=1,nT_h(iparm) sumH=sumH+hfin(t,ib) enddo if (sumH.gt.0.0d0) then do j=1,nQ jj = mod(liczba,nbin1) liczba=liczba/nbin1 write (iout,'(f6.3,$)') dmin+(jj+0.5d0)*delta if (histfile) & write (ihist,'(f6.3,$)') dmin+(jj+0.5d0)*delta enddo do ib=1,nT_h(iparm) write (iout,'(e20.10,$)') hfin(t,ib) if (histfile) write (ihist,'(e20.10,$)') hfin(t,ib) enddo write (iout,'(i5)') iparm if (histfile) write (ihist,'(i5)') iparm endif enddo endif if (entfile) then if (nslice.eq.1) then if (separate_parset) then write(licz3,"(bz,i3.3)") myparm histname=prefix(:ilen(prefix))//"_par"//licz3//'.ent' else histname=prefix(:ilen(prefix))//'.ent' endif else if (separate_parset) then write(licz3,"(bz,i3.3)") myparm histname=prefix(:ilen(prefix))//'par_'//licz3// & '_slice_'//licz2//'.ent' else histname=prefix(:ilen(prefix))//'_slice_'//licz2//'.ent' endif endif #if defined(AIX) || defined(PGI) open (ihist,file=histname,position='append') #else open (ihist,file=histname,access='append') #endif write (ihist,'(a)') "# Microcanonical entropy" do i=0,upindE write (ihist,'(f8.0,$)') dint(potEmin)+i if (histE(i).gt.0.0e0) then write (ihist,'(f15.5,$)') dlog(histE(i)) else write (ihist,'(f15.5,$)') 0.0d0 endif enddo write (ihist,*) close(ihist) endif write (iout,*) "Microcanonical entropy" do i=0,upindE write (iout,'(f8.0,$)') dint(potEmin)+i if (histE(i).gt.0.0e0) then write (iout,'(f15.5,$)') dlog(histE(i)) else write (iout,'(f15.5,$)') 0.0d0 endif write (iout,*) enddo if (rmsrgymap) then if (nslice.eq.1) then if (separate_parset) then write(licz3,"(bz,i3.3)") myparm histname=prefix(:ilen(prefix))//'_par'//licz3//'.rmsrgy' else histname=prefix(:ilen(prefix))//'.rmsrgy' endif else if (separate_parset) then write(licz3,"(bz,i3.3)") myparm histname=prefix(:ilen(prefix))//'_par'//licz3// & '_slice_'//licz2//'.rmsrgy' else histname=prefix(:ilen(prefix))//'_slice_'//licz2//'.rmsrgy' endif endif #if defined(AIX) || defined(PGI) open (ihist,file=histname,position='append') #else open (ihist,file=histname,access='append') #endif do i=0,nbin_rms do j=0,nbin_rgy write(ihist,'(2f8.2,$)') & rgymin+deltrgy*j,rmsmin+deltrms*i do ib=1,nT_h(iparm) if (hrmsrgy(j,i,ib).gt.0.0d0) then write(ihist,'(e14.5,$)') & -dlog(hrmsrgy(j,i,ib))/beta_h(ib,iparm) & +potEmin else write(ihist,'(e14.5,$)') 1.0d6 endif enddo write (ihist,'(i2)') iparm enddo enddo close(ihist) endif endif enddo ! iparm #ifdef MPI call MPI_Reduce(hfin_ent_p(0),hfin_ent(0),nbin, & MPI_DOUBLE_PRECISION,MPI_SUM,Master,WHAM_COMM,IERROR) call MPI_Reduce(sumW_p(0,1),sumW(0,1),(nGridT+1)*nParmSet, & MPI_DOUBLE_PRECISION,MPI_SUM,Master,WHAM_COMM,IERROR) call MPI_Reduce(sumE_p(0,1),sumE(0,1),(nGridT+1)*nParmSet, & MPI_DOUBLE_PRECISION,MPI_SUM,Master,WHAM_COMM,IERROR) call MPI_Reduce(sumEbis_p(0,1),sumEbis(0,1),(nGridT+1)*nParmSet, & MPI_DOUBLE_PRECISION,MPI_SUM,Master,WHAM_COMM,IERROR) call MPI_Reduce(sumEsq_p(0,1),sumEsq(0,1),(nGridT+1)*nParmSet, & MPI_DOUBLE_PRECISION,MPI_SUM,Master,WHAM_COMM,IERROR) call MPI_Reduce(sumQ_p(1,0,1),sumQ(1,0,1), & MaxQ1*(nGridT+1)*nParmSet,MPI_DOUBLE_PRECISION,MPI_SUM,Master, & WHAM_COMM,IERROR) call MPI_Reduce(sumQsq_p(1,0,1),sumQsq(1,0,1), & MaxQ1*(nGridT+1)*nParmSet,MPI_DOUBLE_PRECISION,MPI_SUM,Master, & WHAM_COMM,IERROR) call MPI_Reduce(sumEQ_p(1,0,1),sumEQ(1,0,1), & MaxQ1*(nGridT+1)*nParmSet,MPI_DOUBLE_PRECISION,MPI_SUM,Master, & WHAM_COMM,IERROR) if (me.eq.master) then #endif write (iout,'(/a)') 'Thermal characteristics of folding' if (nslice.eq.1) then nazwa=prefix else nazwa=prefix(:ilen(prefix))//"_slice_"//licz2 endif iln=ilen(nazwa) if (nparmset.eq.1 .and. .not.separate_parset) then nazwa=nazwa(:iln)//".thermal" else if (nparmset.eq.1 .and. separate_parset) then write(licz3,"(bz,i3.3)") myparm nazwa=nazwa(:iln)//"_par_"//licz3//".thermal" endif do iparm=1,nParmSet if (nparmset.gt.1) then write(licz3,"(bz,i3.3)") iparm nazwa=nazwa(:iln)//"_par_"//licz3//".thermal" endif open(34,file=nazwa) if (separate_parset) then write (iout,'(a,i3)') "Parameter set",myparm else write (iout,'(a,i3)') "Parameter set",iparm endif do i=0,NGridT sumE(i,iparm)=sumE(i,iparm)/sumW(i,iparm) sumEbis(i,iparm)=(startGridT+i*delta_T)*sumEbis(i,iparm)/ & sumW(i,iparm) sumEsq(i,iparm)=(sumEsq(i,iparm)/sumW(i,iparm) & -sumE(i,iparm)**2)/(1.987D-3*(startGridT+i*delta_T)**2) do j=1,nQ+2 sumQ(j,i,iparm)=sumQ(j,i,iparm)/sumW(i,iparm) sumQsq(j,i,iparm)=sumQsq(j,i,iparm)/sumW(i,iparm) & -sumQ(j,i,iparm)**2 sumEQ(j,i,iparm)=sumEQ(j,i,iparm)/sumW(i,iparm) & -sumQ(j,i,iparm)*sumE(i,iparm) enddo sumW(i,iparm)=-dlog(sumW(i,iparm))*(1.987D-3* & (startGridT+i*delta_T))+potEmin write (iout,'(f7.1,2f15.5,$)') startGridT+i*delta_T, & sumW(i,iparm),sumE(i,iparm) write (iout,'(f10.5,$)') (sumQ(j,i,iparm),j=1,nQ+2) write (iout,'(e15.5,$)') sumEsq(i,iparm)-sumEbis(i,iparm), & (sumQsq(j,i,iparm),j=1,nQ+2),(sumEQ(j,i,iparm),j=1,nQ+2) write (iout,*) write (34,'(f7.1,2f15.5,$)') startGridT+i*delta_T, & sumW(i,iparm),sumE(i,iparm) write (34,'(f10.5,$)') (sumQ(j,i,iparm),j=1,nQ+2) write (34,'(e15.5,$)') sumEsq(i,iparm)-sumEbis(i,iparm), & (sumQsq(j,i,iparm),j=1,nQ+2),(sumEQ(j,i,iparm),j=1,nQ+2) write (34,*) call flush(34) enddo close(34) enddo if (histout) then do t=0,tmax if (hfin_ent(t).gt.0.0d0) then liczba=t jj = mod(liczba,nbin1) write (iout,'(f6.3,e20.10," ent")') dmin+(jj+0.5d0)*delta, & hfin_ent(t) if (histfile) write (ihist,'(f6.3,e20.10," ent")') & dmin+(jj+0.5d0)*delta, & hfin_ent(t) endif enddo if (histfile) close(ihist) endif #ifdef ZSCORE ! Write data for zscore if (nslice.eq.1) then zscname=prefix(:ilen(prefix))//".zsc" else zscname=prefix(:ilen(prefix))//"_slice_"//licz2//".zsc" endif #if defined(AIX) || defined(PGI) open (izsc,file=prefix(:ilen(prefix))//'.zsc',position='append') #else open (izsc,file=prefix(:ilen(prefix))//'.zsc',access='append') #endif write (izsc,'("NQ=",i1," NPARM=",i1)') nQ,nParmSet do iparm=1,nParmSet write (izsc,'("NT=",i1)') nT_h(iparm) do ib=1,nT_h(iparm) write (izsc,'("TEMP=",f6.1," NR=",i2," SNK=",$)') & 1.0d0/(beta_h(ib,iparm)*1.987D-3),nR(ib,iparm) jj = min0(nR(ib,iparm),7) write (izsc,'(i8,$)') (snk(i,ib,iparm,islice),i=1,jj) write (izsc,'(a1,$)') (" ",i=22+8*jj+1,79) write (izsc,'("&")') if (nR(ib,iparm).gt.7) then do ii=8,nR(ib,iparm),9 jj = min0(nR(ib,iparm),ii+8) write (izsc,'(i8,$)') (snk(i,ib,iparm,islice),i=ii,jj) write (izsc,'(a1,$') (" ",i=(jj-ii+1)*8+1,79) write (izsc,'("&")') enddo endif write (izsc,'("FI=",$)') jj=min0(nR(ib,iparm),7) write (izsc,'(f10.5,$)') (fi(i,ib,iparm),i=1,jj) write (izsc,'(a1,$)') (" ",i=3+10*jj+1,79) write (izsc,'("&")') if (nR(ib,iparm).gt.7) then do ii=8,nR(ib,iparm),9 jj = min0(nR(ib,iparm),ii+8) write (izsc,'(f10.5,$)') (fi(i,ib,iparm),i=ii,jj) if (jj.eq.nR(ib,iparm)) then write (izsc,*) else write (izsc,'(a1,$)') (" ",i=10*(jj-ii+1)+1,79) write (izsc,'(t80,"&")') endif enddo endif do i=1,nR(ib,iparm) write (izsc,'("KH=",$)') write (izsc,'(f7.2,$)') (Kh(j,i,ib,iparm),j=1,nQ) write (izsc,'(" Q0=",$)') write (izsc,'(f7.5,$)') (q0(j,i,ib,iparm),j=1,nQ) write (izsc,*) enddo enddo enddo close(izsc) #endif #ifdef MPI endif #endif return end