static A jttayamp(J jt,A w,B nf,A x,A h){A y;B ng=!nf;I j,n;V*v=VAV(h);
ASSERT(AR(x)<=(nf?v->lr:v->rr),EVRANK);
switch(v->id){
case CPLUS: R tpoly(over(x,one));
case CMINUS: R tpoly(nf?over(x,num[-1]):over(negate(x),one));
case CSTAR: R tpoly(over(zero,x));
case CDIV: ASSERT(ng,EVDOMAIN); R tpoly(over(zero,recip(x)));
case CJDOT: R tpoly(nf?over(x,a0j1):over(jdot1(x),one));
case CPOLY: ASSERT(nf,EVDOMAIN); R tpoly(BOX&AT(x)?poly1(x):x);
case CHGEOM: ASSERT(nf,EVDOMAIN); RE(j=i0(x)); ASSERT(0<=j,EVDOMAIN);
y=IX(j);
R tpoly(divide(hgcoeff(y,h),fact(y)));
case CBANG: ASSERT(nf,EVDOMAIN); RE(j=i0(x)); ASSERT(0<=j,EVDOMAIN);
R tpoly(divide(poly1(box(iota(x))),fact(x)));
case CEXP: if(nf)R eva(x,"(^.x)&^ % !");
RE(n=i0(x));
R 0<=n?tpoly(over(reshape(x,zero),one)):atop(ds(CDIV),amp(h,sc(-n)));
case CFIT: ASSERT(nf&&CPOLY==ID(v->f),EVDOMAIN);
y=over(x,IX(IC(x)));
R tpoly(mdiv(df2(x,y,h),atab(CEXP,y,IX(IC(x)))));
case CCIRCLE:
switch(i0(x)){
case 1: R eval("{&0 1 0 _1@(4&|) % !");
case -3: R eval("{&0 1 0 _1@(4&|) % ]");
case 2: R eval("{&1 0 _1 0@(4&|) % !");
case 5: R eval("2&| % !");
case -7: R eval("2&| % ]");
case 6: R eval("2&|@>: % !");
case -1: R eval("(2&| % ]) * ([: */ (1&+ % 2&+)@(i.@<.&.-:))\"0");
case -5: R eval("({&0 1 0 _1@(4&|) % ]) * ([: */ (1&+ % 2&+)@(i.@<.&.-:))\"0");
}}
ASSERT(0,EVDOMAIN);
}
void ResetQ(void* a) /* transforma coada in coada vida */
{ ACel p = IC(a), aux; /* p - adresa primei celule */
while(p)
{ aux = p; p = p->urm;
free(aux);
}
IC(a) = SC(a) = NULL;
}
int ExtrQ(void *ac, void *ae) /* extrage primul element din coada la adresa ae */
{ ACel p = IC(ac); /* p - adresa primei celule */
if(!p) return 0; /* coada vida -> "esec" */
memcpy(ae, &(p->info), DIME(ac)); /* preia informatia */
IC(ac) = p->urm; /* prima celula este scoasa din coada */
free(p); /* si distrusa */
if(!IC(ac)) SC(ac) = NULL; /* prin extragere coada a devenit vida */
return 1; /* operatie reusita -> "succes" */
}
static F1(jtfacit){A c;V*u,*v;
RZ(c=coeff(w));
if(AN(c))R tpoly(tymes(c,fact(AT(c)&XNUM+RAT?xco1(IX(IC(c))):IX(IC(c)))));
v=VAV(w);
if(CFORK==v->id)switch(ID(v->g)){
case CDIV:
if(CBANG==ID(v->h))R v->f;
break;
case CSTAR:
if(CFORK==ID(v->h)&&(u=VAV(v->h),CDIV==ID(u->g)&&CBANG==ID(u->h)))R folk(v->f,v->g,u->f);
RZ(c=atop(ds(CDIV),ds(CBANG)));
if(equ(c,v->f))R v->h;
if(equ(c,v->h))R v->f;
}
R folk(ds(CBANG),ds(CSTAR),w);
}
void Stwart(vector<long>&IA,vector<long>&MA,
vector<long>&JROW,vector<long>&JCOL)
{
static long M = JROW.size();
vector<long> IR(M),IC(M),R(M),C(M),IP(M),JP(M),IW(M);
static long KERNS, MEND, LG, IER, L = MA[M];
// printf("M = %ld\n",M);
// forVector(MA,i) printf("MA[%ld] = %ld\n",i,MA[i]);
// forVector(IA,i) printf("IA[%ld] = %ld\n",i,IA[i]);
// printf("L=%ld\n",L);
// forVector(R,i) printf("R[%ld] = %ld\n",i,R[i]);
// forVector(C,i) printf("C[%ld] = %ld\n",i,C[i]);
// forVector(IR,i) printf("IR[%ld] = %ld\n",i,IR[i]);
// forVector(IC,i) printf("IC[%ld] = %ld\n",i,IC[i]);
// forVector(JROW,i) printf("JROW[%ld] = %ld\n",i,JROW[i]);
// forVector(JCOL,i) printf("JCOL[%ld] = %ld\n",i,JCOL[i]);
// forVector(IP,i) printf("IP[%ld] = %ld\n",i,IP[i]);
// forVector(JP,i) printf("JP[%ld] = %ld\n",i,JP[i]);
// forVector(IW,i) printf("IW[%ld] = %ld\n",i,IW[i]);
printf("KERNS = %ld, MEND = %ld, LG = %ld, IER = %ld\n",
KERNS, MEND, LG, IER);
stwart_(&IA[0],&L,&MA[0],&M,&R[0],&C[0],&IR[0],&IC[0],
&JROW[0],&JCOL[0],&IP[0],&JP[0],&KERNS,&MEND,&IW[0],&LG,&IER);
}
/* output solution status for PPP --------------------------------------------*/
extern void pppoutsolstat(rtk_t *rtk, int level, FILE *fp)
{
ssat_t *ssat;
double tow,pos[3],vel[3],acc[3];
int i,j,week,nfreq=1;
char id[32];
if (level<=0||!fp) return;
trace(3,"pppoutsolstat:\n");
tow=time2gpst(rtk->sol.time,&week);
/* receiver position */
fprintf(fp,"$POS,%d,%.3f,%d,%.4f,%.4f,%.4f,%.4f,%.4f,%.4f\n",week,tow,
rtk->sol.stat,rtk->x[0],rtk->x[1],rtk->x[2],0.0,0.0,0.0);
/* receiver velocity and acceleration */
if (rtk->opt.dynamics) {
ecef2pos(rtk->sol.rr,pos);
ecef2enu(pos,rtk->x+3,vel);
ecef2enu(pos,rtk->x+6,acc);
fprintf(fp,"$VELACC,%d,%.3f,%d,%.4f,%.4f,%.4f,%.5f,%.5f,%.5f,%.4f,%.4f,%.4f,%.5f,%.5f,%.5f\n",
week,tow,rtk->sol.stat,vel[0],vel[1],vel[2],acc[0],acc[1],acc[2],
0.0,0.0,0.0,0.0,0.0,0.0);
}
/* receiver clocks */
i=IC(0,&rtk->opt);
fprintf(fp,"$CLK,%d,%.3f,%d,%d,%.3f,%.3f,%.3f,%.3f\n",
week,tow,rtk->sol.stat,1,rtk->x[i]*1E9/CLIGHT,rtk->x[i+1]*1E9/CLIGHT,
0.0,0.0);
/* tropospheric parameters */
if (rtk->opt.tropopt==TROPOPT_EST||rtk->opt.tropopt==TROPOPT_ESTG) {
i=IT(&rtk->opt);
fprintf(fp,"$TROP,%d,%.3f,%d,%d,%.4f,%.4f\n",week,tow,rtk->sol.stat,
1,rtk->x[i],0.0);
}
if (rtk->opt.tropopt==TROPOPT_ESTG) {
i=IT(&rtk->opt);
fprintf(fp,"$TRPG,%d,%.3f,%d,%d,%.5f,%.5f,%.5f,%.5f\n",week,tow,
rtk->sol.stat,1,rtk->x[i+1],rtk->x[i+2],0.0,0.0);
}
if (rtk->sol.stat==SOLQ_NONE||level<=1) return;
/* residuals and status */
for (i=0;i<MAXSAT;i++) {
ssat=rtk->ssat+i;
if (!ssat->vs) continue;
satno2id(i+1,id);
for (j=0;j<nfreq;j++) {
fprintf(fp,"$SAT,%d,%.3f,%s,%d,%.1f,%.1f,%.4f,%.4f,%d,%.0f,%d,%d,%d,%d,%d,%d\n",
week,tow,id,j+1,ssat->azel[0]*R2D,ssat->azel[1]*R2D,
ssat->resp[j],ssat->resc[j],ssat->vsat[j],ssat->snr[j]*0.25,
ssat->fix[j],ssat->slip[j]&3,ssat->lock[j],ssat->outc[j],
ssat->slipc[j],ssat->rejc[j]);
}
}
}
size_t PrelQ(void* a, TF1 f) /* prelucreaza elementele, folosind functia f;
rezultat = numarul de elemente pentru care f != 0 */
{ size_t n = 0;
ACel p = IC(a); /* p - adresa primei celule */
for(; p; p = p->urm)
if(f(&(p->info))) n++;
return n;
}
static DF1(jtgsuffix){A h,*hv,z,*zv;I m,n,r;
RZ(w);
if(jt->rank&&jt->rank[1]<AR(w)){r=jt->rank[1]; jt->rank=0; R rank1ex(w,self,jt->rank[1],jtgsuffix);}
jt->rank=0;
n=IC(w);
h=VAV(self)->h; hv=AAV(h); m=AN(h);
GATV(z,BOX,n,1,0); zv=AAV(z);
DO(n, RZ(zv[i]=df1(drop(sc(i),w),hv[i%m])););
void Dummy (EifList<int> &TestList) {
int ex = TestList.ItCnt();
EifIterator it1 (TestList), it2, it3;
ex++;
IC (ex, TestList);
it2 = it3;
IC (ex, TestList);
it2 = it1;
ex++;
IC (ex, TestList);
}
void DummyIt(int Exp, EifIterator it, EifList<int> &TestList) {
Exp++;
IC (Exp, TestList);
EifList <char *> StrList (True);
StrList.Add ("Trevor");
}
static DF1(insert){PROLOG;A hs,*hv,z;I hn,j,k,m,n;
RZ(w);
m=IC(w); hs=VAV(self)->h; hn=AN(hs); hv=AAV(hs);
if(!m)R df1(w,iden(*hv));
j=n=MAX(hn,m-1);
RZ(z=AR(w)?from(sc(n%m),w):ca(w));
if(1==n)R z;
DO(n, --j; k=j%hn; RZ(z=(VAV(hv[k])->f2)(from(sc(j%m),w),z,hv[k])));
EPILOG(z);
}
double R(double x){ //Mills Ratio
if (x > 4){
return 1.0 / x;
}
double N = IC(x);
double D = IN(x);
if (D < pow(10,-15)){ //machine epsilon
return 1.0 / pow(10,-15);
}
return exp(log(1. - N)-log(D));
}
static DF1(reduce){PROLOG;DECLF;A y,z;C*u,*v;I c,k,m,old,t;
RZ(w);
m=IC(w);
if(!m)R df1(w,iden(fs));
RZ(z=tail(w));
if(1==m)R z;
t=AT(w); c=AN(z);
GA(y,t,c,AR(z),AS(z)); u=CAV(y); k=SZT(t,c); v=CAV(w)+k*(m-1);
old=tbase+ttop;
DO(m-1, MC(u,v-=k,k); RZ(z=f2(y,z,fs)); gc(z,old));
EPILOG(z);
}
int IntrQ(void *ac, void *ae) /* adauga element la sfarsitul cozii */
{ ACel aux = (ACel)malloc(DIMC(ac)); /* aloca o noua celula */
if(!aux) return 0; /* alocare imposibila -> "esec" */
memcpy(&(aux->info), ae, DIME(ac)); /* completeaza celula */
aux->urm = NULL;
if(SC(ac) == NULL) /* coada vida */
IC(ac) = aux; /* -> noua celula se afla la inceputul cozii */
else /* coada nevida */
SC(ac)->urm = aux; /* -> leaga celula dupa ultima din coada */
SC(ac) = aux; /* actualizeaza sfarsitul cozii */
return 1; /* operatie reusita -> "succes" */
}
static A jtmerge1(J jt,A w,A ind){A z;B*b;C*wc,*zc;D*wd,*zd;I c,it,j,k,m,r,*s,t,*u,*wi,*zi;
RZ(w&&ind);
r=MAX(0,AR(w)-1); s=1+AS(w); t=AT(w); k=bp(t); m=IC(w); c=aii(w);
ASSERT(!(t&SPARSE),EVNONCE);
ASSERT(r==AR(ind),EVRANK);
ASSERT(!ICMP(s,AS(ind),r),EVLENGTH);
GA(z,t,c,r,s);
if(!(AT(ind)&B01+INT))RZ(ind=cvt(INT,ind));
it=AT(ind); u=AV(ind); b=(B*)u;
ASSERT(!c||1<m||!(it&B01),EVINDEX);
ASSERT(!c||1!=m||!memchr(b,C1,c),EVINDEX);
zi=AV(z); zc=(C*)zi; zd=(D*)zc;
wi=AV(w); wc=(C*)wi; wd=(D*)wc;
switch(MCASE(it,k)){
case MCASE(B01,sizeof(C)): DO(c, *zc++=wc[*b++?i+c:i];); break;
case MCASE(B01,sizeof(I)): DO(c, *zi++=wi[*b++?i+c:i];); break;
int MutaQ(void *ad, void *as)
{
ACel aux = IC(as); //initializare (1)
if(IC(as)==NULL) return 0; //lista vida
aux = IC(as);
if (IC(as)->urm == NULL) //lista sursa are un element (2)
{
IC(as)=NULL;
SC(as)=NULL;
}else
IC(as)=IC(as)->urm;
aux->urm = NULL; // (3)
if(IC(ad)==NULL)
{
IC(ad)=aux; //daca coada destinatie era vida
SC(ad)=aux;
}else{
SC(ad)->urm=aux; // (4)
SC(ad)=aux; // (5)
}
return 1;
}
/* temporal update of clock --------------------------------------------------*/
static void udclk_ppp(rtk_t *rtk)
{
double dtr;
int i;
trace(3,"udclk_ppp:\n");
/* initialize every epoch for clock (white noise) */
for (i=0;i<NSYS;i++) {
if (rtk->opt.sateph==EPHOPT_PREC) {
/* time of prec ephemeris is based gpst */
/* negelect receiver inter-system bias */
dtr=rtk->sol.dtr[0];
}
else {
dtr=i==0?rtk->sol.dtr[0]:rtk->sol.dtr[0]+rtk->sol.dtr[i];
}
initx(rtk,CLIGHT*dtr,VAR_CLK,IC(i,&rtk->opt));
}
}
static DF1(breduce){A z;B b,*u,*v,*x,*xx;I c,cv,d,m;SF f2;VA*p;
RZ(w); /* AN(w)&&1<IC(w) */
m=IC(w); RZ(z=tail(w)); c=AN(z); x=BAV(z); v=BAV(w);
p=vap(self);
switch(1<c?0:p->bf){
case V0001: *x=memchr(v,C0,m)?0:1; R z;
case V0111: *x=memchr(v,C1,m)?1:0; R z;
case V1110: u=memchr(v,C0,m); d=u?u-v:m; *x=d%2!=d<m-1; R z;
case V1000: u=memchr(v,C1,m); d=u?u-v:m; *x=d%2==d<m-1; R z;
case V0010: u=memchr(v,C0,m); *x=(u?u-v:m)%2?1:0; R z;
case V1011: u=memchr(v,C1,m); *x=(u?u-v:m)%2?0:1; R z;
case V0100: *x= *(v+m-1)&&!memchr(v,C1,m-1)?1:0; R z;
case V1101: *x=!*(v+m-1)&&!memchr(v,C0,m-1)?0:1; R z;
case V0110: b=0; DO(m, b=b!=*v++); *x=b; R z;
case V1001: b=1; DO(m, b=b==*v++); *x=b; R z;
}
switch(p->id){I*x,*xx;
case CPLUS:
RZ(z=cvt(INT,z)); x=AV(z);
if(1==c){d=0; DO(m, if(*v++)d++); *x=d;}
else{xx=x+=c; v+=c*(m-1); DO(m-1, DO(c, --x; *x=*--v+*x); x=xx);}
static F1(jttrc){A bot,p,*v,x,y;B b;C*bv,c,ul,ll,*pv;I j,k,m,*s,xn,*xv,yn,*yv;
RZ(w);
s=AS(w); v=AAV(w);
xn=s[0]; RZ(x=apv(xn,0L,0L)); xv=AV(x);
yn=s[1]; RZ(y=apv(yn,0L,0L)); yv=AV(y);
j=0; DO(xn, xv[i]=IC(v[j]); j+=yn;);
/* phase and code residuals --------------------------------------------------*/
static int res_ppp(int iter, const obsd_t *obs, int n, const double *rs,
const double *dts, const double *vare, const int *svh,
const nav_t *nav, const double *x, rtk_t *rtk, double *v,
double *H, double *R, double *azel)
{
prcopt_t *opt=&rtk->opt;
double r,rr[3],disp[3],pos[3],e[3],meas[2],dtdx[3],dantr[NFREQ]={0};
double dants[NFREQ]={0},var[MAXOBS*2],dtrp=0.0,vart=0.0,varm[2]={0};
int i,j,k,sat,sys,nv=0,nx=rtk->nx,brk,tideopt;
trace(3,"res_ppp : n=%d nx=%d\n",n,nx);
for (i=0;i<MAXSAT;i++) rtk->ssat[i].vsat[0]=0;
for (i=0;i<3;i++) rr[i]=x[i];
/* earth tides correction */
if (opt->tidecorr) {
tideopt=opt->tidecorr==1?1:7; /* 1:solid, 2:solid+otl+pole */
tidedisp(gpst2utc(obs[0].time),rr,tideopt,&nav->erp,opt->odisp[0],
disp);
for (i=0;i<3;i++) rr[i]+=disp[i];
}
ecef2pos(rr,pos);
for (i=0;i<n&&i<MAXOBS;i++) {
sat=obs[i].sat;
if (!(sys=satsys(sat,NULL))||!rtk->ssat[sat-1].vs) continue;
/* geometric distance/azimuth/elevation angle */
if ((r=geodist(rs+i*6,rr,e))<=0.0||
satazel(pos,e,azel+i*2)<opt->elmin) continue;
/* excluded satellite? */
if (satexclude(obs[i].sat,svh[i],opt)) continue;
/* tropospheric delay correction */
if (opt->tropopt==TROPOPT_SAAS) {
dtrp=tropmodel(obs[i].time,pos,azel+i*2,REL_HUMI);
vart=SQR(ERR_SAAS);
}
else if (opt->tropopt==TROPOPT_SBAS) {
dtrp=sbstropcorr(obs[i].time,pos,azel+i*2,&vart);
}
else if (opt->tropopt==TROPOPT_EST||opt->tropopt==TROPOPT_ESTG) {
dtrp=prectrop(obs[i].time,pos,azel+i*2,opt,x+IT(opt),dtdx,&vart);
}
else if (opt->tropopt==TROPOPT_COR||opt->tropopt==TROPOPT_CORG) {
dtrp=prectrop(obs[i].time,pos,azel+i*2,opt,x,dtdx,&vart);
}
/* satellite antenna model */
if (opt->posopt[0]) {
satantpcv(rs+i*6,rr,nav->pcvs+sat-1,dants);
}
/* receiver antenna model */
antmodel(opt->pcvr,opt->antdel[0],azel+i*2,opt->posopt[1],dantr);
/* phase windup correction */
if (opt->posopt[2]) {
windupcorr(rtk->sol.time,rs+i*6,rr,&rtk->ssat[sat-1].phw);
}
/* ionosphere and antenna phase corrected measurements */
if (!corrmeas(obs+i,nav,pos,azel+i*2,&rtk->opt,dantr,dants,
rtk->ssat[sat-1].phw,meas,varm,&brk)) {
continue;
}
/* satellite clock and tropospheric delay */
r+=-CLIGHT*dts[i*2]+dtrp;
trace(5,"sat=%2d azel=%6.1f %5.1f dtrp=%.3f dantr=%6.3f %6.3f dants=%6.3f %6.3f phw=%6.3f\n",
sat,azel[i*2]*R2D,azel[1+i*2]*R2D,dtrp,dantr[0],dantr[1],dants[0],
dants[1],rtk->ssat[sat-1].phw);
for (j=0;j<2;j++) { /* for phase and code */
if (meas[j]==0.0) continue;
for (k=0;k<nx;k++) H[k+nx*nv]=0.0;
v[nv]=meas[j]-r;
for (k=0;k<3;k++) H[k+nx*nv]=-e[k];
if (sys!=SYS_GLO) {
v[nv]-=x[IC(0,opt)];
H[IC(0,opt)+nx*nv]=1.0;
}
else {
v[nv]-=x[IC(1,opt)];
H[IC(1,opt)+nx*nv]=1.0;
}
if (opt->tropopt>=TROPOPT_EST) {
for (k=0;k<(opt->tropopt>=TROPOPT_ESTG?3:1);k++) {
H[IT(opt)+k+nx*nv]=dtdx[k];
}
}
if (j==0) {
v[nv]-=x[IB(obs[i].sat,opt)];
H[IB(obs[i].sat,opt)+nx*nv]=1.0;
}
var[nv]=varerr(obs[i].sat,sys,azel[1+i*2],j,opt)+varm[j]+vare[i]+vart;
if (j==0) rtk->ssat[sat-1].resc[0]=v[nv];
else rtk->ssat[sat-1].resp[0]=v[nv];
/* test innovation */
#if 0
if (opt->maxinno>0.0&&fabs(v[nv])>opt->maxinno) {
#else
if (opt->maxinno>0.0&&fabs(v[nv])>opt->maxinno&&sys!=SYS_GLO) {
#endif
trace(2,"ppp outlier rejected %s sat=%2d type=%d v=%.3f\n",
time_str(obs[i].time,0),sat,j,v[nv]);
rtk->ssat[sat-1].rejc[0]++;
continue;
}
if (j==0) rtk->ssat[sat-1].vsat[0]=1;
nv++;
}
}
for (i=0;i<nv;i++) for (j=0;j<nv;j++) {
R[i+j*nv]=i==j?var[i]:0.0;
}
trace(5,"x=\n"); tracemat(5,x, 1,nx,8,3);
trace(5,"v=\n"); tracemat(5,v, 1,nv,8,3);
trace(5,"H=\n"); tracemat(5,H,nx,nv,8,3);
trace(5,"R=\n"); tracemat(5,R,nv,nv,8,5);
return nv;
}
/* number of estimated states ------------------------------------------------*/
extern int pppnx(const prcopt_t *opt)
{
return NX(opt);
}
/* precise point positioning -------------------------------------------------*/
extern void pppos(rtk_t *rtk, const obsd_t *obs, int n, const nav_t *nav)
{
const prcopt_t *opt=&rtk->opt;
double *rs,*dts,*var,*v,*H,*R,*azel,*xp,*Pp;
int i,nv,info,svh[MAXOBS],stat=SOLQ_SINGLE;
trace(3,"pppos : nx=%d n=%d\n",rtk->nx,n);
rs=mat(6,n); dts=mat(2,n); var=mat(1,n); azel=zeros(2,n);
for (i=0;i<MAXSAT;i++) rtk->ssat[i].fix[0]=0;
/* temporal update of states */
udstate_ppp(rtk,obs,n,nav);
trace(4,"x(0)="); tracemat(4,rtk->x,1,NR(opt),13,4);
/* satellite positions and clocks */
satposs(obs[0].time,obs,n,nav,rtk->opt.sateph,rs,dts,var,svh);
/* exclude measurements of eclipsing satellite */
if (rtk->opt.posopt[3]) {
testeclipse(obs,n,nav,rs);
}
xp=mat(rtk->nx,1); Pp=zeros(rtk->nx,rtk->nx);
matcpy(xp,rtk->x,rtk->nx,1);
nv=n*rtk->opt.nf*2; v=mat(nv,1); H=mat(rtk->nx,nv); R=mat(nv,nv);
for (i=0;i<rtk->opt.niter;i++) {
/* phase and code residuals */
if ((nv=res_ppp(i,obs,n,rs,dts,var,svh,nav,xp,rtk,v,H,R,azel))<=0) break;
/* measurement update */
matcpy(Pp,rtk->P,rtk->nx,rtk->nx);
if ((info=filter(xp,Pp,H,v,R,rtk->nx,nv))) {
trace(2,"ppp filter error %s info=%d\n",time_str(rtk->sol.time,0),
info);
break;
}
trace(4,"x(%d)=",i+1); tracemat(4,xp,1,NR(opt),13,4);
stat=SOLQ_PPP;
}
if (stat==SOLQ_PPP) {
/* postfit residuals */
res_ppp(1,obs,n,rs,dts,var,svh,nav,xp,rtk,v,H,R,azel);
/* update state and covariance matrix */
matcpy(rtk->x,xp,rtk->nx,1);
matcpy(rtk->P,Pp,rtk->nx,rtk->nx);
/* ambiguity resolution in ppp */
if (opt->modear==ARMODE_PPPAR||opt->modear==ARMODE_PPPAR_ILS) {
if (pppamb(rtk,obs,n,nav,azel)) stat=SOLQ_FIX;
}
/* update solution status */
rtk->sol.ns=0;
for (i=0;i<n&&i<MAXOBS;i++) {
if (!rtk->ssat[obs[i].sat-1].vsat[0]) continue;
rtk->ssat[obs[i].sat-1].lock[0]++;
rtk->ssat[obs[i].sat-1].outc[0]=0;
rtk->ssat[obs[i].sat-1].fix [0]=4;
rtk->sol.ns++;
}
rtk->sol.stat=stat;
for (i=0;i<3;i++) {
rtk->sol.rr[i]=rtk->x[i];
rtk->sol.qr[i]=(float)rtk->P[i+i*rtk->nx];
}
rtk->sol.qr[3]=(float)rtk->P[1];
rtk->sol.qr[4]=(float)rtk->P[2+rtk->nx];
rtk->sol.qr[5]=(float)rtk->P[2];
rtk->sol.dtr[0]=rtk->x[IC(0,opt)];
rtk->sol.dtr[1]=rtk->x[IC(1,opt)]-rtk->x[IC(0,opt)];
for (i=0;i<n&&i<MAXOBS;i++) {
rtk->ssat[obs[i].sat-1].snr[0]=MIN(obs[i].SNR[0],obs[i].SNR[1]);
}
for (i=0;i<MAXSAT;i++) {
if (rtk->ssat[i].slip[0]&3) rtk->ssat[i].slipc[0]++;
}
}
free(rs); free(dts); free(var); free(azel);
free(xp); free(Pp); free(v); free(H); free(R);
}
IGL_INLINE void igl::resolve_duplicated_faces(
const Eigen::PlainObjectBase<DerivedF1>& F1,
Eigen::PlainObjectBase<DerivedF2>& F2,
Eigen::PlainObjectBase<DerivedJ>& J) {
//typedef typename DerivedF1::Scalar Index;
Eigen::VectorXi IA,IC;
DerivedF1 uF;
igl::unique_simplices(F1,uF,IA,IC);
const size_t num_faces = F1.rows();
const size_t num_unique_faces = uF.rows();
assert((size_t) IA.rows() == num_unique_faces);
// faces on top of each unique face
std::vector<std::vector<int> > uF2F(num_unique_faces);
// signed counts
Eigen::VectorXi counts = Eigen::VectorXi::Zero(num_unique_faces);
Eigen::VectorXi ucounts = Eigen::VectorXi::Zero(num_unique_faces);
// loop over all faces
for (size_t i=0; i<num_faces; i++) {
const size_t ui = IC(i);
const bool consistent =
(F1(i,0) == uF(ui, 0) && F1(i,1) == uF(ui, 1) && F1(i,2) == uF(ui, 2)) ||
(F1(i,0) == uF(ui, 1) && F1(i,1) == uF(ui, 2) && F1(i,2) == uF(ui, 0)) ||
(F1(i,0) == uF(ui, 2) && F1(i,1) == uF(ui, 0) && F1(i,2) == uF(ui, 1));
uF2F[ui].push_back(int(i+1) * (consistent?1:-1));
counts(ui) += consistent ? 1:-1;
ucounts(ui)++;
}
std::vector<size_t> kept_faces;
for (size_t i=0; i<num_unique_faces; i++) {
if (ucounts[i] == 1) {
kept_faces.push_back(abs(uF2F[i][0])-1);
continue;
}
if (counts[i] == 1) {
bool found = false;
for (auto fid : uF2F[i]) {
if (fid > 0) {
kept_faces.push_back(abs(fid)-1);
found = true;
break;
}
}
assert(found);
} else if (counts[i] == -1) {
bool found = false;
for (auto fid : uF2F[i]) {
if (fid < 0) {
kept_faces.push_back(abs(fid)-1);
found = true;
break;
}
}
assert(found);
} else {
assert(counts[i] == 0);
}
}
const size_t num_kept = kept_faces.size();
J.resize(num_kept, 1);
std::copy(kept_faces.begin(), kept_faces.end(), J.data());
igl::slice(F1, J, 1, F2);
}
Z K3(zframesend){PC(x); PC(y); TC2(z,-KI,-KJ); IC(z); ZTK(zframe_t,f); R kj(zframe_send(&f, VSK(y), zi));}
int main() {
//SPECIFY PARMETERS, SEE CONFIG.H FILE FOR ENUM OPTIONS
Configurations config;
//Algorithm for solving for h
config.algorithm_solve_h = NEWTON_BRUTE_FORCE;
//Total time in years
config.total_time = 20;
// Initial time step in seconds
config.initial_time_step = 30000;//12592000;
// Injection stop
config.injection_time = 100;
// Pressure update interval in years
config.pressure_update_injection = 2;
config.pressure_update_migration = 5;
// Permeability type
config.perm_type = PERM_CONSTANT;
// Name of formation
config.formation_name = UTSIRA;
// Parameter beta for the Corey permeability function
config.beta = 0.4;
//////////////////////////////////////////////////////////////////////////
if (config.formation_name == UTSIRA){
config.formation = "utsira";
config.formation_dir = "Utsira";
} else if(config.formation_name == JOHANSEN){
config.formation = "johansen";
config.formation_dir = "Johansen";
} else {
printf("ERROR: No data for this formation");
}
//Initialize some variables
int nx, ny, nz;
float dt, dz;
float t, tf;
int year = 60*60*24*365;
// Device properties
cudaSetDevice(0);
int device;
cudaGetDevice(&device);
cudaDeviceProp p;
cudaGetDeviceProperties(&p, device);
printf("Device name: %s\n", p.name);
// Set directory path of output and input files
size_t buff_size= 100;
char buffer[buff_size];
const char* path;
readlink("/proc/self/exe", buffer, buff_size);
printf("PATH %s", buffer);
char* output_dir_path = "FullyIntegratedVESimulatorMATLAB/SimulationData/ResultData/";
char* input_dir_path = "FullyIntegratedVESimulatorMATLAB/SimulationData/FormationData/";
std::cout << "Trying to open " << input_dir_path << std::endl;
std::cout << "Output will end up in " << output_dir_path << std::endl;
// Filename strings
char dir_input[300];
char filename_input[300];
char dir_output[300];
strcpy(dir_input, input_dir_path);
strcat(dir_input, config.formation_dir);
strcat(dir_input, "/");
strcpy(dir_output, output_dir_path);
strcat(dir_output, config.formation_dir);
strcat(dir_output, "/");
strcpy(filename_input, dir_input);
strcat (filename_input, "dimensions.mat");
// Output txt files with results
FILE* matlab_file_h;
FILE* matlab_file_coarse_satu;
FILE* matlab_file_volume;
// Create output files for coarse saturation, interface height and volume
// Files are stored in the directory
createOutputFiles(matlab_file_h, matlab_file_coarse_satu, matlab_file_volume, dir_output);
readDimensionsFromMATLABFile(filename_input, nx, ny, nz);
InitialConditions IC(nx, ny, 5);
printf("nx: %i, ny: %i nz: %i dt: %.10f", nx, ny, nz, dt);
// Cpu pointers to store formation data from MATLAB
CpuPtr_2D H(nx, ny, 0, true);
CpuPtr_2D top_surface(nx, ny, 0, true);
CpuPtr_2D h(nx, ny, 0, true);
CpuPtr_2D normal_z(nx, ny, 0, true);
CpuPtr_3D perm3D(nx, ny, nz + 1, 0, true);
CpuPtr_3D poro3D(nx, ny, nz + 1, 0, true);
CpuPtr_2D pv(nx, ny, 0, true);
CpuPtr_2D flux_north(nx, ny, IC.border, true);
CpuPtr_2D flux_east(nx, ny, IC.border, true);
CpuPtr_2D source(nx, ny, 0, true);
CpuPtr_2D grav_north(nx, ny, 0, true);
CpuPtr_2D grav_east(nx, ny, 0, true);
CpuPtr_2D K_face_north(nx, ny, 0, true);
CpuPtr_2D K_face_east(nx, ny, 0, true);
CpuPtr_2D active_east(nx, ny, 0, true);
CpuPtr_2D active_north(nx, ny, 0, true);
CpuPtr_2D volume(nx, ny, 0,true);
strcpy(filename_input, dir_input);
strcat (filename_input, "data.mat");
readFormationDataFromMATLABFile(filename_input, H.getPtr(), top_surface.getPtr(),
h.getPtr(), normal_z.getPtr(), perm3D.getPtr(), poro3D.getPtr(),
pv.getPtr(), flux_north.getPtr(), flux_east.getPtr(),
grav_north.getPtr(), grav_east.getPtr(), K_face_north.getPtr(),
K_face_east.getPtr(), dz);
strcpy(filename_input, dir_input);
strcat (filename_input, "active_cells.mat");
readActiveCellsFromMATLABFile(filename_input, active_east.getPtr(), active_north.getPtr());
//readDtTableFromMATLABFile(filename, dt_table, size_dt_table);
Engine *ep;
if (!(ep = engOpen(""))) {
fprintf(stderr, "\nCan't start MATLAB engine\n");
return EXIT_FAILURE;
}
startMatlabEngine(ep, config.formation_dir);
// Create double precision array for the data exchange between the GPU program and the MATLAB program
mxArray *h_matrix = NULL, *flux_east_matrix = NULL, *flux_north_matrix=NULL;
mxArray *source_matrix = NULL, *open_well = NULL;
open_well = mxCreateLogicalScalar(true);
engPutVariable(ep, "open_well", open_well);
double * h_matlab_matrix;
h_matlab_matrix = new double[nx*ny];
h_matrix = mxCreateDoubleMatrix(nx, ny, mxREAL);
flux_east_matrix = mxCreateDoubleMatrix(nx+2*IC.border,ny+2*IC.border,mxREAL);
flux_north_matrix = mxCreateDoubleMatrix(nx+2*IC.border,ny+2*IC.border,mxREAL);
source_matrix = mxCreateDoubleMatrix(nx, ny, mxREAL);
// Cpu Pointer to store the results
CpuPtr_2D zeros(nx, ny, 0, true);
//Initial Conditions
IC.dz = dz;
IC.createnIntervalsTable(H);
IC.createScalingParameterTable(H, config.beta);
IC.createInitialCoarseSatu(H, h);
IC.computeAllGridBlocks();
IC.createDtVec();
// Create mask for sparse grid on GPU
std::vector<int> active_block_indexes;
std::vector<int> active_block_indexes_flux;
int n_active_blocks = 0;
int n_active_blocks_flux = 0;
createGridMask(H, IC.grid, IC.block, nx, ny, active_block_indexes,
n_active_blocks);
createGridMaskFlux(H, IC.grid_flux, IC.block_flux, nx, ny, active_block_indexes_flux,
n_active_blocks_flux);
// Print grid mask properties
printf("\n nBlocks: %i nActiveBlocks: %i fraction: %.5f\n", IC.grid.x * IC.grid.y,
n_active_blocks, (float) n_active_blocks / (IC.grid.x * IC.grid.y));
printf("nBlocks: %i nActiveBlocks: %i fraction: %.5f\n", IC.grid_flux.x * IC.grid_flux.y,
n_active_blocks_flux, (float) n_active_blocks_flux / (IC.grid_flux.x * IC.grid_flux.y));
printf("dz: %.3f\n", IC.dz);
dim3 new_sparse_grid(n_active_blocks, 1, 1);
dim3 new_sparse_grid_flux(n_active_blocks_flux, 1, 1);
CommonArgs common_args;
CoarseMobIntegrationKernelArgs coarse_mob_int_args;
CoarsePermIntegrationKernelArgs coarse_perm_int_args;
FluxKernelArgs flux_kernel_args;
TimeIntegrationKernelArgs time_int_kernel_args;
TimestepReductionKernelArgs time_red_kernel_args;
SolveForhProblemCellsKernelArgs solve_problem_cells_args;
printf("Cuda error 0.5: %s\n", cudaGetErrorString(cudaGetLastError()));
initAllocate(&common_args, &coarse_perm_int_args, &coarse_mob_int_args,
&flux_kernel_args, &time_int_kernel_args, &time_red_kernel_args, &solve_problem_cells_args);
h.convertToDoublePointer(h_matlab_matrix);
memcpy((void *)mxGetPr(h_matrix), (void *)h_matlab_matrix, sizeof(double)*nx*ny);
engPutVariable(ep, "h_matrix", h_matrix);
printf("Cuda error 1: %s\n", cudaGetErrorString(cudaGetLastError()));
// Allocate and set data on the GPU
GpuPtr_3D perm3D_device(nx, ny, nz + 1, 0, perm3D.getPtr());
GpuPtr_2D Lambda_c_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D Lambda_b_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D dLambda_c_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D dLambda_b_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D scaling_parameter_C_device(nx, ny, 0, IC.scaling_parameter.getPtr());
GpuPtr_2D K_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D H_device(nx, ny, 0, H.getPtr());
GpuPtr_2D h_device(nx, ny, 0, h.getPtr());
GpuPtr_2D top_surface_device(nx, ny, 0, top_surface.getPtr());
GpuPtr_2D z_diff_east_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D z_diff_north_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D nInterval_device(nx, ny, 0, IC.nIntervals.getPtr());
GpuPtr_2D U_x_device(nx, ny, IC.border, flux_east.getPtr());
GpuPtr_2D U_y_device(nx, ny, IC.border, flux_north.getPtr());
GpuPtr_2D source_device(nx, ny, 0, source.getPtr());
GpuPtr_2D K_face_east_device(nx, ny, 0, K_face_east.getPtr());
GpuPtr_2D K_face_north_device(nx, ny, 0, K_face_north.getPtr());
GpuPtr_2D grav_east_device(nx, ny, 0, grav_east.getPtr());
GpuPtr_2D grav_north_device(nx, ny, 0, grav_north.getPtr());
GpuPtr_2D normal_z_device(nx, ny, 0, normal_z.getPtr());
GpuPtr_2D R_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D pv_device(nx, ny, 0, pv.getPtr());
GpuPtr_2D output_test_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D active_east_device(nx, ny, 0, active_east.getPtr());
GpuPtr_2D active_north_device(nx, ny, 0, active_north.getPtr());
GpuPtr_2D vol_old_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D vol_new_device(nx, ny, 0, zeros.getPtr());
GpuPtr_2D coarse_satu_device(nx, ny, 0, IC.initial_coarse_satu_c.getPtr());
GpuPtr_1D global_dt_device(3, IC.global_time_data);
GpuPtr_1D dt_vector_device(IC.nElements, IC.dt_vector);
GpuPtrInt_1D active_block_indexes_device(n_active_blocks,
&active_block_indexes[0]);
GpuPtrInt_1D active_block_indexes_flux_device(n_active_blocks_flux,
&active_block_indexes_flux[0]);
cudaCheckError();
setCommonArgs(&common_args, IC.p_ci, IC.delta_rho, IC.g, IC.mu_c, IC.mu_b,
IC.s_c_res, IC.s_b_res, IC.lambda_end_point_c, IC.lambda_end_point_b,
active_east_device.getRawPtr(), active_north_device.getRawPtr(),
H_device.getRawPtr(), pv_device.getRawPtr(),
nx, ny, IC.border);
setupGPU(&common_args);
setCoarsePermIntegrationKernelArgs(&coarse_perm_int_args,
K_device.getRawPtr(), perm3D_device.getRawPtr(),
nInterval_device.getRawPtr(), IC.dz);
callCoarsePermIntegrationKernel(IC.grid, IC.block, &coarse_perm_int_args);
setCoarseMobIntegrationKernelArgs(&coarse_mob_int_args,
Lambda_c_device.getRawPtr(), Lambda_b_device.getRawPtr(),
dLambda_c_device.getRawPtr(), dLambda_b_device.getRawPtr(),
h_device.getRawPtr(), perm3D_device.getRawPtr(),
K_device.getRawPtr(), nInterval_device.getRawPtr(),
scaling_parameter_C_device.getRawPtr(),
active_block_indexes_device.getRawPtr(), IC.p_ci, IC.dz, config.perm_type);
setFluxKernelArgs(&flux_kernel_args,
Lambda_c_device.getRawPtr(),Lambda_b_device.getRawPtr(),
dLambda_c_device.getRawPtr(), dLambda_b_device.getRawPtr(),
U_x_device.getRawPtr(), U_y_device.getRawPtr(), source_device.getRawPtr(),
h_device.getRawPtr(),top_surface_device.getRawPtr(),
normal_z_device.getRawPtr(),
K_face_east_device.getRawPtr(), K_face_north_device.getRawPtr(),
grav_east_device.getRawPtr(), grav_north_device.getRawPtr(),
R_device.getRawPtr(), dt_vector_device.getRawPtr(), active_block_indexes_flux_device.getRawPtr(),
output_test_device.getRawPtr());
setTimestepReductionKernelArgs(&time_red_kernel_args, TIME_THREADS, IC.nElements, global_dt_device.getRawPtr(),
IC.cfl_scale, dt_vector_device.getRawPtr());
//CUDPP
CUDPPHandle plan;
unsigned int* d_isValid;
int* d_in;
int* d_out;
size_t* d_numValid = NULL;
size_t numValidElements;
unsigned int numElements=nx*ny;
cudaCheckError();
cudaMalloc((void**) &d_isValid, sizeof(unsigned int)*numElements);
cudaMalloc((void**) &d_in, sizeof(int)*numElements);
cudaMalloc((void**) &d_out, sizeof(int)*numElements);
cudaMalloc((void**) &d_numValid, sizeof(size_t));
cudaCheckError();
CUDPPHandle theCudpp;
cudppCreate(&theCudpp);
setUpCUDPP(theCudpp, plan, nx, ny, d_isValid, d_in, d_out, numElements);
printf("\nCudaMemcpy error: %s", cudaGetErrorString(cudaGetLastError()));
cudaCheckError();
setTimeIntegrationKernelArgs(&time_int_kernel_args, global_dt_device.getRawPtr(),
IC.integral_res,pv_device.getRawPtr(), h_device.getRawPtr(),
R_device.getRawPtr(),coarse_satu_device.getRawPtr(),
scaling_parameter_C_device.getRawPtr(),
vol_old_device.getRawPtr(), vol_new_device.getRawPtr(),
d_isValid, d_in);
cudaCheckError();
setSolveForhProblemCellsKernelArgs(&solve_problem_cells_args, h_device.getRawPtr(),
coarse_satu_device.getRawPtr(), scaling_parameter_C_device.getRawPtr(),
d_out, IC.integral_res, d_numValid);
cudaCheckError();
//Compute start volume
callCoarseMobIntegrationKernel(new_sparse_grid, IC.block, IC.grid.x, &coarse_mob_int_args);
callFluxKernel(new_sparse_grid_flux, IC.block_flux, IC.grid_flux.x, &flux_kernel_args);
callTimeIntegration(new_sparse_grid, IC.block, IC.grid.x, &time_int_kernel_args);
cudaCheckError();
vol_old_device.download(volume.getPtr(), 0, 0, nx, ny);
float total_volume_old = computeTotalVolume(volume, nx, ny);
t = 0;
double t2 = 0;
tf = IC.global_time_data[2];
int iter_outer_loop = 0;
int iter_inner_loop = 0;
int iter_total = 0;
int iter_total_lim = 5000;
float time = 0;
float injected = 0;
int table_index = 1;
double time_start = getWallTime();
double time_start_iter;
double total_time_gpu = 0;
while (time < config.total_time && iter_total < iter_total_lim){
t = 0;
iter_inner_loop = 0;
h_device.download(h.getPtr(), 0, 0, nx, ny);
h.convertToDoublePointer(h_matlab_matrix);
memcpy((void *)mxGetPr(h_matrix), (void *)h_matlab_matrix, sizeof(double)*nx*ny);
engPutVariable(ep, "h_matrix", h_matrix);
if (time >= config.injection_time){
open_well = mxCreateLogicalScalar(false);
engPutVariable(ep, "open_well", open_well);
IC.global_time_data[2] = 31536000*config.pressure_update_migration;
tf = IC.global_time_data[2];
cudaMemcpy(global_dt_device.getRawPtr(), IC.global_time_data, sizeof(float)*3, cudaMemcpyHostToDevice);
}
// MATLAB call
engEvalString(ep, "[source, east_flux, north_flux] = pressureFunctionToRunfromCpp(h_matrix, variables, open_well);");
// Get variables from MATLABs pressure solver
flux_east_matrix = engGetVariable(ep, "east_flux");
flux_north_matrix = engGetVariable(ep, "north_flux");
source_matrix = engGetVariable(ep, "source");
memcpy((void *)flux_east.getPtr(), (void *)mxGetPr(flux_east_matrix), sizeof(float)*(nx+2*IC.border)*(ny+2*IC.border));
memcpy((void *)flux_north.getPtr(), (void *)mxGetPr(flux_north_matrix), sizeof(float)*(nx+2*IC.border)*(ny+2*IC.border));
memcpy((void *)source.getPtr(), (void *)mxGetPr(source_matrix), sizeof(float)*nx*ny);
source_device.upload(source.getPtr(), 0, 0, nx, ny);
U_x_device.upload(flux_east.getPtr(), 0, 0, nx+2*IC.border, ny+2*IC.border);
U_y_device.upload(flux_north.getPtr(), 0, 0,nx+2*IC.border, ny+2*IC.border);
time_start_iter = getWallTime();
while (t < tf && iter_total < iter_total_lim){
cudaCheckError();
callCoarseMobIntegrationKernel(new_sparse_grid, IC.block, IC.grid.x, &coarse_mob_int_args);
cudaCheckError();
callFluxKernel(new_sparse_grid_flux, IC.block_flux, IC.grid_flux.x, &flux_kernel_args);
cudaCheckError();
callTimestepReductionKernel(TIME_THREADS, &time_red_kernel_args);
cudaCheckError();
// Set the initial time step
if (iter_total < 1 && iter_inner_loop == 0){
IC.global_time_data[0] = config.initial_time_step;
IC.global_time_data[1] += config.initial_time_step;
cudaMemcpy(global_dt_device.getRawPtr(), IC.global_time_data, sizeof(float)*3, cudaMemcpyHostToDevice);
}
//For precomputed time step insertion
/*
IC.global_time_data[0] = (float)dt_table[table_index];
IC.global_time_data[1] += (float)dt_table[table_index];
cudaMemcpy(global_dt_device.getRawPtr(), IC.global_time_data, sizeof(float)*3, cudaMemcpyHostToDevice);
*/
cudaCheckError();
if (config.algorithm_solve_h == BRUTE_FORCE)
callTimeIntegration(new_sparse_grid, IC.block, IC.grid.x, &time_int_kernel_args);
else if (config.algorithm_solve_h == NEWTON_BRUTE_FORCE){
callTimeIntegrationNewton(new_sparse_grid, IC.block, IC.grid.x, &time_int_kernel_args);
cudppCompact(plan, d_out, d_numValid, d_in, d_isValid, numElements);
callSolveForhProblemCells(IC.grid_pc, IC.block_pc, &solve_problem_cells_args);
} else if (config.algorithm_solve_h == NEWTON_BISECTION){
callTimeIntegrationNewton(new_sparse_grid, IC.block, IC.grid.x, &time_int_kernel_args);
cudppCompact(plan, d_out, d_numValid, d_in, d_isValid, numElements);
callSolveForhProblemCellsBisection(IC.grid_pc, IC.block_pc, &solve_problem_cells_args);
}
cudaCheckError();
cudaMemcpy(IC.global_time_data, global_dt_device.getRawPtr(), sizeof(float)*3, cudaMemcpyDeviceToHost);
cudaCheckError();
// Keep track of injected volume, insert injection coordinate and rate
injected += IC.global_time_data[0]*source(50,50);
t += IC.global_time_data[0];
//table_index++;
iter_inner_loop++;
iter_total++;
}
total_time_gpu += getWallTime() - time_start_iter;
printf("Total time in years: %.3f time in this round %.3f timestep %.3f GPU time %.3f time per iter %.8f \n", time, t, IC.global_time_data[0], getWallTime() - time_start_iter, (getWallTime() - time_start_iter)/iter_inner_loop);
time += t/(year);
iter_outer_loop++;
}
printf("Elapsed time program: %.5f gpu part: %.5f", getWallTime() - time_start, total_time_gpu);
engClose(ep);
h_device.download(zeros.getPtr(), 0, 0, nx, ny);
zeros.printToFile(matlab_file_h);
coarse_satu_device.download(zeros.getPtr(), 0, 0, nx, ny);
// Divide by the formation thickness H to get the actual coarse satu
for (int i = 0; i < nx; i++){
for (int j = 0; j < ny; j++){
if (H(i,j) != 0)
zeros(i,j) = zeros(i,j)/H(i,j);
}
}
zeros.printToFile(matlab_file_coarse_satu);
vol_new_device.download(zeros.getPtr(), 0, 0, nx, ny);
zeros.printToFile(matlab_file_volume);
float total_volume_new = computeTotalVolume(zeros, nx, ny);
printf("total volume new %.2f total volume old %.2f injected %.1f injected fraction %.10f",
total_volume_new, total_volume_old, injected, (total_volume_new-injected)/(injected));
printf("volume fraction %.10f",(total_volume_new-total_volume_old)/(total_volume_old));
printf("\nCudaMemcpy error: %s", cudaGetErrorString(cudaGetLastError()));
printf("FINITO precise time %.6f iter_total %i", time, iter_total);
}
int main() {
EifList<int> TestList (True), TestList2 (True);
EifIterator TestIt, *TestItptr;
int d[] = {1, 2, 3, 1, 4, 5, 6, 7, 8, 8, 9, 10, -1}, i;
for (i = 0; d [i] != -1; i++) {
cout << "\n" << d[i];
TestList.Add (d[i]);
}
cout << "\n\nList Test\n";
cout << "Should be 10\t" << TestList.Count() << "\n";
TestList.Remove (11);
cout << "Should be 10\t" << TestList.Count() << "\n";
TestList.Remove (9);
cout << "Should be 9\t" << TestList.Count() << "\n";
TestList.Search (9);
cout <<"Should be 0\t" << TestList.Found() << "\n";
TestList.Search (6);
cout <<"Should be 1:\t" << TestList.Found() << "\n";
cout << "Should be 6:\t" << TestList.FoundItem() << "\n";
IC (0, TestList);
TestItptr = new EifIterator (TestList);
IC (1, TestList);
EifIterator TestIt2 = *TestItptr;
IC (2, TestList);
TestIt2 = TestIt;
IC (1, TestList);
DummyIt (1, *TestItptr, TestList);
delete TestItptr;
TestItptr = NULL;
IC (0, TestList);
Dummy (TestList);
IC (0, TestList);
TestIt = TestList.Iterator();
IC (1, TestList);
while (!TestIt.Finished()) {
cout <<'\n' << TestList.Item (TestIt);
TestIt.Forth();
}
IC (0, TestList);
TestIt = TestList2.Iterator();
IC (0, TestList2);
TestList2.Add (1);
TestIt.First();
IC (1, TestList2);
TestIt.Forth();
IC (0, TestList2);
return 0;
}
static DF1(jtsuffix){DECLF;I r;
RZ(w);
if(jt->rank&&jt->rank[1]<AR(w)){r=jt->rank[1]; jt->rank=0; R rank1ex(w,self,r,jtsuffix);}
jt->rank=0;
R eachl(IX(IC(w)),w,atop(fs,ds(CDROP)));
} /* f\."r w for general f */
static void generate_round_keys(int rnds,uint32_t* K, uint32_t* x, uint32_t* z)
{
COMPUTE_Z;
K[1]=S5[IA(z[2])]^S6[IB(z[2])]^S7[ID(z[1])]^S8[IC(z[1])]^S5[IC(z[0])];
K[2]=S5[IC(z[2])]^S6[ID(z[2])]^S7[IB(z[1])]^S8[IA(z[1])]^S6[IC(z[1])];
K[3]=S5[IA(z[3])]^S6[IB(z[3])]^S7[ID(z[0])]^S8[IC(z[0])]^S7[IB(z[2])];
K[4]=S5[IC(z[3])]^S6[ID(z[3])]^S7[IB(z[0])]^S8[IA(z[0])]^S8[IA(z[3])];
COMPUTE_X;
K[5]=S5[ID(x[0])]^S6[IC(x[0])]^S7[IA(x[3])]^S8[IB(x[3])]^S5[IA(x[2])];
K[6]=S5[IB(x[0])]^S6[IA(x[0])]^S7[IC(x[3])]^S8[ID(x[3])]^S6[IB(x[3])];
K[7]=S5[ID(x[1])]^S6[IC(x[1])]^S7[IA(x[2])]^S8[IB(x[2])]^S7[ID(x[0])];
K[8]=S5[IB(x[1])]^S6[IA(x[1])]^S7[IC(x[2])]^S8[ID(x[2])]^S8[ID(x[1])];
COMPUTE_Z;
K[9]=S5[ID(z[0])]^S6[IC(z[0])]^S7[IA(z[3])]^S8[IB(z[3])]^S5[IB(z[2])];
K[10]=S5[IB(z[0])]^S6[IA(z[0])]^S7[IC(z[3])]^S8[ID(z[3])]^S6[IA(z[3])];
K[11]=S5[ID(z[1])]^S6[IC(z[1])]^S7[IA(z[2])]^S8[IB(z[2])]^S7[IC(z[0])];
K[12]=S5[IB(z[1])]^S6[IA(z[1])]^S7[IC(z[2])]^S8[ID(z[2])]^S8[IC(z[1])];
COMPUTE_X;
if (rnds==16) {
K[13]=S5[IA(x[2])]^S6[IB(x[2])]^S7[ID(x[1])]^S8[IC(x[1])]^S5[ID(x[0])];
K[14]=S5[IC(x[2])]^S6[ID(x[2])]^S7[IB(x[1])]^S8[IA(x[1])]^S6[ID(x[1])];
K[15]=S5[IA(x[3])]^S6[IB(x[3])]^S7[ID(x[0])]^S8[IC(x[0])]^S7[IA(x[2])];
K[16]=S5[IC(x[3])]^S6[ID(x[3])]^S7[IB(x[0])]^S8[IA(x[0])]^S8[IB(x[3])];
}
}
static char *
split_lujvo (char *x)
{
static char result[2048];
int first=1; /* At start, so we can handle r and n hyphens */
result[0] = 0;
while (1) {
/* Advance x to look at the tail of the string. */
int len = strlen(x);
switch (len) {
case 0:
case 1:
case 2:
/* Impossible */
return NULL;
break;
case 3:
case 4:
{
char *comp = lookup_component(x);
if (comp) {
if (!first) strcat(result, "+");
strcat(result,comp);
}
return result;
}
break;
case 5:
if (!first) strcat(result, "+");
strcat(result, x);
return result;
break;
default:
{
if (x[3] == 'y') {
strip_leading_rafsi(&x, 3, 1, first, result);
} else if (x[4] == 'y') {
strip_leading_rafsi(&x, 4, 1, first, result);
} else if (x[5] == 'y') {
/* Handle cultural rafsi of form CCVVCy */
strip_leading_rafsi(&x, 5, 1, first, result);
} else if ((x[6] == 'y') && (x[3] == '\'')) {
/* Handle cultural rafsi of form CCV'VCy */
strip_leading_rafsi(&x, 6, 1, first, result);
} else {
/* Have to decide if there's an r/n hyphen */
if (!is_consonant(x[0])) return NULL; /* can't ever happen */
if (IC(x[1]) && IV(x[2])) { /* CCV at start */
strip_leading_rafsi(&x, 3, 0, first, result);
} else if (IV(x[1]) && IC(x[2])) {
strip_leading_rafsi(&x, 3, 0, first, result);
} else if (IV(x[1])) {
if (x[2] == '\'') {
/* Everything pushed up by 1 */
if (IV(x[3])) {
if (((x[4] == 'r') && (IC(x[5]))) ||
((x[4] == 'n') && (x[5] == 'r'))) {
/* This pattern can only occur if a hyphen is there. */
strip_leading_rafsi(&x, 4, 1, first, result);
} else {
strip_leading_rafsi(&x, 4, 0, first, result);
}
} else {
/* CV'C form, impossible. */
return NULL;
}
} else {
if (IV(x[2])) {
if (((x[3] == 'r') && (IC(x[4]))) ||
((x[3] == 'n') && (x[4] == 'r'))) {
/* This pattern can only occur if a hyphen is there. */
strip_leading_rafsi(&x, 3, 1, first, result);
} else {
strip_leading_rafsi(&x, 3, 0, first, result);
}
} else {
/* CVC form, no r/n hyphen */
strip_leading_rafsi(&x, 3, 0, first, result);
}
}
} else {
/* Impossible */
return NULL;
}
}
}
break;
}
first = 0;
}
}
size_t NrEQ(void* a) /* numarul de elemente din coada */
{ size_t n = 0;
ACel p = IC(a); /* p - adresa primei celule */
for(; p; p = p->urm) n++;
return n;
}
int PrimQ(void *ac, void *ae) /* copiaza primul element din coada la adresa ae */
{ if(!IC(ac)) return 0; /* coada vida -> "esec" */
memcpy(ae, &(IC(ac)->info), DIME(ac)); /* preia informatia */
return 1; /* operatie reusita -> "succes" */
}
Z K3(device){TC2(x,-KI,-KJ); IC(x); PC(y); PC(z); R kj(zmq_device(xi, VSK(y), VSK(z)));}
