paradiseo/eo/contrib/irace/fastga.cpp
nojhan 02eb0e967d fix eoAlgoFoundry management of numeric parameters
- Use a variant to avoid implicit casting to integer when selecting with brace-initialization.
- Add more doc around parameter forges.
2022-01-26 10:40:50 +01:00

689 lines
27 KiB
C++

#include <filesystem>
#include <iostream>
#include <cstdlib>
#include <string>
#include <memory>
#include <eo>
#include <ga.h>
#include <utils/checkpointing>
#include <eoInt.h>
#include <problems/eval/eoEvalIOH.h>
#include <ioh.hpp>
/*****************************************************************************
* ParadisEO algorithmic grammar definition.
*****************************************************************************/
using Bits = eoBit<eoMaximizingFitnessT<int>, int>;
// by enumerating candidate operators and their parameters.
eoAlgoFoundryFastGA<Bits>& make_foundry(
eoFunctorStore& store,
eoInit<Bits>& init,
eoEvalFunc<Bits>& eval,
const size_t max_evals,
const size_t generations,
const double optimum
)
{
// FIXME using max_restarts>1 does not allow to honor max evals.
auto& foundry = store.pack< eoAlgoFoundryFastGA<Bits> >(init, eval, max_evals, /*max_restarts=*/1);
/***** Continuators ****/
auto& fitcont = store.pack< eoFitContinue<Bits> >(optimum);
auto& gencont = store.pack< eoGenContinue<Bits> >(generations);
auto combconts = std::make_shared< std::vector<eoContinue<Bits>*> >();
combconts->push_back( &fitcont );
combconts->push_back( &gencont );
foundry.continuators.add< eoCombinedContinue<Bits> >( *combconts );
// for(size_t i=1; i<10; i++) {
// foundry.continuators.add< eoGenContinue<Bits> >(i);
// }
// for(size_t i=10; i < 100; i+=2 ) {
// foundry.continuators.add< eoSteadyFitContinue<Bits> >(10,i);
// }
// for(double i=0.0; i<1.0; i+=0.2) {
// foundry.crossover_rates.add<double>(i);
// foundry.mutation_rates.add<double>(i);
// }
/***** Offsprings size *****/
// for(size_t i=5; i<100; i+=10) {
// foundry.offspring_sizes.add<size_t>(i);
// }
foundry.offspring_sizes.setup(0,100); // 0 = use parents fixed pop size.
/***** Crossovers ****/
for(double i=0.1; i<1.0; i+=0.2) {
foundry.crossovers.add< eoUBitXover<Bits> >(i); // preference over 1
}
for(size_t i=1; i < 10; i+=2) {
foundry.crossovers.add< eoNPtsBitXover<Bits> >(i); // nb of points
}
// foundry.crossovers.add< eo1PtBitXover<Bits> >(); // Same as NPts=1
/***** Mutations ****/
double p = 1.0; // Probability of flipping each bit.
// proba of flipping k bits, k drawn in uniform distrib
foundry.mutations.add< eoUniformBitMutation<Bits> >(p);
// proba of flipping k bits, k drawn in binomial distrib
foundry.mutations.add< eoStandardBitMutation<Bits> >(p);
// proba of flipping k bits, k drawn in binomial distrib, minus zero
foundry.mutations.add< eoConditionalBitMutation<Bits> >(p);
// proba of flipping k bits, k drawn in binomial distrib, changing zeros to one
foundry.mutations.add< eoShiftedBitMutation<Bits> >(p);
// proba of flipping k bits, k drawn in normal distrib
foundry.mutations.add< eoNormalBitMutation<Bits> >(p);
// proba of flipping k bits, k drawn in powerlaw distrib
foundry.mutations.add< eoFastBitMutation<Bits> >(p);
for(size_t i=1; i < 11; i+=2) {
// mutate k bits without duplicates
foundry.mutations.add< eoDetSingleBitFlip<Bits> >(i);
}
/***** Selectors *****/
for(eoOperatorFoundry<eoSelectOne<Bits>>& ops :
{std::ref(foundry.crossover_selectors),
std::ref(foundry.mutation_selectors) }) {
ops.add< eoRandomSelect<Bits> >();
ops.add< eoStochTournamentSelect<Bits> >(0.5);
ops.add< eoSequentialSelect<Bits> >();
ops.add< eoProportionalSelect<Bits> >();
for(size_t i=2; i < 11; i+=4) {
ops.add< eoDetTournamentSelect<Bits> >(i);
}
}
foundry.aftercross_selectors.add< eoRandomSelect<Bits> >();
/***** Replacements ****/
foundry.replacements.add< eoPlusReplacement<Bits> >();
foundry.replacements.add< eoCommaReplacement<Bits> >();
foundry.replacements.add< eoSSGAWorseReplacement<Bits> >();
for(double i=0.51; i<0.92; i+=0.2) {
foundry.replacements.add< eoSSGAStochTournamentReplacement<Bits> >(i);
}
for(size_t i=2; i < 11; i+=2) {
foundry.replacements.add< eoSSGADetTournamentReplacement<Bits> >(i);
}
return foundry;
}
/*****************************************************************************
* irace helper functions.
*****************************************************************************/
Bits::Fitness fake_func(const Bits&) { return 0; }
void print_irace_categorical(const eoParam& param, const size_t slot_size, std::string type="c", std::ostream& out = std::cout)
{
// If there is no choice to be made on this operator, comment it out.
if(slot_size - 1 <= 0) {
out << "# ";
}
// irace doesn't support "-" in names.
std::string irace_name = param.longName();
irace_name.erase(std::remove(irace_name.begin(), irace_name.end(), '-'), irace_name.end());
out << irace_name
<< "\t\"--" << param.longName() << "=\""
<< "\t" << type;
out << "\t(0";
for(size_t i=1; i<slot_size; ++i) {
out << "," << i;
}
out << ")" << std::endl;
}
template<class T>
void print_irace_ranged(const eoParam& param, const T min, const T max, std::string type="r", std::ostream& out = std::cout)
{
// If there is no choice to be made on this operator, comment it out.
if(max - min <= 0) {
out << "# ";
}
// irace doesn't support "-" in names.
std::string irace_name = param.longName();
irace_name.erase(std::remove(irace_name.begin(), irace_name.end(), '-'), irace_name.end());
out << irace_name
<< "\t\"--" << param.longName() << "=\""
<< "\t" << type;
if(max-min <= 0) {
out << "\t(?)";
} else {
out << "\t(" << min << "," << max << ")";
}
out << std::endl;
}
template<class ITF>
void print_irace_oper(const eoParam& param, const eoOperatorFoundry<ITF>& op_foundry, std::ostream& out = std::cout)
{
print_irace_categorical(param, op_foundry.size(), "c", out);
}
// FIXME generalize to any scalar type with enable_if
// template<class ITF>
void print_irace_param(
const eoParam& param,
// const eoParameterFoundry<typename std::enable_if< std::is_floating_point<ITF>::value >::type>& op_foundry,
const eoParameterFoundry<double>& op_foundry,
std::ostream& out)
{
print_irace_ranged(param, op_foundry.min(), op_foundry.max(), "r", out);
}
// template<class ITF>
void print_irace_param(
const eoParam& param,
// const eoParameterFoundry<typename std::enable_if< std::is_integral<ITF>::value >::type>& op_foundry,
const eoParameterFoundry<size_t>& op_foundry,
std::ostream& out)
{
print_irace_ranged(param, op_foundry.min(), op_foundry.max(), "i", out);
}
template<class ITF>
void print_irace(const eoParam& param, const eoOperatorFoundry<ITF>& op_foundry, std::ostream& out = std::cout)
{
print_irace_oper<ITF>(param, op_foundry, out);
}
template<class ITF>
void print_irace(const eoParam& param, const eoParameterFoundry<ITF>& op_foundry, std::ostream& out = std::cout)
{
print_irace_param/*<ITF>*/(param, op_foundry, out);
}
void print_irace(const eoParam& param, const size_t min, const size_t max, std::ostream& out = std::cout)
{
print_irace_ranged(param, min, max, "i", out);
}
void print_operator_typed(const eoFunctorBase& op, std::ostream& out)
{
out << op.className();
}
void print_operator_typed(const double& op, std::ostream& out)
{
out << op;
}
template<class ITF>
void print_operators(const eoParam& param, eoOperatorFoundry<ITF>& op_foundry, std::ostream& out = std::cout, std::string indent=" ")
{
out << indent << op_foundry.size() << " " << param.longName() << ":" << std::endl;
for(size_t i=0; i < op_foundry.size(); ++i) {
out << indent << indent << i << ": ";
auto& op = op_foundry.instantiate(i);
print_operator_typed(op, out);
out << std::endl;
}
}
template<class T>
void print_operators(const eoParam& param, T min, T max, std::ostream& out = std::cout, std::string indent=" ")
{
out << indent << "[" << min << "," << max << "] " << param.longName() << "." << std::endl;
}
template<class ITF>
void print_operators(const eoParam& param, eoParameterFoundry<ITF>& op_foundry, std::ostream& out = std::cout, std::string indent=" ")
{
print_operators(param, op_foundry.min(), op_foundry.max(), out, indent);
}
// Problem configuration.
struct Problem {
double dummy;
size_t epistasis;
size_t neutrality;
size_t ruggedness;
size_t max_target;
size_t dimension;
friend std::ostream& operator<<(std::ostream& os, const Problem& pb);
};
std::ostream& operator<<(std::ostream& os, const Problem& pb)
{
os << "u=" << pb.dummy << "_"
<< "e=" << pb.epistasis << "_"
<< "n=" << pb.neutrality << "_"
<< "r=" << pb.ruggedness << "_"
<< "t=" << pb.max_target << "_"
<< "d=" << pb.dimension;
return os;
}
/*****************************************************************************
* IOH problem adaptation.
*****************************************************************************/
class WModelFlat : public ioh::problem::wmodel::WModelOneMax
{
public:
WModelFlat(const int instance, const int n_variables,
const double dummy_para, const int epistasis_para, const int neutrality_para,
const int ruggedness_para)
: WModelOneMax(instance, n_variables, dummy_para, epistasis_para, neutrality_para, ruggedness_para)
{ }
protected:
double transform_objectives(const double y) override
{ // Disable objective function shift & scaling.
return y;
}
};
/*****************************************************************************
* Command line interface.
*****************************************************************************/
int main(int argc, char* argv[])
{
/***** Global parameters. *****/
enum { NO_ERROR = 0, ERROR_USAGE = 100 };
std::map<size_t, Problem> benchmark {
/* ┌ problem index in the map
* │ ┌ problem ID in IOH experimenter
* │ │ ┌ dummy
* │ │ │ ┌ epistasis
* │ │ │ │ ┌ neutrality
* │ │ │ │ │ ┌ ruggedness
* │ │ │ │ │ │ ┌ max target
* │ │ │ │ │ │ │ ┌ dimension (bitstring length) */
{ 0 /* 1*/, {0, 6, 2, 10, 10, 20 }},
{ 1 /* 2*/, {0, 6, 2, 18, 10, 20 }},
{ 2 /* 3*/, {0, 5, 1, 72, 16, 16 }},
{ 3 /* 4*/, {0, 9, 3, 72, 16, 48 }},
{ 4 /* 5*/, {0, 23, 1, 90, 25, 25 }},
{ 5 /* 6*/, {0, 2, 1, 397, 32, 32 }},
{ 6 /* 7*/, {0, 11, 4, 0, 32, 128 }},
{ 7 /* 8*/, {0, 14, 4, 0, 32, 128 }},
{ 8 /* 9*/, {0, 8, 4, 128, 32, 128 }},
{ 9 /*10*/, {0, 36, 1, 245, 50, 50 }},
{10 /*11*/, {0, 21, 2, 256, 50, 100 }},
{11 /*12*/, {0, 16, 3, 613, 50, 150 }},
{12 /*13*/, {0, 32, 2, 256, 64, 128 }},
{13 /*14*/, {0, 21, 3, 16, 64, 192 }},
{14 /*15*/, {0, 21, 3, 256, 64, 192 }},
{15 /*16*/, {0, 21, 3, 403, 64, 192 }},
{16 /*17*/, {0, 52, 4, 2, 64, 256 }},
{17 /*18*/, {0, 60, 1, 16, 75, 75 }},
{18 /*19*/, {0, 32, 2, 4, 75, 150 }}
};
eoFunctorStore store;
eoParser parser(argc, argv, "FastGA interface for iRace");
/***** Problem parameters *****/
auto problem_p = parser.getORcreateParam<size_t>(0,
"problem", "Problem ID",
'p', "Problem", /*required=*/true);
const size_t problem = problem_p.value();
assert(problem < benchmark.size());
// const size_t dimension = parser.getORcreateParam<size_t>(1000,
// "dimension", "Dimension size",
// 'd', "Problem").value();
const size_t dimension = benchmark[problem].dimension;
auto instance_p = parser.getORcreateParam<size_t>(0,
"instance", "Instance ID",
'i', "Instance", /*required=*/false);
const size_t instance = instance_p.value();
const size_t max_evals = parser.getORcreateParam<size_t>(5 * dimension,
"max-evals", "Maximum number of evaluations (default: 5*dim, else the given value)",
'e', "Stopping criterion").value();
const size_t buckets = parser.getORcreateParam<size_t>(100,
"buckets", "Number of buckets for discretizing the ECDF",
'b', "Performance estimation").value();
/***** Generic options *****/
uint32_t seed =
parser.getORcreateParam<uint32_t>(0,
"seed", "Random number seed (0 = epoch)",
'S').value();
if(seed == 0) {
seed = time(0);
}
// rng is a global
rng.reseed(seed);
bool full_log =
parser.getORcreateParam<bool>(0,
"full-log", "Log the full search in CSV files"/* (using the IOH profiler format)"*/,
'F').value();
bool output_mat =
parser.getORcreateParam<bool>(0,
"output-mat", "Output the aggregated attainment matrix instead of its scalar sum (fancy colormap on stderr, parsable CSV on stdout).",
'A').value();
/***** populations sizes *****/
auto pop_size_p = parser.getORcreateParam<size_t>(5,
"pop-size", "Population size",
'P', "Operator Choice", /*required=*/false);
const size_t pop_size = pop_size_p.value();
const size_t pop_size_max = 200;
auto offspring_size_p = parser.getORcreateParam<size_t>(0,
"offspring-size", "Offsprings size (0 = same size than the parents pop, see --pop-size)",
'O', "Operator Choice", /*required=*/false); // Single alternative, not required.
const size_t offspring_size = offspring_size_p.value();
size_t generations = static_cast<size_t>(std::floor(
static_cast<double>(max_evals) / static_cast<double>(pop_size)));
// const size_t generations = std::numeric_limits<size_t>::max();
if(generations < 1) {
generations = 1;
}
/***** metric parameters *****/
auto crossover_rate_p = parser.getORcreateParam<double>(0.5,
"crossover-rate", "",
'C', "Operator Choice", /*required=*/true);
const double crossover_rate = crossover_rate_p.value();
auto mutation_rate_p = parser.getORcreateParam<double>(0,
"mutation-rate", "",
'M', "Operator Choice", /*required=*/true);
const double mutation_rate = mutation_rate_p.value();
/***** operators *****/
auto continuator_p = parser.getORcreateParam<size_t>(0,
"continuator", "Stopping criterion",
'o', "Operator Choice", /*required=*/false); // Single alternative, not required.
const size_t continuator = continuator_p.value();
auto crossover_selector_p = parser.getORcreateParam<size_t>(0,
"cross-selector", "How to selects candidates for cross-over",
's', "Operator Choice", /*required=*/true);
const size_t crossover_selector = crossover_selector_p.value();
auto crossover_p = parser.getORcreateParam<size_t>(0,
"crossover", "",
'c', "Operator Choice", /*required=*/true);
const size_t crossover = crossover_p.value();
auto aftercross_selector_p = parser.getORcreateParam<size_t>(0,
"aftercross-selector", "How to selects between the two individuals altered by cross-over which one will mutate",
'a', "Operator Choice", /*required=*/false); // Single alternative, not required.
const size_t aftercross_selector = aftercross_selector_p.value();
auto mutation_selector_p = parser.getORcreateParam<size_t>(0,
"mut-selector", "How to selects candidate for mutation",
'u', "Operator Choice", /*required=*/true);
const size_t mutation_selector = mutation_selector_p.value();
auto mutation_p = parser.getORcreateParam<size_t>(0,
"mutation", "",
'm', "Operator Choice", /*required=*/true);
const size_t mutation = mutation_p.value();
auto replacement_p = parser.getORcreateParam<size_t>(0,
"replacement", "",
'r', "Operator Choice", /*required=*/true);
const size_t replacement = replacement_p.value();
// Help + Verbose routines
make_verbose(parser);
make_help(parser, /*exit_after*/false, std::clog);
if(parser.userNeedsHelp()) {
// Fake operators, just to be able to call make_foundry
// to get the configured operators slots.
eoEvalFuncPtr<Bits> fake_eval(fake_func);
eoUniformGenerator<int> fake_gen(0, 1);
eoInitFixedLength<Bits> fake_init(/*bitstring size=*/1, fake_gen);
auto fake_foundry = make_foundry(store, fake_init, fake_eval, max_evals, /*generations=*/ 1, 0);
std::clog << std::endl << "Available operators:" << std::endl;
print_operators( continuator_p, fake_foundry.continuators , std::clog);
print_operators( crossover_rate_p, fake_foundry.crossover_rates , std::clog);
print_operators( crossover_selector_p, fake_foundry.crossover_selectors , std::clog);
print_operators(aftercross_selector_p, fake_foundry.aftercross_selectors, std::clog);
print_operators( crossover_p, fake_foundry.crossovers , std::clog);
print_operators( mutation_rate_p, fake_foundry.mutation_rates , std::clog);
print_operators( mutation_selector_p, fake_foundry.mutation_selectors , std::clog);
print_operators( mutation_p, fake_foundry.mutations , std::clog);
print_operators( replacement_p, fake_foundry.replacements , std::clog);
print_operators( offspring_size_p, fake_foundry.offspring_sizes , std::clog);
print_operators( pop_size_p, (size_t)1, pop_size_max , std::clog);
std::clog << std::endl;
// If we were to make a DoE sampling numeric parameters,
// we would use that many samples:
size_t fake_sample_size = 10;
std::clog << "With " << fake_sample_size << " samples for numeric parameters..." << std::endl;
size_t n =
fake_sample_size //crossover_rates
* fake_foundry.crossover_selectors.size()
* fake_foundry.crossovers.size()
* fake_foundry.aftercross_selectors.size()
* fake_sample_size //mutation_rates
* fake_foundry.mutation_selectors.size()
* fake_foundry.mutations.size()
* fake_foundry.replacements.size()
* fake_foundry.continuators.size()
* fake_sample_size //offspring_sizes
* fake_sample_size //pop_size
;
std::clog << "~" << n << " possible algorithms configurations." << std::endl;
std::clog << "Ranges of configurable parameters (redirect the stdout in a file to use it with iRace): " << std::endl;
// Do not print problem and instances, as they are managed separately by irace.
std::cout << "# name\tswitch\ttype\trange" << std::endl;
print_irace( continuator_p, fake_foundry.continuators , std::cout);
print_irace( crossover_rate_p, fake_foundry.crossover_rates , std::cout);
print_irace( crossover_selector_p, fake_foundry.crossover_selectors , std::cout);
print_irace(aftercross_selector_p, fake_foundry.aftercross_selectors, std::cout);
print_irace( crossover_p, fake_foundry.crossovers , std::cout);
print_irace( mutation_rate_p, fake_foundry.mutation_rates , std::cout);
print_irace( mutation_selector_p, fake_foundry.mutation_selectors , std::cout);
print_irace( mutation_p, fake_foundry.mutations , std::cout);
print_irace( replacement_p, fake_foundry.replacements , std::cout);
print_irace( offspring_size_p, fake_foundry.offspring_sizes , std::cout);
print_irace( pop_size_p, 1, pop_size_max , std::cout);
// std::ofstream irace_param("fastga.params");
// irace_param << "# name\tswitch\ttype\tvalues" << std::endl;
exit(NO_ERROR);
}
eo::log << eo::debug << "Maximum number of evaluations: " << max_evals << std::endl;
eo::log << eo::debug << "Number of generations: " << generations << std::endl;
/*****************************************************************************
* IOH stuff.
*****************************************************************************/
/***** IOH logger *****/
auto max_target = benchmark[problem].max_target;
ioh::logger::eah::Log10Scale<double> target_range(0, max_target, buckets);
ioh::logger::eah::Log10Scale<size_t> budget_range(0, max_evals, buckets);
ioh::logger::EAH eah_logger(target_range, budget_range);
ioh::logger::Combine loggers(eah_logger);
std::shared_ptr<ioh::logger::FlatFile> csv_logger = nullptr;
if(full_log) {
// Build up an algorithm name from main parameters.
std::ostringstream name;
name << "FastGA";
for(auto& p : {
crossover_selector_p,
crossover_p,
aftercross_selector_p,
mutation_selector_p,
mutation_p,
replacement_p }) {
name << "_" << p.shortName() << "=" << p.getValue();
}
for(auto& p : {
crossover_rate_p,
mutation_rate_p }) {
name << "_" << p.shortName() << "=" << p.getValue();
}
for(auto& p : {pop_size_p,
offspring_size_p }) {
name << "_" << p.shortName() << "=" << p.getValue();
}
std::clog << name.str() << std::endl;
// Build up a problem description.
std::ostringstream desc;
desc << "pb=" << problem << "_";
desc << benchmark[problem]; // Use the `operator<<` above.
std::clog << desc.str() << std::endl;
std::filesystem::path folder = desc.str();
std::filesystem::create_directories(folder);
ioh::trigger::OnImprovement on_improvement;
ioh::watch::Evaluations evaluations;
ioh::watch::TransformedYBest transformed_y_best;
std::vector<std::reference_wrapper<ioh::logger::Trigger >> t = {on_improvement};
std::vector<std::reference_wrapper<ioh::logger::Property>> w = {evaluations,transformed_y_best};
csv_logger = std::make_shared<ioh::logger::FlatFile>(
// {std::ref(on_improvement)},
// {std::ref(evaluations),std::ref(transformed_y_best)},
t, w,
name.str(),
folder
);
loggers.append(*csv_logger);
}
/***** IOH problem *****/
double w_dummy = benchmark[problem].dummy;
int w_epitasis = benchmark[problem].epistasis;
int w_neutrality = benchmark[problem].neutrality;
int w_ruggedness = benchmark[problem].ruggedness;
// std::string problem_name = "OneMax";
// problem_name = problem_name
// + "_D" + std::to_string((int)(w_dummy * dimension))
// + "_E" + std::to_string(w_epitasis)
// + "_N" + std::to_string(w_neutrality)
// + "_R" + std::to_string(w_ruggedness);
// ioh::problem::wmodel::WModelOneMax w_model_om(
WModelFlat w_model_om(
instance,
dimension,
w_dummy,
w_epitasis,
w_neutrality,
w_ruggedness);
/***** Bindings *****/
w_model_om.attach_logger(loggers);
/*****************************************************************************
* Binding everything together.
*****************************************************************************/
eoEvalIOHproblem<Bits> onemax_pb(w_model_om, loggers);
// eoEvalPrint<Bits> eval_print(onemax_pb, std::clog, "\n");
// eoEvalFuncCounter<Bits> eval_count(onemax_pb);
eoEvalCounterThrowException<Bits> eval_count(onemax_pb, max_evals);
eoPopLoopEval<Bits> onemax_eval(eval_count);
/***** Instanciate and run the algo *****/
eoBooleanGenerator<int> bgen;
eoInitFixedLength<Bits> onemax_init(/*bitstring size=*/dimension, bgen);
auto& foundry = make_foundry(store, onemax_init, eval_count, max_evals, generations, max_target);
eoAlgoFoundry<Bits>::Encodings encoded_algo(foundry.size());
encoded_algo[foundry.crossover_rates .index()] = crossover_rate;
encoded_algo[foundry.crossover_selectors .index()] = crossover_selector;
encoded_algo[foundry.crossovers .index()] = crossover;
encoded_algo[foundry.aftercross_selectors.index()] = aftercross_selector;
encoded_algo[foundry.mutation_rates .index()] = mutation_rate;
encoded_algo[foundry.mutation_selectors .index()] = mutation_selector;
encoded_algo[foundry.mutations .index()] = mutation;
encoded_algo[foundry.replacements .index()] = replacement;
encoded_algo[foundry.continuators .index()] = continuator;
encoded_algo[foundry.offspring_sizes .index()] = offspring_size;
// std::clog << "Encoded algorithm:" << std::endl;
foundry.select(encoded_algo);
std::clog << foundry.name() << std::endl;
/*****************************************************************************
* Run and output results.
*****************************************************************************/
eoPop<Bits> pop;
pop.append(pop_size, onemax_init);
try {
onemax_eval(pop,pop);
foundry(pop); // Actually run the selected algorithm.
} catch(eoMaxEvalException & e) {
eo::log << eo::debug << "Reached maximum evaluations: " << eval_count.getValue() << " / " << max_evals << std::endl;
}
/***** IOH perf stats *****/
double perf = ioh::logger::eah::stat::under_curve::volume(eah_logger);
if(perf == 0 or perf > max_target * max_evals * 1.0) {
std::cerr << "WARNING: illogical performance? " << perf
<< " Check the bounds or the algorithm." << std::endl;
}
// std::clog << "After " << eval_count.getValue() << " / " << max_evals << " evaluations" << std::endl;
if(output_mat) {
std::vector<std::vector<double>> mat = ioh::logger::eah::stat::distribution(eah_logger);
// Fancy color map on clog.
std::clog << ioh::logger::eah::colormap(mat) << std::endl;
// Parsable CSV on cout.
std::clog << "Attainment matrix distribution: " << std::endl;
assert(mat.size() > 0);
assert(mat[0].size() > 1);
for(size_t i = mat.size()-1; i > 0; --i) {
assert(mat[i].size() >= 1);
std::cout << mat[i][0];
for(size_t j = 1; j < mat[i].size(); ++j) {
std::cout << "," << mat[i][j];
}
std::cout << std::endl;
}
} else {
// iRace expects minimization
std::cout << -1 * perf << std::endl;
}
}