You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

410 lines
13 KiB

//==============================================================================
//
// Copyright (c) 2002-
// Authors:
// * Dave Parker <david.parker@comlab.ox.ac.uk> (University of Oxford, formerly University of Birmingham)
//
//------------------------------------------------------------------------------
//
// This file is part of PRISM.
//
// PRISM 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 2 of the License, or
// (at your option) any later version.
//
// PRISM 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 PRISM; if not, write to the Free Software Foundation,
// Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
//
//==============================================================================
// includes
#include "PrismHybrid.h"
#include <math.h>
#include <util.h>
#include <cudd.h>
#include <dd.h>
#include <odd.h>
#include <dv.h>
#include <prism.h>
#include "sparse.h"
#include "hybrid.h"
#include "PrismHybridGlob.h"
#include "jnipointer.h"
#include <new>
// local prototypes
static void mult_rec(HDDNode *hdd, int level, int row_offset, int col_offset);
static void mult_cm(CMSparseMatrix *cmsm, int row_offset, int col_offset);
static void mult_cmsc(CMSCSparseMatrix *cmscsm, int row_offset, int col_offset);
// globals (used by local functions)
static HDDNode *zero;
static int num_levels;
static bool compact_sm;
static double *sm_dist;
static int sm_dist_shift;
static int sm_dist_mask;
static double *soln = NULL, *soln2 = NULL;
static double unif;
//------------------------------------------------------------------------------
JNIEXPORT jlong __jlongpointer JNICALL Java_hybrid_PrismHybrid_PH_1StochTransient
(
JNIEnv *env,
jclass cls,
jlong __jlongpointer tr, // trans matrix
jlong __jlongpointer od, // odd
jlong __jlongpointer in, // initial distribution (note: this will be deleted afterwards)
jlong __jlongpointer rv, // row vars
jint num_rvars,
jlong __jlongpointer cv, // col vars
jint num_cvars,
jdouble time // time bound
)
{
// cast function parameters
DdNode *trans = jlong_to_DdNode(tr); // trans matrix
ODDNode *odd = jlong_to_ODDNode(od); // odd
double *init = jlong_to_double(in); // initial distribution
DdNode **rvars = jlong_to_DdNode_array(rv); // row vars
DdNode **cvars = jlong_to_DdNode_array(cv); // col vars
// model stats
int n;
// flags
bool compact_d;
// matrix mtbdd
HDDMatrix *hddm = NULL;
HDDNode *hdd = NULL;
// vectors
double *diags = NULL, *tmpsoln = NULL, *sum = NULL;
DistVector *diags_dist = NULL;
// fox glynn stuff
FoxGlynnWeights fgw;
// timing stuff
long start1, start2, start3, stop;
double time_taken, time_for_setup, time_for_iters;
// misc
bool done;
int i, iters, num_iters;
double kb, kbt, max_diag, weight, term_crit_param_unif;
// exception handling around whole function
try {
// start clocks
start1 = start2 = util_cpu_time();
// get number of states from odd
n = odd->eoff + odd->toff;
// build hdd for matrix
PH_PrintToMainLog(env, "\nBuilding hybrid MTBDD matrix... ");
hddm = build_hdd_matrix(trans, rvars, cvars, num_rvars, odd, false);
hdd = hddm->top;
zero = hddm->zero;
num_levels = hddm->num_levels;
kb = hddm->mem_nodes;
kbt = kb;
PH_PrintToMainLog(env, "[levels=%d, nodes=%d] ", hddm->num_levels, hddm->num_nodes);
PH_PrintMemoryToMainLog(env, "[", kb, "]\n");
// add sparse matrices
PH_PrintToMainLog(env, "Adding explicit sparse matrices... ");
add_sparse_matrices(hddm, compact, false);
compact_sm = hddm->compact_sm;
if (compact_sm) {
sm_dist = hddm->dist;
sm_dist_shift = hddm->dist_shift;
sm_dist_mask = hddm->dist_mask;
}
kb = hddm->mem_sm;
kbt += kb;
PH_PrintToMainLog(env, "[levels=%d, num=%d%s] ", hddm->l_sm, hddm->num_sm, compact_sm?", compact":"");
PH_PrintMemoryToMainLog(env, "[", kb, "]\n");
// get vector of diagonals
PH_PrintToMainLog(env, "Creating vector for diagonals... ");
diags = hdd_negative_row_sums(hddm, n);
compact_d = false;
// try and convert to compact form if required
if (compact) {
if (diags_dist = double_vector_to_dist(diags, n)) {
compact_d = true;
delete[] diags; diags = NULL;
}
}
kb = (!compact_d) ? n*8.0/1024.0 : (diags_dist->num_dist*8.0+n*2.0)/1024.0;
kbt += kb;
if (compact_d) PH_PrintToMainLog(env, "[dist=%d, compact] ", diags_dist->num_dist);
PH_PrintMemoryToMainLog(env, "[", kb, "]\n");
//for(i = 0; i < n; i++) printf("%f ", (!compact_d)?(diags[i]):(diags_dist->dist[diags_dist->ptrs[i]])); printf("\n");
// find max diagonal element
if (!compact_d) {
max_diag = diags[0];
for (i = 1; i < n; i++) if (diags[i] < max_diag) max_diag = diags[i];
} else {
max_diag = diags_dist->dist[0];
for (i = 1; i < diags_dist->num_dist; i++) if (diags_dist->dist[i] < max_diag) max_diag = diags_dist->dist[i];
}
max_diag = -max_diag;
// constant for uniformization
unif = 1.02*max_diag;
last_unif = unif;
// modify diagonals
if (!compact_d) {
for (i = 0; i < n; i++) diags[i] = diags[i] / unif + 1;
} else {
for (i = 0; i < diags_dist->num_dist; i++) diags_dist->dist[i] = diags_dist->dist[i] / unif + 1;
}
// create solution/iteration vectors
PH_PrintToMainLog(env, "Allocating iteration vectors... ");
// for soln, we just use init (since we are free to modify/delete this vector)
// we also report the memory usage of this vector here, even though it has already been created
soln = init;
soln2 = new double[n];
sum = new double[n];
kb = n*8.0/1024.0;
kbt += 3*kb;
PH_PrintMemoryToMainLog(env, "[3 x ", kb, "]\n");
// print total memory usage
PH_PrintMemoryToMainLog(env, "TOTAL: [", kbt, "]\n");
// compute new termination criterion parameter (epsilon/8)
term_crit_param_unif = term_crit_param / 8.0;
// compute poisson probabilities (fox/glynn)
PH_PrintToMainLog(env, "\nUniformisation: q.t = %f x %f = %f\n", unif, time, unif * time);
fgw = fox_glynn(unif * time, 1.0e-300, 1.0e+300, term_crit_param_unif);
if (fgw.right < 0) throw "Overflow in Fox-Glynn computation (time bound too big?)";
for (i = fgw.left; i <= fgw.right; i++) {
fgw.weights[i-fgw.left] /= fgw.total_weight;
}
PH_PrintToMainLog(env, "Fox-Glynn: left = %d, right = %d\n", fgw.left, fgw.right);
// set up vectors
for (i = 0; i < n; i++) {
sum[i] = 0.0;
}
// get setup time
stop = util_cpu_time();
time_for_setup = (double)(stop - start2)/1000;
start2 = stop;
// start transient analysis
done = false;
num_iters = -1;
PH_PrintToMainLog(env, "\nStarting iterations...\n");
// if necessary, do 0th element of summation (doesn't require any matrix powers)
if (fgw.left == 0) for (i = 0; i < n; i++) {
sum[i] += fgw.weights[0] * soln[i];
}
// note that we ignore max_iters as we know how any iterations _should_ be performed
for (iters = 1; (iters <= fgw.right) && !done; iters++) {
// PH_PrintToMainLog(env, "Iteration %d: ", iters);
// start3 = util_cpu_time();
// initialise vector
if (!compact_d) {
for (i = 0; i < n; i++) soln2[i] = diags[i] * soln[i];
} else {
for (i = 0; i < n; i++) soln2[i] = diags_dist->dist[diags_dist->ptrs[i]] * soln[i];
}
// do matrix vector multiply bit
mult_rec(hdd, 0, 0, 0);
// check for steady state convergence
if (do_ss_detect) switch (term_crit) {
case TERM_CRIT_ABSOLUTE:
done = true;
for (i = 0; i < n; i++) {
if (fabs(soln2[i] - soln[i]) > term_crit_param_unif) {
done = false;
break;
}
}
break;
case TERM_CRIT_RELATIVE:
done = true;
for (i = 0; i < n; i++) {
if (fabs((soln2[i] - soln[i])/soln2[i]) > term_crit_param_unif) {
done = false;
break;
}
}
break;
}
// special case when finished early (steady-state detected)
if (done) {
// work out sum of remaining poisson probabilities
if (iters <= fgw.left) {
weight = 1.0;
} else {
weight = 0.0;
for (i = iters; i <= fgw.right; i++) {
weight += fgw.weights[i-fgw.left];
}
}
// add to sum
for (i = 0; i < n; i++) sum[i] += weight * soln2[i];
PH_PrintToMainLog(env, "\nSteady state detected at iteration %d\n", iters);
num_iters = iters;
break;
}
// prepare for next iteration
tmpsoln = soln;
soln = soln2;
soln2 = tmpsoln;
// add to sum
if (iters >= fgw.left) {
for (i = 0; i < n; i++) sum[i] += fgw.weights[iters-fgw.left] * soln[i];
}
// PH_PrintToMainLog(env, "%.2f %.2f sec\n", ((double)(util_cpu_time() - start3)/1000), ((double)(util_cpu_time() - start2)/1000)/iters);
}
// stop clocks
stop = util_cpu_time();
time_for_iters = (double)(stop - start2)/1000;
time_taken = (double)(stop - start1)/1000;
// print iters/timing info
if (num_iters == -1) num_iters = fgw.right;
PH_PrintToMainLog(env, "\nIterative method: %d iterations in %.2f seconds (average %.6f, setup %.2f)\n", num_iters, time_taken, time_for_iters/num_iters, time_for_setup);
// catch exceptions: register error, free memory
} catch (std::bad_alloc e) {
PH_SetErrorMessage("Out of memory");
if (sum) delete[] sum;
sum = 0;
} catch (const char *err) {
PH_SetErrorMessage(err);
if (sum) delete sum;
sum = 0;
}
// free memory
if (hddm) delete hddm;
if (diags) delete[] diags;
if (diags_dist) delete diags_dist;
// nb: we *do* free soln (which was originally init)
if (soln) delete[] soln;
if (soln2) delete[] soln2;
return ptr_to_jlong(sum);
}
//------------------------------------------------------------------------------
static void mult_rec(HDDNode *hdd, int level, int row_offset, int col_offset)
{
HDDNode *e, *t;
// if it's the zero node
if (hdd == zero) {
return;
}
// or if we've reached a submatrix
// (check for non-null ptr but, equivalently, we could just check if level==l_sm)
else if (hdd->sm.ptr) {
if (!compact_sm) {
mult_cm((CMSparseMatrix *)hdd->sm.ptr, row_offset, col_offset);
} else {
mult_cmsc((CMSCSparseMatrix *)hdd->sm.ptr, row_offset, col_offset);
}
return;
}
// or if we've reached the bottom
else if (level == num_levels) {
//printf("(%d,%d)=%f\n", col_offset, row_offset, hdd->type.val);
soln2[col_offset] += soln[row_offset] * (hdd->type.val / unif);
return;
}
// otherwise recurse
e = hdd->type.kids.e;
if (e != zero) {
mult_rec(e->type.kids.e, level+1, row_offset, col_offset);
mult_rec(e->type.kids.t, level+1, row_offset, col_offset+e->off.val);
}
t = hdd->type.kids.t;
if (t != zero) {
mult_rec(t->type.kids.e, level+1, row_offset+hdd->off.val, col_offset);
mult_rec(t->type.kids.t, level+1, row_offset+hdd->off.val, col_offset+t->off.val);
}
}
//-----------------------------------------------------------------------------------
static void mult_cm(CMSparseMatrix *cmsm, int row_offset, int col_offset)
{
int i2, j2, l2, h2;
int sm_n = cmsm->n;
int sm_nnz = cmsm->nnz;
double *sm_non_zeros = cmsm->non_zeros;
unsigned char *sm_col_counts = cmsm->col_counts;
int *sm_col_starts = (int *)cmsm->col_counts;
bool sm_use_counts = cmsm->use_counts;
unsigned int *sm_rows = cmsm->rows;
// loop through columns of submatrix
l2 = sm_nnz; h2 = 0;
for (i2 = 0; i2 < sm_n; i2++) {
// loop through entries in this column
if (!sm_use_counts) { l2 = sm_col_starts[i2]; h2 = sm_col_starts[i2+1]; }
else { l2 = h2; h2 += sm_col_counts[i2]; }
for (j2 = l2; j2 < h2; j2++) {
soln2[col_offset + i2] += soln[row_offset + sm_rows[j2]] * (sm_non_zeros[j2] / unif);
//printf("(%d,%d)=%f\n", col_offset + sm_rows[j2], row_offset + i2, sm_non_zeros[j2]);
}
}
}
//-----------------------------------------------------------------------------------
static void mult_cmsc(CMSCSparseMatrix *cmscsm, int row_offset, int col_offset)
{
int i2, j2, l2, h2;
int sm_n = cmscsm->n;
int sm_nnz = cmscsm->nnz;
unsigned char *sm_col_counts = cmscsm->col_counts;
int *sm_col_starts = (int *)cmscsm->col_counts;
bool sm_use_counts = cmscsm->use_counts;
unsigned int *sm_rows = cmscsm->rows;
// loop through columns of submatrix
l2 = sm_nnz; h2 = 0;
for (i2 = 0; i2 < sm_n; i2++) {
// loop through entries in this column
if (!sm_use_counts) { l2 = sm_col_starts[i2]; h2 = sm_col_starts[i2+1]; }
else { l2 = h2; h2 += sm_col_counts[i2]; }
for (j2 = l2; j2 < h2; j2++) {
soln2[col_offset + i2] += soln[row_offset + (int)(sm_rows[j2] >> sm_dist_shift)] * (sm_dist[(int)(sm_rows[j2] & sm_dist_mask)] / unif);
//printf("(%d,%d)=%f\n", col_offset + (int)(sm_rows[j2] >> sm_dist_shift), row_offset + i2, sm_dist[(int)(sm_rows[j2] & sm_dist_mask)]);
}
}
}
//------------------------------------------------------------------------------