nojhan/eodev
Archived
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

MPI Multistart: everybody loves comments, except the one who writes them.

Browse source
This commit is contained in:
Benjamin Bouvier 2012-07-26 16:01:04 +02:00
commit 61c31a4a71
2 changed files with 252 additions and 108 deletions
Download patch file Download diff file Expand all files Collapse all files

235
eo/src/mpi/eoMultiStart.h
View file

@ -4,10 +4,50 @@
# include <eo>
# include "eoMpi.h"
/**
* @ingroup MPI
* @{
*/
/**
* @file eoMultiStart.h
*
* Contains implementation of a MPI job which consists in a multi start, which basically consists in the following:
* the same eoAlgo is launched on computers of a clusters, with different seeds for each. As the eoAlgo are most of
* the time stochastics, the results won't be the same. It is fully equivalent to launch the same program but with
* different seeds.
*
* It follows the structure of a MPI job, as described in eoMpi.h. The basic algorithm is trivial:
* - Loop while we have a run to perform.
* - Worker performs runs and send their best solution (individual with best fitness) to the master.
* - Master retrieves the best solution and adds it to a eoPop of best solutions (the user can chooses what he does
* with this population, for instance: retrieve the best element, etc.)
*
* The principal concerns about this algorithm are:
* - How do we reinitialize the algorithm? An eoAlgo can have several forms, and initializations have to be performed
* before each "start". We can hence decide whether we reinits the population or keep the same population obtained
* after the previous start, we have to reinitialize continuator, etc. This is customizable in the store.
*
* - Which seeds should be chosen? If we want the run to be re-runnable with the same results, we need to be sure that
* the seeds are the same. But user can not care about this, and just want random seeds. This is customizable in the
* store.
*
* These concerns are handled by functors, inheriting from MultiStartStore<EOT>::ResetAlgo (for the first concern), and
* MultiStartStore<EOT>::GetSeeds (for the second one). There are default implementations, but there is no problem about
* specializing them or coding your own, by directly inheriting from them.
*
* @ingroup MPI
*/
namespace eo
{
namespace mpi
{
/**
* @brief Data used by the Multi Start job.
*
* This data is shared between the different Job functors. More details are given for each attribute.
*/
template< class EOT >
struct MultiStartData
{
@ -22,17 +62,49 @@ namespace eo
}
// dynamic parameters
/**
* @brief Total remaining number of runs.
*
* It's decremented as the runs are performed.
*/
int runs;
/**
* @brief eoPop of the best individuals, which are the one sent by the workers.
*/
eoPop< EOT > bests;
/**
* @brief eoPop on which the worker is working.
*/
eoPop< EOT > pop;
// static parameters
/**
* @brief Communicator, used to send and retrieve messages.
*/
bmpi::communicator& comm;
/**
* @brief Algorithm which will be performed by the worker.
*/
eoAlgo<EOT>& algo;
/**
* @brief Reset Algo functor, which defines how to reset the algo (above) before re running it.
*/
ResetAlgo& resetAlgo;
// Rank of master
int masterRank;
};
/**
* @brief Send task (master side) in the Multi Start job.
*
* It only consists in decrementing the number of runs, as the worker already have the population and
* all the necessary parameters to run the eoAlgo.
*/
template< class EOT >
class SendTaskMultiStart : public SendTaskFunction< MultiStartData< EOT > >
{
@ -41,10 +113,17 @@ namespace eo
void operator()( int wrkRank )
{
wrkRank++; // unused
--(_data->runs);
}
};
/**
* @brief Handle Response (master side) in the Multi Start job.
*
* It consists in retrieving the best solution sent by the worker and adds it to a population of best
* solutions.
*/
template< class EOT >
class HandleResponseMultiStart : public HandleResponseFunction< MultiStartData< EOT > >
{
@ -60,6 +139,12 @@ namespace eo
}
};
/**
* @brief Process Task (worker side) in the Multi Start job.
*
* Consists in resetting the algorithm and launching it on the population, then
* send the best individual (the one with the best fitness) to the master.
*/
template< class EOT >
class ProcessTaskMultiStart : public ProcessTaskFunction< MultiStartData< EOT > >
{
@ -74,6 +159,11 @@ namespace eo
}
};
/**
* @brief Is Finished (master side) in the Multi Start job.
*
* The job is finished if and only if all the runs have been performed.
*/
template< class EOT >
class IsFinishedMultiStart : public IsFinishedFunction< MultiStartData< EOT > >
{
@ -86,14 +176,41 @@ namespace eo
}
};
/**
* @brief Store for the Multi Start job.
*
* Contains the data used by the workers (algo,...) and functor to
* send the seeds.
*/
template< class EOT >
class MultiStartStore : public JobStore< MultiStartData< EOT > >
{
public:
/**
* @brief Generic functor to reset an algorithm before it's launched by
* the worker.
*
* This reset algorithm should reinits population (if necessary), continuator, etc.
*/
typedef typename MultiStartData<EOT>::ResetAlgo ResetAlgo;
/**
* @brief Generic functor which returns a vector of seeds for the workers.
*
* If this vector hasn't enough seeds to send, random ones are generated and
* sent to the workers.
*/
typedef eoUF< int, std::vector<int> > GetSeeds;
/**
* @brief Default ctor for MultiStartStore.
*
* @param algo The algorithm to launch in parallel
* @param masterRank The MPI rank of the master
* @param resetAlgo The ResetAlgo functor
* @param getSeeds The GetSeeds functor
*/
MultiStartStore(
eoAlgo<EOT> & algo,
int masterRank,
@ -104,6 +221,7 @@ namespace eo
_getSeeds( getSeeds ),
_masterRank( masterRank )
{
// Default job functors for this one.
this->_iff = new IsFinishedMultiStart< EOT >;
this->_iff->needDelete(true);
this->_stf = new SendTaskMultiStart< EOT >;
@ -114,6 +232,15 @@ namespace eo
this->_ptf->needDelete(true);
}
/**
* @brief Send new seeds to the workers before a job.
*
* Uses the GetSeeds functor given in constructor. If there's not
* enough seeds to send, random seeds are sent to the workers.
*
* @param workers Vector of MPI ranks of the workers
* @param runs The number of runs to perform
*/
void init( const std::vector<int>& workers, int runs )
{
_data.runs = runs;
@ -156,8 +283,51 @@ namespace eo
int _masterRank;
};
/**
* @brief MultiStart job, created for convenience.
*
* This is an OneShotJob, which means workers leave it along with
* the master.
*/
template< class EOT >
class MultiStart : public OneShotJob< MultiStartData< EOT > >
{
public:
MultiStart( AssignmentAlgorithm & algo,
int masterRank,
MultiStartStore< EOT > & store,
// dynamic parameters
int runs,
const std::vector<int>& seeds = std::vector<int>() ) :
OneShotJob< MultiStartData< EOT > >( algo, masterRank, store )
{
store.init( algo.idles(), runs );
}
/**
* @brief Returns the best solution, at the end of the job.
*
* Warning: if you call this function from a worker, or from the master before the
* launch of the job, you will only get an empty population!
*
* @return Population of best individuals retrieved by the master.
*/
eoPop<EOT>& best_individuals()
{
return this->store.data()->bests;
}
};
/*************************************
* DEFAULT GET SEEDS IMPLEMENTATIONS *
************************************/
/**
* @brief Uses the internal default seed generator to get seeds,
* which means: random seeds are sent.
*/
template<class EOT>
// No seeds! Use default generator
struct DummyGetSeeds : public MultiStartStore<EOT>::GetSeeds
{
std::vector<int> operator()( int n )
@ -166,8 +336,14 @@ namespace eo
}
};
/**
* @brief Sends seeds to the workers, which are multiple of a number
* given by the master. If no number is given, a random one is used.
*
* This functor ensures that even if the same store is used with
* different jobs, the seeds will be different.
*/
template<class EOT>
// Multiple of a seed
struct MultiplesOfNumber : public MultiStartStore<EOT>::GetSeeds
{
MultiplesOfNumber ( int n = 0 )
@ -196,6 +372,10 @@ namespace eo
unsigned int _i;
};
/**
* @brief Returns random seeds to the workers. We can controle which seeds are generated
* by precising the seed of the master.
*/
template<class EOT>
struct GetRandomSeeds : public MultiStartStore<EOT>::GetSeeds
{
@ -215,6 +395,17 @@ namespace eo
}
};
/**************************************
* DEFAULT RESET ALGO IMPLEMENTATIONS *
**************************************
/**
* @brief For a Genetic Algorithm, reinits the population by copying the original one
* given in constructor, and reinits the continuator.
*
* The evaluator should also be given, as the population needs to be evaluated
* before each run.
*/
template<class EOT>
struct ReuseOriginalPopEA: public MultiStartStore<EOT>::ResetAlgo
{
@ -231,7 +422,7 @@ namespace eo
void operator()( eoPop<EOT>& pop )
{
pop = _originalPop;
pop = _originalPop; // copies the original population
for(unsigned i = 0, size = pop.size(); i < size; ++i)
{
_eval( pop[i] );
@ -245,6 +436,16 @@ namespace eo
eoEvalFunc<EOT>& _eval;
};
/**
* @brief For a Genetic Algorithm, reuses the same population without
* modifying it after a run.
*
* This means, if you launch a run after another one, you'll make evolve
* the same population.
*
* The evaluator should also be sent, as the population needs to be evaluated
* at the first time.
*/
template< class EOT >
struct ReuseSamePopEA : public MultiStartStore<EOT>::ResetAlgo
{
@ -279,32 +480,12 @@ namespace eo
eoCountContinue<EOT>& _continuator;
eoPop<EOT> _originalPop;
bool _firstTime;
};
template< class EOT >
class MultiStart : public OneShotJob< MultiStartData< EOT > >
{
public:
MultiStart( AssignmentAlgorithm & algo,
int masterRank,
MultiStartStore< EOT > & store,
// dynamic parameters
int runs,
const std::vector<int>& seeds = std::vector<int>() ) :
OneShotJob< MultiStartData< EOT > >( algo, masterRank, store )
{
store.init( algo.idles(), runs );
}
eoPop<EOT>& best_individuals()
{
return this->store.data()->bests;
}
};
} // namespace mpi
} // namespace eo
/**
* @}
*/
# endif // __EO_MULTISTART_H__

103
eo/test/mpi/t-mpi-multistart.cpp
View file

@ -8,9 +8,24 @@ using namespace eo::mpi;
#include <eo>
#include <es.h>
// Use functions from namespace std
/*
* This file is based on the tutorial lesson 1. We'll consider that you know all the EO
* related parts of the algorithm and we'll focus our attention on parallelization.
*
* This file shows an example of multistart applied to a eoSGA (simple genetic
* algorithm). As individuals need to be serialized, we implement a class inheriting
* from eoReal (which is the base individual), so as to manipulate individuals as they
* were eoReal AND serialize them.
*
* The main function shows how to launch a multistart job, with default functors. If you
* don't know which functors to use, these ones should fit the most of your purposes.
*/
using namespace std;
/*
* eoReal is a vector of double: we just have to serializes the value and the fitness.
*/
class SerializableEOReal: public eoReal<double>, public eoserial::Persistent
{
public:
@ -91,99 +106,48 @@ int main(int argc, char **argv)
const double EPSILON = 0.01; // range for real uniform mutation
const float MUT_RATE = 0.5; // mutation rate
// GENERAL
//////////////////////////
// Random seed
//////////////////////////
//reproducible random seed: if you don't change SEED above,
// you'll aways get the same result, NOT a random run
// rng.reseed(SEED);
// EVAL
/////////////////////////////
// Fitness function
////////////////////////////
// Evaluation: from a plain C++ fn to an EvalFunc Object
eoEvalFuncPtr<Indi> eval( real_value );
// INIT
////////////////////////////////
// Initilisation of population
////////////////////////////////
// declare the population
eoPop<Indi> pop;
// fill it!
/*
for (unsigned int igeno=0; igeno<POP_SIZE; igeno++)
{
Indi v; // void individual, to be filled
for (unsigned ivar=0; ivar<VEC_SIZE; ivar++)
{
double r = 2*rng.uniform() - 1; // new value, random in [-1,1)
v.push_back(r); // append that random value to v
}
eval(v); // evaluate it
pop.push_back(v); // and put it in the population
}
*/
eoUniformGenerator< double > generator;
eoInitFixedLength< Indi > init( VEC_SIZE, generator );
// eoInitAndEval< Indi > init( real_init, eval, continuator );
pop = eoPop<Indi>( POP_SIZE, init );
// ENGINE
/////////////////////////////////////
// selection and replacement
////////////////////////////////////
// SELECT
// The robust tournament selection
eoDetTournamentSelect<Indi> select(T_SIZE); // T_SIZE in [2,POP_SIZE]
// REPLACE
// eoSGA uses generational replacement by default
// so no replacement procedure has to be given
// OPERATORS
//////////////////////////////////////
// The variation operators
//////////////////////////////////////
// CROSSOVER
// offspring(i) is a linear combination of parent(i)
eoDetTournamentSelect<Indi> select(T_SIZE);
eoSegmentCrossover<Indi> xover;
// MUTATION
// offspring(i) uniformly chosen in [parent(i)-epsilon, parent(i)+epsilon]
eoUniformMutation<Indi> mutation(EPSILON);
// STOP
// CHECKPOINT
//////////////////////////////////////
// termination condition
/////////////////////////////////////
// stop after MAX_GEN generations
eoGenContinue<Indi> continuator(MAX_GEN);
/* Does work too with a steady fit continuator. */
// eoSteadyFitContinue< Indi > continuator( 10, 50 );
// GENERATION
/////////////////////////////////////////
// the algorithm
////////////////////////////////////////
// standard Generational GA requires
// selection, evaluation, crossover and mutation, stopping criterion
eoSGA<Indi> gga(select, xover, CROSS_RATE, mutation, MUT_RATE,
eval, continuator);
/* How to assign tasks, which are starts? */
DynamicAssignmentAlgorithm assignmentAlgo;
/* Before a worker starts its algorithm, how does it reinits the population?
* There are a few default usable functors, defined in eoMultiStart.h.
*
* This one (ReuseSamePopEA) doesn't modify the population after a start, so
* the same population is reevaluated on each multistart: the solution tend
* to get better and better.
*/
ReuseSamePopEA< Indi > resetAlgo( continuator, pop, eval );
/**
* How to send seeds to the workers, at the beginning of the parallel job?
* This functors indicates that seeds should be random values.
*/
GetRandomSeeds< Indi > getSeeds( SEED );
// Builds the store
MultiStartStore< Indi > store(
gga,
DEFAULT_MASTER,
resetAlgo,
getSeeds);
// Creates the multistart job and runs it.
// The last argument indicates that we want to launch 5 runs.
MultiStart< Indi > msjob( assignmentAlgo, DEFAULT_MASTER, store, 5 );
msjob.run();
@ -202,5 +166,4 @@ int main(int argc, char **argv)
std::cout << "Global best individual has fitness " << msjob10.best_individuals().best_element().fitness() << std::endl;
}
return 0;
}