AdaptiveImportanceSampler.h

Go to the documentation of this file.

// -*- lsst-c++ -*-
/*
 * LSST Data Management System
 * Copyright 2008-2013 LSST Corporation.
 *
 * This product includes software developed by the
 * LSST Project (http://www.lsst.org/).
 *
 * This program 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 3 of the License, or
 * (at your option) any later version.
 *
 * This program 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 LSST License Statement and
 * the GNU General Public License along with this program.  If not,
 * see <http://www.lsstcorp.org/LegalNotices/>.
 */
 
#ifndef LSST_MEAS_MODELFIT_AdaptiveImportanceSampler_h_INCLUDED
#define LSST_MEAS_MODELFIT_AdaptiveImportanceSampler_h_INCLUDED
 
#include <map>
 
#include "lsst/pex/config.h"
#include "lsst/afw/table/Schema.h"
#include "lsst/meas/modelfit/Sampler.h"
#include "lsst/meas/modelfit/Mixture.h"
 
namespace lsst { namespace meas { namespace modelfit {

class ImportanceSamplerControl {
public:
    LSST_CONTROL_FIELD(nSamples, int, "Number of Monte Carlo samples to draw");
    LSST_CONTROL_FIELD(nUpdateSteps, int, "Number of Expectation-Maximization update iterations");
    LSST_CONTROL_FIELD(tau1, double, "Damping parameter for E-M update (see Mixture::updateEM)");
    LSST_CONTROL_FIELD(tau2, double, "Damping parameter for E-M update (see Mixture::updateEM)");
    LSST_CONTROL_FIELD(
        targetPerplexity, double,
        "Minimum value for normalized perplexity after this iteration; if the actual value is less "
        "than this, this iteration will be repeated up to maxRepeat times until the target is met. "
        "In addition, if any previous iteration meets this target, this iteration will be skipped."
    );
    LSST_CONTROL_FIELD(
        maxRepeat, int,
        "Maximum number of times this iteration will be repeated to meet the perplexityTarget"
    );
 
    ImportanceSamplerControl() :
        nSamples(2000), nUpdateSteps(2), tau1(1E-4), tau2(0.5), targetPerplexity(1.0), maxRepeat(0)
    {}
};

class AdaptiveImportanceSampler : public Sampler {
public:

    AdaptiveImportanceSampler(
        afw::table::Schema & sampleSchema,
        std::shared_ptr<afw::math::Random> rng,
        std::map<int,ImportanceSamplerControl> const & ctrls,
        bool doSaveIterations=false
    );
 
    void run(
        SamplingObjective const & objective,
        std::shared_ptr<Mixture> proposal,
        afw::table::BaseCatalog & samples
    ) const override;
 
    double computeNormalizedPerplexity(afw::table::BaseCatalog const & samples) const;
 
    double computeEffectiveSampleSizeFraction(afw::table::BaseCatalog const & samples) const;
 
private:
    bool _doSaveIterations;
    std::shared_ptr<afw::math::Random>  _rng;
    std::map<int,ImportanceSamplerControl> _ctrls;
    afw::table::Key<Scalar> _weightKey;
    afw::table::Key<Scalar> _objectiveKey;
    afw::table::Key<Scalar> _proposalKey;
    afw::table::Key< afw::table::Array<Scalar> > _parametersKey;
    afw::table::Key<int> _iterCtrlKey;
    afw::table::Key<int> _iterRepeatKey;
};
 
}}} // namespace lsst::meas::modelfit
 
#endif // !LSST_MEAS_MODELFIT_AdaptiveImportanceSampler_h_INCLUDED