/*
* only used in estimation
*/
double topicprob(int d, int t, int Ttot) {
if ( ddP.theta )
return ddP.theta[d][t];
if ( ddP.PYalpha==H_None )
return ((double)ddS.Ndt[d][t]+ddP.alphapr[t])
/ ((double)ddS.NdT[d]+ddP.alphatot);
if ( ddP.PYalpha==H_HPDD && ddS.TDt[t]==0 ) {
/*
* special case for HPDD with a topic with 0 occupancy
* to handle the introduction of a new topic into a document;
* spread probability over all possible NULL cases
*/
return (ddP.bpar+ddP.apar*Ttot)/((double)ddP.bpar+ddS.NdT[d])
* (ddP.b0+ddP.a0*ddS.TDTnz)/(ddP.b0+ddS.TDT)
/(ddN.T-ddS.TDTnz);
}
/*
* standard PYP result using parent base rate
*/
if ( ddS.Tdt[d][t]==0 ) {
return ((double)ddP.bpar+ddP.apar*Ttot) * alphabasetopicprob(t)
/ ((double)ddP.bpar+ddS.NdT[d]);
}
return (ddS.Ndt[d][t] - ddP.apar*ddS.Tdt[d][t]
+ ((double)ddP.bpar+ddP.apar*Ttot) * alphabasetopicprob(t))
/ ((double)ddP.bpar+ddS.NdT[d]);
}
static double merge_sumapprox_Tdt(int k) {
double x[ddN.DT];
int d;
for (d=0; d<ddN.DT; d++) {
double xp = 0, xm = 0;
int nn, tt, TdT;
nn = ddS.Ndt[d][k];
if ( nn<=1 ) {
x[d] = 0;
continue;
}
tt = ddS.Tdt[d][k];
TdT = getTdT(d);
if ( tt<nn )
xp = (ddP.bpar + ddP.apar*TdT) * S_V(ddC.SX,nn,tt+1)
* alphabasetopicprob(k);
if ( tt>1) {
ddS.TDt[k]--; ddS.TDT--;
xm = 1.0/(ddP.bpar + ddP.apar*(TdT-1)) / S_V(ddC.SX,nn,tt)
/ alphabasetopicprob(k);
ddS.TDt[k]++; ddS.TDT++;
}
if ( xm>xp )
x[d] = xm;
else
x[d] = xp;
}
// ?????????
}
/*
* prob. the doc table indicator is increased, but not forced
*/
void doctableindicatorprob(int d, int t, int Ttot,
double *uone, double *uzero) {
int nn = ddS.Ndt[d][t];
int tt = ddS.Tdt[d][t];
double e1, e0;
e1 = S_UV(ddC.SX,nn,tt+1);
if ( tt==1 )
e0 = nn - ddP.apar;
else
e0 = S_U(ddC.SX,nn,tt);
*uone = e1 * (ddP.bpar+ddP.apar*Ttot) * alphabasetopicprob(t)
* (tt+1)/(nn+1);
*uzero = e0 * (nn-tt+1)/(nn+1);
}
/*
* probability of topic given document
*
* *zerod - set to zero if a new topic is suggested
* *tip - set to prob. indicator would be 1
* Ttot - total tables
*/
double topicfact(int d, int t, int Ttot, uint16_t *zerod, float *tip) {
if ( ddP.PYalpha ) {
double p;
if ( ddP.PYalpha==H_HPDD && ddS.TDt[t]==0 ) {
/*
* special case for HPDD with a topic with 0 occupancy
* to handle the introduction of a new topic into a document
*/
if ( *zerod ) {
/* want to only do first time */
p = (ddP.bpar+ddP.apar*Ttot) *
(ddP.b0+ddP.a0*ddS.TDTnz)/(ddP.b0+ddS.TDT);
*zerod = 0;
} else {
/* subsequent times we set it to zero */
p = 0;
}
*tip = 1.0;
return p;
}
if ( ddS.Tdt[d][t]==0 ) {
#ifndef NDEBUG
if ( ddS.Ndt[d][t]>0 ) {
check_Ndt(d);
assert(ddS.Ndt[d][t]==0);
}
#endif
p = ((double)ddP.bpar+ddP.apar*Ttot) * alphabasetopicprob(t);
*tip = 1.0;
} else {
double uone, uzero;
#ifndef NDEBUG
if ( ddS.Ndt[d][t]==0 ) {
check_Ndt(d);
assert(ddS.Ndt[d][t]>0);
}
#endif
doctableindicatorprob(d, t, Ttot, &uone, &uzero);
p = uone + uzero;
*tip = uone/(uone + uzero);
}
return p;
}
return ((double)ddS.Ndt[d][t]+ddP.alphapr[t]);
}
void get_probs(double *vp) {
int zerod = 1;
int t;
double tot = 0;
if ( ddP.PYalpha==0 ) {
get_probs_alpha(vp);
return;
}
for (t=0; t<ddN.T; t++) {
if ( ddP.PYalpha!=H_HPDD || ddS.TDt[t]>0 || zerod )
tot += vp[t] = alphabasetopicprob(t);
else
vp[t] = 0;
if ( ddP.PYalpha==H_HPDD && ddS.TDt[t]==0 )
zerod = 0;
}
#ifndef NDEBUG
if ( fabs(tot-1.0)>1e-4 ) {
yap_message("get_probs() probs doesn't normalise, get %lf\n", tot);
}
#endif
for (t=0; t<ddN.T; t++)
vp[t] /= tot;
}
