lar_content::ThreeDLinearFitFeatureTool Class Reference

LinearFitFeatureTool class for the calculation of variables related to sliding linear fit. More...

#include "TrackShowerIdFeatureTool.h"

Inheritance diagram for lar_content::ThreeDLinearFitFeatureTool:

Public Types
typedef std::vector< MvaFeatureTool< Ts... > * >	FeatureToolVector

Public Member Functions
	ThreeDLinearFitFeatureTool ()
	Default constructor. More...
void	Run (LArMvaHelper::MvaFeatureVector &featureVector, const pandora::Algorithm *const pAlgorithm, const pandora::ParticleFlowObject *const pInputPfo)
virtual void	Run (MvaTypes::MvaFeatureVector &featureVector, Ts...args)=0
	Run the algorithm tool. More...

Private Member Functions
pandora::StatusCode	ReadSettings (const pandora::TiXmlHandle xmlHandle)
void	CalculateVariablesSlidingLinearFit (const pandora::Cluster *const pCluster, float &straightLineLengthLarge, float &diffWithStraigthLineMean, float &maxFitGapLength, float &rmsSlidingLinearFit) const
	Calculation of several variables related to sliding linear fit. More...

Private Attributes
unsigned int	m_slidingLinearFitWindow
	The sliding linear fit window. More...
unsigned int	m_slidingLinearFitWindowLarge
	The sliding linear fit window - should be large, providing a simple linear fit. More...

Detailed Description

LinearFitFeatureTool class for the calculation of variables related to sliding linear fit.

Definition at line 123 of file TrackShowerIdFeatureTool.h.

Member Typedef Documentation

template<typename... Ts>

typedef std::vector<MvaFeatureTool<Ts...> *> lar_content::MvaFeatureTool< Ts >::FeatureToolVector

inherited

Definition at line 30 of file LArMvaHelper.h.

Constructor & Destructor Documentation

lar_content::ThreeDLinearFitFeatureTool::ThreeDLinearFitFeatureTool

(

)

Default constructor.

Definition at line 253 of file TrackShowerIdFeatureTool.cc.

                                                       :
    m_slidingLinearFitWindow(3),
    m_slidingLinearFitWindowLarge(10000)
{
}

Member Function Documentation

void lar_content::ThreeDLinearFitFeatureTool::CalculateVariablesSlidingLinearFit ( const pandora::Cluster *const  pCluster, float &  straightLineLengthLarge, float &  diffWithStraigthLineMean, float &  maxFitGapLength, float &  rmsSlidingLinearFit  ) const

private

Calculation of several variables related to sliding linear fit.

Parameters

pCluster	the cluster we are characterizing
straightLineLengthLarge	to receive to length reported by the straight line fit
diffWithStraigthLineMean	to receive the difference with straight line mean variable
diffWithStraightLineSigma	to receive the difference with straight line sigma variable
dTdLWidth	to receive the dTdL width variable
maxFitGapLength	to receive the max fit gap length variable
rmsSlidingLinearFit	to receive the RMS from the linear fit

Definition at line 310 of file TrackShowerIdFeatureTool.cc.

References lar_content::LArGeometryHelper::CalculateGapDeltaZ(), f, lar_content::LArClusterHelper::GetClusterHitType(), lar_content::LayerFitResult::GetFitT(), lar_content::TwoDSlidingFitResult::GetGlobalMaxLayerPosition(), lar_content::TwoDSlidingFitResult::GetGlobalMinLayerPosition(), lar_content::TwoDSlidingFitResult::GetGlobalPosition(), lar_content::LayerFitResult::GetGradient(), lar_content::LayerFitResult::GetL(), lar_content::TwoDSlidingFitResult::GetLayer(), lar_content::TwoDSlidingFitResult::GetLayerFitResultMap(), lar_content::LayerFitResult::GetRms(), lar_content::LArGeometryHelper::GetWireZPitch(), m_slidingLinearFitWindow, m_slidingLinearFitWindowLarge, max, and min.

Referenced by Run().

{
    try
    {
        const TwoDSlidingFitResult slidingFitResult(pCluster, m_slidingLinearFitWindow, LArGeometryHelper::GetWireZPitch(this->GetPandora()));
        const TwoDSlidingFitResult slidingFitResultLarge(pCluster, m_slidingLinearFitWindowLarge, LArGeometryHelper::GetWireZPitch(this->GetPandora()));

        if (slidingFitResult.GetLayerFitResultMap().empty())
            throw StatusCodeException(STATUS_CODE_NOT_INITIALIZED);

        straightLineLengthLarge = (slidingFitResultLarge.GetGlobalMaxLayerPosition() - slidingFitResultLarge.GetGlobalMinLayerPosition()).GetMagnitude();
        rmsSlidingLinearFit = 0.f;

        FloatVector diffWithStraightLineVector;
        const HitType hitType(LArClusterHelper::GetClusterHitType(pCluster));
        CartesianVector previousFitPosition(slidingFitResult.GetGlobalMinLayerPosition());
        float dTdLMin(+std::numeric_limits<float>::max()), dTdLMax(-std::numeric_limits<float>::max());

        for (const auto &mapEntry : slidingFitResult.GetLayerFitResultMap())
        {
            const LayerFitResult &layerFitResult(mapEntry.second);
            rmsSlidingLinearFit += layerFitResult.GetRms();

            CartesianVector thisFitPosition(0.f, 0.f, 0.f);
            slidingFitResult.GetGlobalPosition(layerFitResult.GetL(), layerFitResult.GetFitT(), thisFitPosition);

            LayerFitResultMap::const_iterator iterLarge = slidingFitResultLarge.GetLayerFitResultMap().find(slidingFitResultLarge.GetLayer(layerFitResult.GetL()));

            if (slidingFitResultLarge.GetLayerFitResultMap().end() == iterLarge)
                throw StatusCodeException(STATUS_CODE_FAILURE);

            diffWithStraightLineVector.push_back(static_cast<float>(std::fabs(layerFitResult.GetFitT() - iterLarge->second.GetFitT())));

            const float thisGapLength((thisFitPosition - previousFitPosition).GetMagnitude());
            const float minZ(std::min(thisFitPosition.GetZ(), previousFitPosition.GetZ()));
            const float maxZ(std::max(thisFitPosition.GetZ(), previousFitPosition.GetZ()));

            if ((maxZ - minZ) > std::numeric_limits<float>::epsilon())
            {
                const float gapZ(LArGeometryHelper::CalculateGapDeltaZ(this->GetPandora(), minZ, maxZ, hitType));
                const float correctedGapLength(thisGapLength * (1.f - gapZ / (maxZ - minZ)));

                if (correctedGapLength > maxFitGapLength)
                    maxFitGapLength = correctedGapLength;
            }
            else
            {
                maxFitGapLength = 0.f;
            }

            dTdLMin = std::min(dTdLMin, static_cast<float>(layerFitResult.GetGradient()));
            dTdLMax = std::max(dTdLMax, static_cast<float>(layerFitResult.GetGradient()));
            previousFitPosition = thisFitPosition;
        }

        if (diffWithStraightLineVector.empty())
            throw StatusCodeException(STATUS_CODE_FAILURE);

        diffWithStraightLineMean = 0.f;

        for (const float diffWithStraightLine : diffWithStraightLineVector)
            diffWithStraightLineMean += diffWithStraightLine;

        diffWithStraightLineMean /= static_cast<float>(diffWithStraightLineVector.size());

    }
    catch (const StatusCodeException &)
    {
        straightLineLengthLarge = -1.f;
        diffWithStraightLineMean = -1.f;
        maxFitGapLength = -1.f;
        rmsSlidingLinearFit = -1.f;
    }
}

StatusCode lar_content::ThreeDLinearFitFeatureTool::ReadSettings ( const pandora::TiXmlHandle  xmlHandle )

private

Definition at line 388 of file TrackShowerIdFeatureTool.cc.

References m_slidingLinearFitWindow, and m_slidingLinearFitWindowLarge.

{
    PANDORA_RETURN_RESULT_IF_AND_IF(STATUS_CODE_SUCCESS, STATUS_CODE_NOT_FOUND, !=, XmlHelper::ReadValue(xmlHandle,
        "SlidingLinearFitWindow", m_slidingLinearFitWindow));

    PANDORA_RETURN_RESULT_IF_AND_IF(STATUS_CODE_SUCCESS, STATUS_CODE_NOT_FOUND, !=, XmlHelper::ReadValue(xmlHandle,
        "SlidingLinearFitWindowLarge", m_slidingLinearFitWindowLarge));

    return STATUS_CODE_SUCCESS;
}

template<typename... Ts>

virtual void lar_content::MvaFeatureTool< Ts >::Run ( MvaTypes::MvaFeatureVector &  featureVector, Ts...  args  )

pure virtualinherited

Run the algorithm tool.

Parameters

featureVector	the vector of features to append
args	arguments to pass to the tool

void lar_content::ThreeDLinearFitFeatureTool::Run	(	LArMvaHelper::MvaFeatureVector &	featureVector,
		const pandora::Algorithm *const	pAlgorithm,
		const pandora::ParticleFlowObject *const	pInputPfo
	)

Definition at line 261 of file TrackShowerIdFeatureTool.cc.

References CalculateVariablesSlidingLinearFit(), f, lar_content::LArPfoHelper::GetThreeDLengthSquared(), and lar_content::LArPfoHelper::GetTwoDClusterList().

{
    if (PandoraContentApi::GetSettings(*pAlgorithm)->ShouldDisplayAlgorithmInfo())
        std::cout << "----> Running Algorithm Tool: " << this->GetInstanceName() << ", " << this->GetType() << std::endl;

    ClusterList clusterList;
    LArPfoHelper::GetTwoDClusterList(pInputPfo, clusterList);
    float diffWithStraightLineMean(0.f), maxFitGapLength(0.f), rmsSlidingLinearFit(0.f);
    LArMvaHelper::MvaFeature length, diff, gap, rms;
    unsigned int nClustersUsed(0);

    for (const Cluster *const pCluster : clusterList)
    {
        float straightLineLengthLargeCluster(-1.f), diffWithStraightLineMeanCluster(-1.f), maxFitGapLengthCluster(-1.f), rmsSlidingLinearFitCluster(-1.f);

        this->CalculateVariablesSlidingLinearFit(pCluster, straightLineLengthLargeCluster, diffWithStraightLineMeanCluster, maxFitGapLengthCluster, rmsSlidingLinearFitCluster);

        if (straightLineLengthLargeCluster > std::numeric_limits<float>::epsilon())
        {
            diffWithStraightLineMeanCluster  /= straightLineLengthLargeCluster;
            maxFitGapLengthCluster           /= straightLineLengthLargeCluster;
            rmsSlidingLinearFitCluster       /= straightLineLengthLargeCluster;

            diffWithStraightLineMean   += diffWithStraightLineMeanCluster;
            maxFitGapLength            += maxFitGapLengthCluster;
            rmsSlidingLinearFit        += rmsSlidingLinearFitCluster;

            ++nClustersUsed;
        }
    }

    if (nClustersUsed > 0)
    {
        const float nClusters(static_cast<float>(nClustersUsed));
        length = std::sqrt(LArPfoHelper::GetThreeDLengthSquared(pInputPfo));
        diff   = diffWithStraightLineMean / nClusters;
        gap    = maxFitGapLength          / nClusters;
        rms    = rmsSlidingLinearFit      / nClusters;
    }

    featureVector.push_back(length);
    featureVector.push_back(diff);
    featureVector.push_back(gap);
    featureVector.push_back(rms);
}

Member Data Documentation

unsigned int lar_content::ThreeDLinearFitFeatureTool::m_slidingLinearFitWindow

private

The sliding linear fit window.

Definition at line 150 of file TrackShowerIdFeatureTool.h.

Referenced by CalculateVariablesSlidingLinearFit(), and ReadSettings().

unsigned int lar_content::ThreeDLinearFitFeatureTool::m_slidingLinearFitWindowLarge

private

The sliding linear fit window - should be large, providing a simple linear fit.

Definition at line 151 of file TrackShowerIdFeatureTool.h.

Referenced by CalculateVariablesSlidingLinearFit(), and ReadSettings().

---

The documentation for this class was generated from the following files:

• /cvmfs/larsoft.opensciencegrid.org/products/larpandoracontent/v03_14_03/source/larpandoracontent/LArTrackShowerId/TrackShowerIdFeatureTool.h
• /cvmfs/larsoft.opensciencegrid.org/products/larpandoracontent/v03_14_03/source/larpandoracontent/LArTrackShowerId/TrackShowerIdFeatureTool.cc