/home/marsyas/marsyas/src/marsyas/RunningAutocorrelation.cpp

/*
 ** Copyright (C) 2010 Stefaan Lippens
 **
 ** 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 2 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 GNU General Public License
 ** along with this program; if not, write to the Free Software
 ** Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 */

#include "RunningAutocorrelation.h"

// For min().
#include <algorithm>

using namespace std;
using namespace Marsyas;

RunningAutocorrelation::RunningAutocorrelation(mrs_string name) :
        MarSystem("RunningAutocorrelation", name)
{
    addControls();
}

RunningAutocorrelation::RunningAutocorrelation(const RunningAutocorrelation& a) :
        MarSystem(a)
{
    ctrl_maxLag_ = getctrl("mrs_natural/maxLag");
    ctrl_normalize_ = getctrl("mrs_bool/normalize");
    ctrl_doNotNormalizeForLag0_ = getctrl("mrs_bool/doNotNormalizeForLag0");
    ctrl_clear_ = getctrl("mrs_bool/clear");
    ctrl_unfoldToObservations_ = getctrl("mrs_bool/unfoldToObservations");
}

RunningAutocorrelation::~RunningAutocorrelation()
{
}

MarSystem*
RunningAutocorrelation::clone() const
{
    return new RunningAutocorrelation(*this);
}

void RunningAutocorrelation::addControls()
{
    addctrl("mrs_natural/maxLag", 15, ctrl_maxLag_);
    setctrlState("mrs_natural/maxLag", true);
    addctrl("mrs_bool/normalize", false, ctrl_normalize_);
    setctrlState("mrs_bool/normalize", true);
    addctrl("mrs_bool/doNotNormalizeForLag0", false,
            ctrl_doNotNormalizeForLag0_);
    setctrlState("mrs_bool/doNotNormalizeForLag0", true);
    addctrl("mrs_bool/clear", false, ctrl_clear_);
    setctrlState("mrs_bool/clear", true);
    addctrl("mrs_bool/unfoldToObservations", false, ctrl_unfoldToObservations_);
    setctrlState("mrs_bool/unfoldToObservations", true);
}

mrs_string prefixObservationNamesWithAutocorrelationUnfoldingPrefixes_(
    mrs_string inObservationNames, mrs_bool normalize,
    mrs_bool doNotNormalizeForLag0, mrs_natural maxLag)
{
    vector<mrs_string> inObsNames = obsNamesSplit(inObservationNames);
    mrs_string onObsNames("");
    for (vector<mrs_string>::iterator it = inObsNames.begin(); it
            != inObsNames.end(); it++)
    {
        for (int lag = 0; lag <= maxLag; lag++)
        {
            ostringstream oss;
            if (normalize && !(doNotNormalizeForLag0 && lag == 0))
            {
                oss << "Normalized";
            }
            oss << "Autocorr" << lag << "_" << (*it) << ",";
            onObsNames += oss.str();
        }

    }
    return onObsNames;

}

void RunningAutocorrelation::myUpdate(MarControlPtr sender)
{
    MRSDIAG("RunningAutocorrelation.cpp - RunningAutocorrelation:myUpdate");

    // Update the cache of the control values.
    maxLag_ = ctrl_maxLag_->to<mrs_natural> ();
    if (maxLag_ < 0)
    {
        MRSERR("maxLag should be greater than zero.");
        // Setting it to zero to be sure.
        maxLag_ = 0;
    }
    normalize_ = ctrl_normalize_->to<mrs_bool> ();
    doNotNormalizeForLag0_ = ctrl_doNotNormalizeForLag0_->to<mrs_bool> ();
    unfoldToObservations_ = ctrl_unfoldToObservations_->to<mrs_bool> ();

    // Configure output flow.
    mrs_natural onObservations = unfoldToObservations_ ? inObservations_
                                 * (maxLag_ + 1) : inObservations_;
    mrs_natural onSamples = unfoldToObservations_ ? 1 : maxLag_ + 1;
    ctrl_onSamples_->setValue(onSamples, NOUPDATE);
    ctrl_onObservations_->setValue(onObservations, NOUPDATE);
    ctrl_osrate_->setValue(ctrl_israte_, NOUPDATE);
    // Handle the observation names.
    mrs_string onObsNames = ctrl_inObsNames_->to<mrs_string> ();
    if (unfoldToObservations_)
    {
        onObsNames
        = prefixObservationNamesWithAutocorrelationUnfoldingPrefixes_(
              onObsNames, normalize_, doNotNormalizeForLag0_, maxLag_);
    }
    ctrl_onObsNames_->setValue(onObsNames, NOUPDATE);

    // Allocate and initialize the buffers and counters.
    this->acBuffer_.stretch(inObservations_, maxLag_ + 1);
    this->acBuffer_.setval(0.0);
    this->memory_.stretch(inObservations_, maxLag_);
    this->memory_.setval(0.0);

    // We just cleared the internal buffers, so we should reset the clear control.
    ctrl_clear_->setValue(false, NOUPDATE);
}

void RunningAutocorrelation::myProcess(realvec& in, realvec& out)
{
    for (mrs_natural i = 0; i < inObservations_; ++i)
    {
        // Calculate the autocorrelation terms
        for (mrs_natural lag = 0; lag <= maxLag_; lag++)
        {
            // For the first part, we need the memory.
            for (mrs_natural n = 0; n < min(lag, inSamples_); n++)
            {
                acBuffer_(i, lag) += in(i, n) * memory_(i, maxLag_ - lag + n);
            }
            // For the second part, we have enough with the input slice.
            for (mrs_natural n = lag; n < inSamples_; n++)
            {
                acBuffer_(i, lag) += in(i, n) * in(i, n - lag);
            }
        }

        // Store result in output vector.
        mrs_natural u = (mrs_natural) unfoldToObservations_;
        mrs_natural o_base = unfoldToObservations_ ? i * (maxLag_ + 1) : i;
        for (mrs_natural lag = 0; lag <= maxLag_; lag++)
        {
            out(o_base + u * lag, (1 - u) * lag) = acBuffer_(i, lag);

        }
        // Do the (optional) normalization.
        if (normalize_ && acBuffer_(i, 0) > 0.0)
        {
            mrs_natural lag = doNotNormalizeForLag0_ ? 1 : 0;
            for (; lag <= maxLag_; lag++)
            {
                out(o_base + u * lag, (1 - u) * lag) /= acBuffer_(i, 0);
            }
        }

        // Remember the last samples for next time.
        // First, shift samples in memory (if needed).
        for (mrs_natural m = 0; m < maxLag_ - inSamples_; m++)
        {
            memory_(i, m) = memory_(i, m + inSamples_);
        }
        // Second, copy over samples from input.
        for (mrs_natural m = 1; m <= min(maxLag_, inSamples_); m++)
        {
            memory_(i, maxLag_ - m) = in(i, inSamples_ - m);
        }
    }
}

---