Skip to content

FIC_DK_P GA Solving
FIC_DK_P GA Solving
Commit 46fa9dea authored by HUMEAU Jéremie's avatar HUMEAU Jéremie
Browse files
Options

cluster run

parent fdebfb3f
Show whitespace changes
Inline Side-by-side
Showing with 532 additions and 3 deletions
.gitignore
View file @ 46fa9dea
......@@ -3,4 +3,4 @@ build/
howto.pdf
.vscode/
.cmake/
results/
CMakeLists.txt
View file @ 46fa9dea
......@@ -49,6 +49,13 @@ include_directories(${PARADISEO_INCLUDE_DIR})
# the user should choose the build type on windows environments,excepted under cygwin (default=none)
set(CMAKE_DEFAULT_BUILD_TYPE "Release" CACHE STRING "Variable that stores the default CMake build type" FORCE)
add_definitions(-DDEPRECATED_MESSAGES)
set(CMAKE_CXX_FLAGS_DEBUG "-Wunknown-pragmas -O0 -g -fPIC -Wall -Wextra -ansi -pedantic -std=c++11" CACHE STRING "" FORCE)
set(CMAKE_CXX_FLAGS_RELEASE "-Wunknown-pragmas -O2 -fPIC -std=c++11" CACHE STRING "" FORCE)
find_program(MEMORYCHECK_COMMAND
NAMES purify valgrind
PATHS
......
application/main.cpp
View file @ 46fa9dea
......@@ -272,6 +272,12 @@ int main(int argc, char* argv[]){
// ================================= RUN ZONE ===============================
std::string filename=_outputFile + "/bestsol.";
std::ostringstream oss;
oss << _seedParam;
filename+=oss.str();
std::ofstream outputfile(filename.c_str(), std::ios::app);
std::cout << "===== CURRENT POP =====" << std::endl;
pop.best_element().printOn(std::cout);
std::cout << " fitness:" << pop.best_element().fitness() << std::endl;
......@@ -314,6 +320,11 @@ int main(int argc, char* argv[]){
eoEasyEA<QUEEN> algo(cont,eval,select,transform,replace);
algo(pop);
pop.best_element().printOn(outputfile);
outputfile.close();
// -------------------
std::cout << "\n===== BEST INDIVIDU =====" << std::endl;
......
application/rmnkDLHV.cpp 0 → 100644
View file @ 46fa9dea
using namespace std;
// temp
#include "moBitNeighbor.h"
// paradiseo
#include <eo>
#include <ga.h>
#include <mo>
#include <problems/bitString/moBitNeighbor.h>
#include <moeo>
#include <utils/moeoFullEvalByCopy.h>
// problem
#include<rmnkEval.h>
//MPI
#include <mpi.h>
// checks for help demand, and writes the status file and make_help; in libutils
void make_help(eoParser & _parser);
// objective vector of real values
typedef moeoRealObjectiveVector < moeoObjectiveVectorTraits > rmnkObjectiveVector;
// problem representation
typedef moeoBitVector < rmnkObjectiveVector > RMNK;
// neighbor
typedef moBitNeighbor<RMNK,rmnkObjectiveVector> rmnkNeighbor;
// go
int main(int argc, char* argv[]){
try{
/*** init ***/
eoParser parser(argc, argv); // for user-parameter reading
eoState state; // to keep all things allocated
// parameters
eoValueParam<uint32_t>& _seedParam = parser.createParam(uint32_t(0), "seed", "Random number seed", 'S');
std::string _inputFile = parser.createParam(std::string(), "inputfile", "", '\0', "Representation", true).value();
std::string _outputFile = parser.createParam(std::string("/scratch"), "outputfile", "", '\0', "Representation", true).value();
unsigned int PAS = parser.createParam((unsigned int)(1), "pas", "Nb of iteration between 2 print",'p',"Param").value();
unsigned int BEGIN = parser.createParam((unsigned int)(0), "begin", "Nb of iteration between 2 print",'p',"Param").value();
unsigned int _popSize = parser.createParam((unsigned int)(100), "popSize", "Population size", 'P', "Evolution Engine").value();
unsigned int NBCORES = parser.createParam((unsigned int)(32), "nbCores", "Number of cores used",'C',"Param").value();
std::string STRAT = parser.createParam(std::string(""), "strat", "Strategy for the choice of neighbors",'S',"Param").value();
unsigned int NBMUT = parser.createParam((unsigned int)(100), "nbMut", "Number of mutation",'X',"Param").value();
double _pMut = parser.createParam((double)(1.0), "pMut", "mut rate", 'M', "Evolution Engine").value();
unsigned int NBITER = parser.createParam((unsigned int)(100), "nbIter", "Number of iteration",'I',"Param").value();
//for MPI
int numtasks, rank, tag1=1, tag2=2, prev, next;
MPI_Request reqs[4];
MPI_Status stats[4];
double tmp, max, x, y, newx=1.0, newy=1.0, lambda0, lambda1, xref=0, yref=0;
// seed
if (_seedParam.value() == 0) _seedParam.value() = time(0);
rng.reseed(_seedParam.value());
//copy of objective vector for MPI
double front[2][NBCORES];
// eval function
rmnkEval<RMNK,rmnkObjectiveVector> fEval(_inputFile.c_str());
eoEvalFuncCounter<RMNK> eval(fEval);
// number of objectives
std::vector <bool> bObj(fEval.getM(), false);
moeoObjectiveVectorTraits::setup(fEval.getM(), bObj);
// initialiser
eoUniformGenerator<bool> uGen;
eoInitFixedLength<RMNK> init(fEval.getN(),uGen);
// mutation (prob = _pMut/N)
eoBitMutation<RMNK> mut (_pMut, true);
// generate initial population
RMNK sol;
moeoUnboundedArchive<RMNK> archive;
moeoUnboundedArchive<RMNK> finalArchive;
while(archive.size()<NBCORES){
init(sol);
eval(sol);
archive.push_back(sol);
}
//Print initial population
/*
std::cout << "POINTS INITIALS" << std::endl;
for (unsigned int i=0; i<archive.size(); i++){
cout << archive[i].objectiveVector() << " " << arch[i].fitness() << endl;
}
*/
// Copie des valeurs des objectifs
for(int i=0; i<NBCORES; i++){
front[0][i]=archive[i].objectiveVector()[0];
front[1][i]=archive[i].objectiveVector()[1];
}
// stockage local des objectif vector sur les coeurs P, P+1, P-1
double local[3][2];
//Init MPI
MPI_Init(&argc,&argv);
MPI_Comm_rank(MPI_COMM_WORLD, &rank);
MPI_Comm_size(MPI_COMM_WORLD, &numtasks);
//On repartie les valeurs sur chaque coeurs
MPI_Scatter(&front[0][0],1,MPI_DOUBLE,&local[0][0],1,MPI_DOUBLE,1,MPI_COMM_WORLD);
MPI_Scatter(&front[1][0],1,MPI_DOUBLE,&local[0][1],1,MPI_DOUBLE,1,MPI_COMM_WORLD);
//Print du front de dpart
//std::cout << rank << " " << local[0][0] << " " << local[0][1] << std::endl;
//define the neighbor (P+1 et P-1)
prev=rank-1;
next=rank+1;
// Keep best
RMNK best;
max=0;
// main loop
for(unsigned int i=0;i<NBITER; i++){
// Communication with prev and next
if(rank==0){
//prev n'existe pas
local[2][0]=0;
local[2][1]=0;
//envoie des valeurs objectif au next
MPI_Isend(&local[0][0], 2, MPI_DOUBLE, next, tag1, MPI_COMM_WORLD, &reqs[0]);
//Reception des valeurs objectif du next
MPI_Irecv(&local[1][0], 2, MPI_DOUBLE, next, tag2, MPI_COMM_WORLD, &reqs[1]);
MPI_Waitall(2, reqs, stats);
}//endif(rank==0)
else if(rank==(NBCORES-1)){
//next n'existe pas
local[1][0]=0;
local[1][1]=0;
//envoie des valeurs objectif au prev
MPI_Isend(&local[0][0], 2, MPI_DOUBLE, prev, tag2, MPI_COMM_WORLD, &reqs[0]);
//Reception des valeurs objectif du prev
MPI_Irecv(&local[2][0], 2, MPI_DOUBLE, prev, tag1, MPI_COMM_WORLD, &reqs[1]);
MPI_Waitall(2, reqs, stats);
}//endif(rank==NBCORES-1)
else{
//envoie des valeurs objectif au next
MPI_Isend(&local[0][0], 2, MPI_DOUBLE, next, tag1, MPI_COMM_WORLD, &reqs[0]);
//Reception des valeurs objectif du next
MPI_Irecv(&local[1][0], 2, MPI_DOUBLE, next, tag2, MPI_COMM_WORLD, &reqs[1]);
//envoie des valeurs objectif au prev
MPI_Isend(&local[0][0], 2, MPI_DOUBLE, prev, tag2, MPI_COMM_WORLD, &reqs[2]);
//Reception des valeurs objectif du prev
MPI_Irecv(&local[2][0], 2, MPI_DOUBLE, prev, tag1, MPI_COMM_WORLD, &reqs[3]);
MPI_Waitall(4, reqs, stats);
}
// find a new solution
bool better=false;
//on calcul le score de la solution actuelle par rapport ces nouveaux voisins
if(rank!=0 && rank!=(NBCORES-1)){
x=archive[rank].objectiveVector()[0];
y=archive[rank].objectiveVector()[1];
if(STRAT=="hyp"){
// si on apporte rien l'hypervolume on prend une direction ortogonal par rapport aux voisins
if((x<=local[2][0]) || (y<=local[1][1])){
lambda0=std::abs(local[1][1]-local[2][1]);
lambda1=std::abs(local[1][0]-local[2][0]);
tmp=(lambda0*x + lambda1*y) * 0.000001;
}
// sinon on prend l'hypervolume
else{
tmp=(x-local[2][0]) * (y-local[1][1]);
}
}
else if(STRAT=="ortho"){
lambda0=std::abs(local[1][1]-local[2][1]);
lambda1=std::abs(local[1][0]-local[2][0]);
tmp=lambda0*x + lambda1*y;
}
max=tmp;
}
else{
max=-2;
}
// mutations 1/l
for(unsigned int j=0; j<NBMUT; j++){
// copie de la solution
RMNK tmpSol;
tmpSol=archive[rank];
double tmp1;
double tmp2;
// mutation de la copie
mut(tmpSol);
// on invalide la copie
tmpSol.invalidate();
// on evalue les valeurs objectifs de la copie
eval(tmpSol);
//on copie ces valeurs dans x et y (plus simple a ecrire quand mm)
x=tmpSol.objectiveVector()[0];
y=tmpSol.objectiveVector()[1];
// extreme
if(rank==0){
if((y >local[0][1]) || ((y==local[0][1])&& (x>local[0][0]))){
if(y > local[0][1])
tmp=y;
else
tmp=-1*1/x;
if(i==0 || tmp>max){
max=tmp;
best=tmpSol;
newx=x;
newy=y;
better=true;
}
}
}
else if(rank==(NBCORES-1)){
if((x >local[0][0]) || ((x ==local[0][0]) && (y >local[0][0])) ){
if(x >local[0][0])
tmp=x;
else
tmp=-1*1/y;
if(i==0 || tmp>max){
max=tmp;
best=tmpSol;
newx=x;
newy=y;
better=true;
}
}
}
//cas gnral
else{
if(STRAT=="hyp"){
// si on apporte rien l'hypervolume on prend une direction ortogonal par rapport aux voisins
if((x<=local[2][0]) || (y<=local[1][1])){
lambda0=std::abs(local[1][1]-local[2][1]);
lambda1=std::abs(local[1][0]-local[2][0]);
tmp=(lambda0*x + lambda1*y) * 0.000001;
}
// sinon on prend l'hypervolume
else{
tmp=(x-local[2][0]) * (y-local[1][1]);
}
}
else if(STRAT=="ortho"){
lambda0=std::abs(local[1][1]-local[2][1]);
lambda1=std::abs(local[1][0]-local[2][0]);
tmp=lambda0*x + lambda1*y;
}
if(tmp>max){
// si on a trouv une meilleur solution qu'avant on la sauvegarde
better=true;
max=tmp;
best=tmpSol;
newx=x;
newy=y;
}
}
}
// on affecte la meilleur solution trouv
if(better){
local[0][0]=newx;
local[0][1]=newy;
archive[rank]=best;
}
//on attend tout le monde
MPI_Barrier(MPI_COMM_WORLD);
//on save le front rgulierement
if((i%PAS==0 && i>=BEGIN) || i==NBITER-1){
std::string filename=_outputFile + "/rank.";
std::ostringstream oss;
if(rank<10)
oss << "00" << rank;
else if(rank<100)
oss << "0" << rank;
else
oss << rank;
filename+=oss.str();
std::ofstream outputfile(filename.c_str(), std::ios::app);
outputfile << local[0][0] << " " << local[0][1] << std::endl;
outputfile.close();
}
}
std::string filename=_outputFile + "/finalpop.";
std::ostringstream oss;
oss << rank;
filename+=oss.str();
std::ofstream outputfile(filename.c_str(), std::ios::app);
archive[rank].printOn(outputfile);
outputfile << std::endl;
outputfile.close();
filename=_outputFile + "/finalpoint.";
oss << rank;
filename+=oss.str();
std::ofstream outputfile2(filename.c_str(), std::ios::app);
outputfile2 << archive[rank].objectiveVector()[0] << " " << archive[rank].objectiveVector()[1] << std::endl;
outputfile2.close();
make_help(parser);
MPI_Finalize();
}
catch (exception& e){
cout << e.what() << endl;
}
return EXIT_SUCCESS;
}
cmake/module/FindParadiseo.cmake
View file @ 46fa9dea
......@@ -48,6 +48,7 @@ set(BUILD_DIR build)
# Path
set(PARADISEO_SRC_PATHS
/home/jeremie.humeau/private/ParadisEO-2.0/
/home/jerem/Softs/ParadisEO-2.0/
${PARADISEO_ROOT}
$ENV{PARADISEO_ROOT}
......@@ -102,6 +103,7 @@ endif()
set(PARADISEO_FOUND true) # will be set to false if one of the required modules is not found
set(FIND_PARADISEO_LIB_PATHS
/home/jeremie.humeau/private/ParadisEO-2.0/build
/home/jerem/Softs/ParadisEO-2.0/build/
${PARADISEO_ROOT}/${BUILD_DIR}
$ENV{PARADISEO_ROOT}
......@@ -163,3 +165,4 @@ else()
# include directory or library not found
message(FATAL_ERROR "Could NOT find ParadisEO (missing : ${FIND_PARADISEO_MISSING})")
endif()
run/job.sh 0 → 100644
View file @ 46fa9dea
#!/bin/sh
#SBATCH -o main.out
#SBATCH -N 1
#SBATCH -n 1
#SBATCH --mem=10G
#SBATCH -p defq
/home/jeremie.humeau/private/gitProject/csdvp-evolutionary-algorithm-optimization/build/application/ceao -B=50 -S=$S -C=$C -c=$c -A=$A --outputfile=$O -n=$A
exit 0
run/job.sh~ 0 → 100644
View file @ 46fa9dea
#!/bin/sh
#SBATCH -o main.out
#SBATCH -N 1
#SBATCH -n 1
#SBATCH --mem=10G
#SBATCH -p defq
/home/jeremie.humeau/private/gitProject/csdvp-evolutionary-algorithm-optimization/build/application/ceao -B=50 -S=$S -C=$C -c=$c -A=$A --outputfile=$O
exit 0
run/startExp.sh 0 → 100644
View file @ 46fa9dea
#!/bin/bash
for nbCours in 15
do
for nbComps in 10 20
do
for cbyTF in 2
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
# sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
for nbCours in 25
do
for nbComps in 10
do
for cbyTF in 3
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
# sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
for nbCours in 20
do
for nbComps in 20
do
for cbyTF in 3
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
# sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
for nbCours in 30
do
for nbComps in 40
do
for cbyTF in 4
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
# sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
exit 0
run/startExp.sh~ 0 → 100644
View file @ 46fa9dea
#!/bin/bash
for nbCours in 15
do
for nbComps in 10 20
do
for cbyTF in 2
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
# sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
for nbCours in 25
do
for nbComps in 10
do
for cbyTF in 3
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
# sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
for nbCours in 20
do
for nbComps in 20
do
for cbyTF in 3
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
# sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
for nbCours in 30
do
for nbComps in 40
do
for cbyTF in 4
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
mkdir $RESDIR
for run in 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
do
RESDIR="${nbCours}_${nbComps}_${cbyTF}"
# sbatch --job-name=${RESDIR}_${run} --output=${RESDIR}_${run}.out --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
sbatch --job-name=${RESDIR}_${run} --export=C=$nbCours,c=$nbComps,A=$cbyTF,O=$RESDIR,S=$run job.sh
done
done
done
done
exit 0
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment