1
0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-10 19:36:56 +01:00

Improved the gamma quite a bit.

Serial rng's which are set on node zero and broadcaste
This commit is contained in:
Peter Boyle
2015-04-24 20:21:40 +01:00
parent 2a67214f9d
commit fc32450360
11 changed files with 151 additions and 28 deletions

View File

@ -26,7 +26,70 @@ namespace Grid {
}
};
class GridRNG {
class GridRNGbase {
public:
GridRNGbase() : _uniform{0,1}, _gaussian(0.0,1.0) {};
int _seeded;
// One generator per site.
// Uniform and Gaussian distributions from these generators.
std::vector<std::ranlux48> _generators;
std::uniform_real_distribution<double> _uniform;
std::normal_distribution<double> _gaussian;
};
class GridSerialRNG : public GridRNGbase {
public:
// FIXME ... do we require lockstep draws of randoms
// from all nodes keeping seeds consistent.
// place a barrier/broadcast in the fill routine
template<class source> void Seed(source &src)
{
typename source::result_type init = src();
CartesianCommunicator::BroadcastWorld(0,(void *)&init,sizeof(init));
_generators[0] = std::ranlux48(init);
_seeded=1;
}
GridSerialRNG() : GridRNGbase() {
_generators.resize(1);
_seeded=0;
}
template <class sobj,class distribution> inline void fill(sobj &l,distribution &dist){
typedef typename sobj::scalar_type scalar_type;
int words = sizeof(sobj)/sizeof(scalar_type);
scalar_type *buf = (scalar_type *) & l;
for(int idx=0;idx<words;idx++){
buf[idx] = dist(_generators[0]);
}
CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
};
void SeedRandomDevice(void){
std::random_device rd;
Seed(rd);
}
void SeedFixedIntegers(std::vector<int> &seeds){
fixedSeed src(seeds);
Seed(src);
}
};
class GridParallelRNG : public GridRNGbase {
public:
// One generator per site.
std::vector<std::ranlux48> _generators;
@ -42,25 +105,13 @@ namespace Grid {
return is*_grid->oSites()+os;
}
GridRNG(GridBase *grid) : _uniform{0,1}, _gaussian(0.0,1.0) {
GridParallelRNG(GridBase *grid) : GridRNGbase() {
_grid=grid;
_vol =_grid->iSites()*_grid->oSites();
_generators.resize(_vol);
// SeedFixedIntegers(seeds);
// worst case we seed properly but non-deterministically
SeedRandomDevice();
_seeded=0;
}
// FIXME: drive seeding from node zero and transmit to all
// to get unique randoms on each node
void SeedRandomDevice(void){
std::random_device rd;
Seed(rd);
}
void SeedFixedIntegers(std::vector<int> &seeds){
fixedSeed src(seeds);
Seed(src);
}
// This loop could be made faster to avoid the Ahmdahl by
// i) seed generators on each timeslice, for x=y=z=0;
@ -86,6 +137,7 @@ namespace Grid {
_generators[l_idx] = std::ranlux48(init);
}
}
_seeded=1;
}
//FIXME implement generic IO and create state save/restore
@ -122,15 +174,34 @@ namespace Grid {
merge(l._odata[ss],pointers);
}
};
void SeedRandomDevice(void){
std::random_device rd;
Seed(rd);
}
void SeedFixedIntegers(std::vector<int> &seeds){
fixedSeed src(seeds);
Seed(src);
}
};
// FIXME Implement a consistent seed management strategy
template <class vobj> inline void random(GridRNG &rng,Lattice<vobj> &l){
template <class vobj> inline void random(GridParallelRNG &rng,Lattice<vobj> &l){
rng.fill(l,rng._uniform);
}
template <class vobj> inline void gaussian(GridRNG &rng,Lattice<vobj> &l){
template <class vobj> inline void gaussian(GridParallelRNG &rng,Lattice<vobj> &l){
rng.fill(l,rng._gaussian);
}
template <class sobj> inline void random(GridSerialRNG &rng,sobj &l){
rng.fill(l,rng._uniform);
}
template <class sobj> inline void gaussian(GridSerialRNG &rng,sobj &l){
rng.fill(l,rng._gaussian);
}
}
#endif