PXDClusterPositionCalibrationAlgorithm Class Reference

Class implementing the PXD cluster position calibration algorithm. More...

#include <PXDClusterPositionCalibrationAlgorithm.h>

Inheritance diagram for PXDClusterPositionCalibrationAlgorithm:

Public Types
enum	EResult {   c_OK ,   c_Iterate ,   c_NotEnoughData ,   c_Failure ,   c_Undefined }
	The result of calibration. More...

Public Member Functions
	PXDClusterPositionCalibrationAlgorithm ()
	Constructor set the prefix to PXDClusterPositionCalibrationAlgorithm.
std::string	getPrefix () const
	Get the prefix used for getting calibration data.
bool	checkPyExpRun (PyObject *pyObj)
	Checks that a PyObject can be successfully converted to an ExpRun type.
Calibration::ExpRun	convertPyExpRun (PyObject *pyObj)
	Performs the conversion of PyObject to ExpRun.
std::string	getCollectorName () const
	Alias for prefix.
void	setPrefix (const std::string &prefix)
	Set the prefix used to identify datastore objects.
void	setInputFileNames (PyObject *inputFileNames)
	Set the input file names used for this algorithm from a Python list.
PyObject *	getInputFileNames ()
	Get the input file names used for this algorithm and pass them out as a Python list of unicode strings.
std::vector< Calibration::ExpRun >	getRunListFromAllData () const
	Get the complete list of runs from inspection of collected data.
RunRange	getRunRangeFromAllData () const
	Get the complete RunRange from inspection of collected data.
IntervalOfValidity	getIovFromAllData () const
	Get the complete IoV from inspection of collected data.
void	fillRunToInputFilesMap ()
	Fill the mapping of ExpRun -> Files.
std::string	getGranularity () const
	Get the granularity of collected data.
EResult	execute (std::vector< Calibration::ExpRun > runs={}, int iteration=0, IntervalOfValidity iov=IntervalOfValidity())
	Runs calibration over vector of runs for a given iteration.
EResult	execute (PyObject *runs, int iteration=0, IntervalOfValidity iov=IntervalOfValidity())
	Runs calibration over Python list of runs. Converts to C++ and then calls the other execute() function.
std::list< Database::DBImportQuery > &	getPayloads ()
	Get constants (in TObjects) for database update from last execution.
std::list< Database::DBImportQuery >	getPayloadValues ()
	Get constants (in TObjects) for database update from last execution but passed by VALUE.
bool	commit ()
	Submit constants from last calibration into database.
bool	commit (std::list< Database::DBImportQuery > payloads)
	Submit constants from a (potentially previous) set of payloads.
const std::string &	getDescription () const
	Get the description of the algorithm (set by developers in constructor)
bool	loadInputJson (const std::string &jsonString)
	Load the m_inputJson variable from a string (useful from Python interface). The return bool indicates success or failure.
const std::string	dumpOutputJson () const
	Dump the JSON string of the output JSON object.
const std::vector< Calibration::ExpRun >	findPayloadBoundaries (std::vector< Calibration::ExpRun > runs, int iteration=0)
	Used to discover the ExpRun boundaries that you want the Python CAF to execute on. This is optional and only used in some.
template<>
std::shared_ptr< TTree >	getObjectPtr (const std::string &name, const std::vector< Calibration::ExpRun > &requestedRuns)
	Specialization of getObjectPtr<TTree>.

Public Attributes
int	minClusterForShapeLikelyhood
	Minimum number of collected clusters for estimating shape likelyhood.
int	minClusterForPositionOffset
	Minimum number of collected clusters for estimating cluster position offsets.
int	maxEtaBins
	Maximum number of eta bins for estimating cluster position offsets.
std::vector< int >	clusterKinds
	Vector of clusterkinds to calibrate.

Protected Member Functions
virtual EResult	calibrate () override
	Run algo on data.
void	setInputFileNames (std::vector< std::string > inputFileNames)
	Set the input file names used for this algorithm.
virtual bool	isBoundaryRequired (const Calibration::ExpRun &)
	Given the current collector data, make a decision about whether or not this run should be the start of a payload boundary.
virtual void	boundaryFindingSetup (std::vector< Calibration::ExpRun >, int)
	If you need to make some changes to your algorithm class before 'findPayloadBoundaries' is run, make them in this function.
virtual void	boundaryFindingTearDown ()
	Put your algorithm back into a state ready for normal execution if you need to.
const std::vector< Calibration::ExpRun > &	getRunList () const
	Get the list of runs for which calibration is called.
int	getIteration () const
	Get current iteration.
std::vector< std::string >	getVecInputFileNames () const
	Get the input file names used for this algorithm as a STL vector.
template<class T>
std::shared_ptr< T >	getObjectPtr (const std::string &name, const std::vector< Calibration::ExpRun > &requestedRuns)
	Get calibration data object by name and list of runs, the Merge function will be called to generate the overall object.
template<class T>
std::shared_ptr< T >	getObjectPtr (std::string name)
	Get calibration data object (for all runs the calibration is requested for) This function will only work during or after execute() has been called once.
template<>
shared_ptr< TTree >	getObjectPtr (const string &name, const vector< ExpRun > &requestedRuns)
	We cheekily cast the TChain to TTree for the returned pointer so that the user never knows Hopefully this doesn't cause issues if people do low level stuff to the tree...
std::string	getGranularityFromData () const
	Get the granularity of collected data.
void	saveCalibration (TClonesArray *data, const std::string &name)
	Store DBArray payload with given name with default IOV.
void	saveCalibration (TClonesArray *data, const std::string &name, const IntervalOfValidity &iov)
	Store DBArray with given name and custom IOV.
void	saveCalibration (TObject *data)
	Store DB payload with default name and default IOV.
void	saveCalibration (TObject *data, const IntervalOfValidity &iov)
	Store DB payload with default name and custom IOV.
void	saveCalibration (TObject *data, const std::string &name)
	Store DB payload with given name with default IOV.
void	saveCalibration (TObject *data, const std::string &name, const IntervalOfValidity &iov)
	Store DB payload with given name and custom IOV.
void	updateDBObjPtrs (const unsigned int event, const int run, const int experiment)
	Updates any DBObjPtrs by calling update(event) for DBStore.
void	setDescription (const std::string &description)
	Set algorithm description (in constructor)
void	clearCalibrationData ()
	Clear calibration data.
Calibration::ExpRun	getAllGranularityExpRun () const
	Returns the Exp,Run pair that means 'Everything'. Currently unused.
void	resetInputJson ()
	Clears the m_inputJson member variable.
void	resetOutputJson ()
	Clears the m_outputJson member variable.
template<class T>
void	setOutputJsonValue (const std::string &key, const T &value)
	Set a key:value pair for the outputJson object, expected to used internally during calibrate()
template<class T>
const T	getOutputJsonValue (const std::string &key) const
	Get a value using a key from the JSON output object, not sure why you would want to do this.
template<class T>
const T	getInputJsonValue (const std::string &key) const
	Get an input JSON value using a key. The normal exceptions are raised when the key doesn't exist.
const nlohmann::json &	getInputJsonObject () const
	Get the entire top level JSON object. We explicitly say this must be of object type so that we might pick.
bool	inputJsonKeyExists (const std::string &key) const
	Test for a key in the input JSON object.

Protected Attributes
std::vector< Calibration::ExpRun >	m_boundaries
	When using the boundaries functionality from isBoundaryRequired, this is used to store the boundaries. It is cleared when.

Private Member Functions
void	createShapeClassifier (std::string treename, PXDClusterShapeClassifierPar *shapeClassifier, PXDClusterShapeIndexPar *shapeIndexer)
	Returns a new classifier and index trained on cluster tree.
PXDClusterShapeClassifierPar	mirrorShapeClassifier (PXDClusterShapeClassifierPar *shapeClassifier, PXDClusterShapeIndexPar *shapeIndexer, int clusterKind)
	Returns a mirrored version of shape classifier.
PXDClusterShapeClassifierPar	localToGlobal (PXDClusterShapeClassifierPar *localShapeClassifier, PXDClusterShapeIndexPar *localShapeIndexer, PXDClusterShapeIndexPar *globalShapeIndexer)
	Returns a shape classifier using global shape indices instead of local ones.
std::string	getExpRunString (Calibration::ExpRun &expRun) const
	Gets the "exp.run" string repr. of (exp,run)
std::string	getFullObjectPath (const std::string &name, Calibration::ExpRun expRun) const
	constructs the full TDirectory + Key name of an object in a TFile based on its name and exprun

Private Attributes
std::string	m_shapeName
	Branches for tree.
std::string	m_mirroredShapeName
	Name of mirrored cluster shape.
float	m_clusterEta
	Eta value of cluster.
float	m_positionOffsetU
	Position offset u of cluster.
float	m_positionOffsetV
	Position offset v of cluster.
int	m_sizeV
	Size in V.
float	m_pitchV
	Branches for pitchtree.
int	m_clusterKind
	Pitch in V.
std::map< int, float >	m_pitchMap
	Helper needed to map the clusterkind to the V pitch of the sensor.
std::map< std::string, std::string >	m_mirrorMap
	Helper needed to map the name of a shape to the name of the mirrored shape.
std::map< std::string, int >	m_sizeMap
	Helper needed to map the name of a shape to the V size of the cluster.
std::set< std::string >	m_shapeSet
	Set of unique shape names.
std::vector< std::string >	m_inputFileNames
	List of input files to the Algorithm, will initially be user defined but then gets the wildcards expanded during execute()
std::map< Calibration::ExpRun, std::vector< std::string > >	m_runsToInputFiles
	Map of Runs to input files. Gets filled when you call getRunRangeFromAllData, gets cleared when setting input files again.
std::string	m_granularityOfData
	Granularity of input data. This only changes when the input files change so it isn't specific to an execution.
ExecutionData	m_data
	Data specific to a SINGLE execution of the algorithm. Gets reset at the beginning of execution.
std::string	m_description {""}
	Description of the algorithm.
std::string	m_prefix {""}
	The name of the TDirectory the collector objects are contained within.
nlohmann::json	m_jsonExecutionInput = nlohmann::json::object()
	Optional input JSON object used to make decisions about how to execute the algorithm code.
nlohmann::json	m_jsonExecutionOutput = nlohmann::json::object()
	Optional output JSON object that can be set during the execution by the underlying algorithm code.

Static Private Attributes
static const Calibration::ExpRun	m_allExpRun = make_pair(-1, -1)
	allExpRun

Detailed Description

Class implementing the PXD cluster position calibration algorithm.

Definition at line 26 of file PXDClusterPositionCalibrationAlgorithm.h.

Member Enumeration Documentation

EResult

enum EResult

inherited

The result of calibration.

Enumerator
c_OK	Finished successfully =0 in Python.
c_Iterate	Needs iteration =1 in Python.
c_NotEnoughData	Needs more data =2 in Python.
c_Failure	Failed =3 in Python.
c_Undefined	Not yet known (before execution) =4 in Python.

Definition at line 40 of file CalibrationAlgorithm.h.

                 {
      c_OK,           
      c_Iterate,      
      c_NotEnoughData,
      c_Failure,      
      c_Undefined     
    };

Constructor & Destructor Documentation

PXDClusterPositionCalibrationAlgorithm()

PXDClusterPositionCalibrationAlgorithm

(

)

Constructor set the prefix to PXDClusterPositionCalibrationAlgorithm.

Definition at line 27 of file PXDClusterPositionCalibrationAlgorithm.cc.

                                                                              :
  CalibrationAlgorithm("PXDClusterPositionCollector"),
  minClusterForShapeLikelyhood(500), minClusterForPositionOffset(2000), maxEtaBins(10),
  // Branches from TTree
  m_clusterEta(0), m_positionOffsetU(0), m_positionOffsetV(0), m_sizeV(0),
  m_pitchV(0), m_clusterKind(0)
{
  setDescription(
    " -------------------------- PXDClusterPositionCalibrationAlgorithm ----------------------\n"
    "                                                                                         \n"
    "  Algorithm for estimating cluster position offsets and shape likelyhoods.               \n"
    " ----------------------------------------------------------------------------------------\n"
  );
}

Member Function Documentation

boundaryFindingSetup()

virtual void boundaryFindingSetup ( std::vector< Calibration::ExpRun > , int  )

inlineprotectedvirtualinherited

If you need to make some changes to your algorithm class before 'findPayloadBoundaries' is run, make them in this function.

Reimplemented in PXDAnalyticGainCalibrationAlgorithm, PXDValidationAlgorithm, SVD3SampleCoGTimeCalibrationAlgorithm, SVD3SampleELSTimeCalibrationAlgorithm, SVDCoGTimeCalibrationAlgorithm, TestBoundarySettingAlgorithm, and TestCalibrationAlgorithm.

Definition at line 252 of file CalibrationAlgorithm.h.

{};

boundaryFindingTearDown()

virtual void boundaryFindingTearDown ( )

inlineprotectedvirtualinherited

Put your algorithm back into a state ready for normal execution if you need to.

Definition at line 257 of file CalibrationAlgorithm.h.

{};

calibrate()

CalibrationAlgorithm::EResult calibrate ( )

overrideprotectedvirtual

Run algo on data.

Implements CalibrationAlgorithm.

Definition at line 42 of file PXDClusterPositionCalibrationAlgorithm.cc.

{
 
  // Read back the V pitch of all cluster kinds in source data
  // This avoids relying on VXD::GeoCache.
  auto pitchtree = getObjectPtr<TTree>("pitchtree");
  pitchtree->SetBranchAddress("PitchV", &m_pitchV);
  pitchtree->SetBranchAddress("ClusterKind", &m_clusterKind);
 
  for (int i = 0; i < pitchtree->GetEntries(); ++i) {
    pitchtree->GetEntry(i);
    m_pitchMap[m_clusterKind] = m_pitchV;
  }
 
  // Buffer temporary payloads for shape calibration for all
  // clusterkinds and angle bins
  typedef tuple<int, int, int> bufferkey_t;
  typedef pair<PXDClusterShapeIndexPar, PXDClusterShapeClassifierPar> buffervalue_t;
  map<bufferkey_t, buffervalue_t> localCalibrationMap;
 
  for (auto clusterKind : clusterKinds) {
 
    B2INFO("Start calibration of clusterkind=" << clusterKind << " ...");
 
    string gridname = str(format("GridKind_%1%") % clusterKind);
    auto grid = getObjectPtr<TH2F>(gridname);
 
    for (auto uBin = 1; uBin <= grid->GetXaxis()->GetNbins(); uBin++) {
      for (auto vBin = 1; vBin <= grid->GetYaxis()->GetNbins(); vBin++) {
 
        // Bin is centered around angles
        auto thetaU = grid->GetXaxis()->GetBinCenter(uBin);
        auto thetaV = grid->GetYaxis()->GetBinCenter(vBin);
 
        if (thetaV < 0) {
          B2INFO("Skip training estimator on thetaU=" << thetaU << ", thetaV=" << thetaV);
          continue;
        } else {
          B2INFO("Start training estimator on thetaU=" << thetaU << ", thetaV=" << thetaV);
        }
 
        string treename = str(format("tree_%1%_%2%_%3%") % clusterKind % uBin % vBin);
 
        auto localShapeIndexer = PXDClusterShapeIndexPar();
        auto localShapeClassifier = PXDClusterShapeClassifierPar();
        createShapeClassifier(treename, &localShapeClassifier, &localShapeIndexer);
 
        bufferkey_t key = std::make_tuple(clusterKind, uBin, vBin);
        localCalibrationMap[key] = buffervalue_t(localShapeIndexer,  localShapeClassifier);
      }
    }
  }
 
  // Create a ShapeIndexer payload
  PXDClusterShapeIndexPar* shapeIndexer = new PXDClusterShapeIndexPar();
 
  int globalShapeIndex = 0;
  for (auto it = m_shapeSet.begin(); it != m_shapeSet.end(); ++it) {
    shapeIndexer->addShape(*it, globalShapeIndex);
    globalShapeIndex++;
  }
 
  B2INFO("Number of cluster shapes is " << globalShapeIndex);
 
  // Save the cluster shape index table to database.
  saveCalibration(shapeIndexer, "PXDClusterShapeIndexPar");
 
  // Create position estimator
  PXDClusterPositionEstimatorPar* positionEstimator = new PXDClusterPositionEstimatorPar();
 
  for (auto clusterKind : clusterKinds) {
 
    string gridname = str(format("GridKind_%1%") % clusterKind);
    auto grid = getObjectPtr<TH2F>(gridname);
 
    positionEstimator->addGrid(clusterKind, *grid);
 
    for (auto uBin = 1; uBin <= grid->GetXaxis()->GetNbins(); uBin++) {
      for (auto vBin = 1; vBin <= grid->GetYaxis()->GetNbins(); vBin++) {
 
        // Bin is centered around angles
        auto thetaU = grid->GetXaxis()->GetBinCenter(uBin);
        auto thetaV = grid->GetYaxis()->GetBinCenter(vBin);
 
        if (thetaV < 0) {
          // We skipped this part in training before
          continue;
        }
 
        // Find the local calibration results
        auto iter = localCalibrationMap.find(std::make_tuple(clusterKind, uBin, vBin));
        auto localShapeIndexer = iter->second.first;
        auto localShapeClassifer = iter->second.second;
 
        // Require that all shape classifiers use a common shape indexer
        auto shapeClassifier = localToGlobal(&localShapeClassifer, &localShapeIndexer, shapeIndexer);
 
        // Mirror the shape classifier along v
        auto mirror_vBin = grid->GetYaxis()->FindBin(-thetaV);
        auto mirroredClassifier = mirrorShapeClassifier(&shapeClassifier, shapeIndexer, clusterKind);
 
        // and fill into position estimator payload
        B2INFO("Add shape classifier for angles thetaU=" << thetaU << ", thetaV=" << thetaV << ", clusterkind=" << clusterKind);
        positionEstimator->setShapeClassifier(shapeClassifier, uBin, vBin, clusterKind);
        B2INFO("Add mirrored shape classifier for angles thetaU=" << thetaU << ", thetaV=" << -thetaV << ", clusterkind=" << clusterKind);
        positionEstimator->setShapeClassifier(mirroredClassifier, uBin, mirror_vBin, clusterKind);
      }
    }
  }
 
  // Save the cluster positions to database.
  saveCalibration(positionEstimator, "PXDClusterPositionEstimatorPar");
 
  B2INFO("PXDClusterPosition Calibration Successful");
  return c_OK;
}

checkPyExpRun()

bool checkPyExpRun ( PyObject * pyObj )

inherited

Checks that a PyObject can be successfully converted to an ExpRun type.

Checks if the PyObject can be converted to ExpRun.

Definition at line 28 of file CalibrationAlgorithm.cc.

{
  // Is it a sequence?
  if (PySequence_Check(pyObj)) {
    Py_ssize_t nObj = PySequence_Length(pyObj);
    // Does it have 2 objects in it?
    if (nObj != 2) {
      B2DEBUG(29, "ExpRun was a Python sequence which didn't have exactly 2 entries!");
      return false;
    }
    PyObject* item1, *item2;
    item1 = PySequence_GetItem(pyObj, 0);
    item2 = PySequence_GetItem(pyObj, 1);
    // Did the GetItem work?
    if ((item1 == NULL) || (item2 == NULL)) {
      B2DEBUG(29, "A PyObject pointer was NULL in the sequence");
      return false;
    }
    // Are they longs?
    if (PyLong_Check(item1) && PyLong_Check(item2)) {
      long value1, value2;
      value1 = PyLong_AsLong(item1);
      value2 = PyLong_AsLong(item2);
      if (((value1 == -1) || (value2 == -1)) && PyErr_Occurred()) {
        B2DEBUG(29, "An error occurred while converting the PyLong to long");
        return false;
      }
    } else {
      B2DEBUG(29, "One or more of the PyObjects in the ExpRun wasn't a long");
      return false;
    }
    // Make sure to kill off the reference GetItem gave us responsibility for
    Py_DECREF(item1);
    Py_DECREF(item2);
  } else {
    B2DEBUG(29, "ExpRun was not a Python sequence.");
    return false;
  }
  return true;
}

clearCalibrationData()

void clearCalibrationData ( )

inlineprotectedinherited

Clear calibration data.

Definition at line 324 of file CalibrationAlgorithm.h.

{m_data.clearCalibrationData();}

commit() [1/2]

bool commit ( )

inherited

Submit constants from last calibration into database.

Definition at line 302 of file CalibrationAlgorithm.cc.

{
  if (getPayloads().empty())
    return false;
  list<Database::DBImportQuery> payloads = getPayloads();
  B2INFO("Committing " << payloads.size()  << " payloads to database.");
  return Database::Instance().storeData(payloads);
}

commit() [2/2]

bool commit ( std::list< Database::DBImportQuery > payloads )

inherited

Submit constants from a (potentially previous) set of payloads.

Definition at line 311 of file CalibrationAlgorithm.cc.

{
  if (payloads.empty())
    return false;
  return Database::Instance().storeData(payloads);
}

convertPyExpRun()

ExpRun convertPyExpRun ( PyObject * pyObj )

inherited

Performs the conversion of PyObject to ExpRun.

Converts the PyObject to an ExpRun. We've preoviously checked the object so this assumes a lot about the PyObject.

Definition at line 70 of file CalibrationAlgorithm.cc.

{
  ExpRun expRun;
  PyObject* itemExp, *itemRun;
  itemExp = PySequence_GetItem(pyObj, 0);
  itemRun = PySequence_GetItem(pyObj, 1);
  expRun.first = PyLong_AsLong(itemExp);
  Py_DECREF(itemExp);
  expRun.second = PyLong_AsLong(itemRun);
  Py_DECREF(itemRun);
  return expRun;
}

createShapeClassifier()

void createShapeClassifier ( std::string treename, PXDClusterShapeClassifierPar * shapeClassifier, PXDClusterShapeIndexPar * shapeIndexer )

private

Returns a new classifier and index trained on cluster tree.

Definition at line 256 of file PXDClusterPositionCalibrationAlgorithm.cc.

{
 
  auto tree = getObjectPtr<TTree>(treename);
 
  const auto nEntries = tree->GetEntries();
  B2INFO("Number of clusters is " << nEntries);
 
  string* shapeNamePtr = &m_shapeName;
  string* mirroredShapeNamePtr = &m_mirroredShapeName;
 
  tree->SetBranchAddress("ShapeName", &shapeNamePtr);
  tree->SetBranchAddress("MirroredShapeName", &mirroredShapeNamePtr);
  tree->SetBranchAddress("ClusterEta", &m_clusterEta);
  tree->SetBranchAddress("OffsetU", &m_positionOffsetU);
  tree->SetBranchAddress("OffsetV", &m_positionOffsetV);
  tree->SetBranchAddress("SizeV", &m_sizeV);
 
  // Vector to enumerate all shapes by unique name and count their
  // occurrence in training data.
  vector< pair<string, float> > shapeList;
 
  for (int i = 0; i < nEntries; ++i) {
    tree->GetEntry(i);
 
    string shapeName = m_shapeName;
 
    auto it = std::find_if(shapeList.begin(), shapeList.end(),
    [&](const pair<string, float>& element) { return element.first == shapeName;});
 
    //Shape name exists in vector
    if (it != shapeList.end()) {
      //increment key in map
      it->second++;
    }
    //Shape name does not exist
    else {
      //Not found, insert in vector
      shapeList.push_back(pair<string, int>(shapeName, 1));
      // Remember the relation between name of a shape and name of mirrored shape
      m_mirrorMap[shapeName] = m_mirroredShapeName;
      // Remember the relation between name of a shape and its size
      m_sizeMap[shapeName] = m_sizeV;
    }
  }
 
  // Loop over shapeList to select shapes for
  // next calibration step
 
  // Vector with eta histograms for selected shapes
  vector< pair<string, TH1D> > etaHistos;
 
  // Index for enumerating selected shapes
  int tmpIndex = 0;
 
  // Coverage of position offsets on training data
  double coverage = 0.0;
 
  for (auto iter : shapeList) {
    auto name = iter.first;
    auto counter = iter.second;
 
    double likelyhood = counter / nEntries;
 
    if (counter >=  minClusterForShapeLikelyhood) {
      //B2INFO("Adding shape " << name << " with index " << tmpIndex << " and shape likelyhood " << 100*likelyhood << "% and count " << counter);
      shapeIndexer->addShape(name, tmpIndex);
      shapeClassifier->addShapeLikelyhood(tmpIndex, likelyhood);
 
      // Add name of shape to global (all clusterkinds + all angle bins) shape set
      m_shapeSet.insert(name);
      // Add name of mirrored shape as well
      m_shapeSet.insert(m_mirrorMap[name]);
      // Increment the index
      tmpIndex++;
    }
 
    if (counter >=  minClusterForPositionOffset) {
      coverage += likelyhood;
      string etaname = str(format("eta_%1%") % name);
 
      // Single pixel case: Eta value is cluster charge
      if (name == "SD0.0") {
        TH1D etaHisto(etaname.c_str(), etaname.c_str(), 255, 0, 255);
        etaHisto.SetDirectory(0);
        etaHistos.push_back(pair<string, TH1D>(name, etaHisto));
      }
      // Multipixel case: Eta value is ratio head/(tail+head) of charges (to be less gain sensitive)
      else {
        TH1D etaHisto(etaname.c_str(), etaname.c_str(), 301, 0, 1);
        etaHisto.SetDirectory(0);
        etaHistos.push_back(pair<string, TH1D>(name, etaHisto));
      }
    }
  }
 
  B2INFO("Offset coverage is "  << 100 * coverage << "%");
 
  // Loop over the tree is to fill the eta histograms for
  // selected shapes.
 
  for (int i = 0; i < nEntries; ++i) {
    tree->GetEntry(i);
 
    string shapeName = m_shapeName;
    auto it = std::find_if(etaHistos.begin(), etaHistos.end(),
    [&](const pair<string, TH1D>& element) { return element.first == shapeName;});
    //Item exists in map
    if (it != etaHistos.end()) {
      // increment key in map
      it->second.Fill(m_clusterEta);
    }
  }
 
  // Vector for offset histograms stored by offset shape name and eta bin
  vector< pair< string, vector<TH2D> > > offsetHistosVec;
 
  for (auto iter : etaHistos) {
    auto name = iter.first;
    auto& histo = iter.second;
    int nClusters = histo.GetEntries();
 
    // Add shape for offset correction
    int shapeIndex = shapeIndexer->getShapeIndex(name);
    shapeClassifier->addShape(shapeIndex);
 
    // Try to split clusters into n bins with minClusterForPositionOffset clusters
    int nEtaBins  = std::max(int(nClusters / minClusterForPositionOffset), 1);
    nEtaBins =  std::min(nEtaBins, maxEtaBins);
 
    //B2INFO("SHAPE NAME:" << name << " WITH BINS " << nEtaBins);
 
    vector< TH2D > offsetHistos;
 
    for (int i = 0; i < nEtaBins; i++) {
      // Position where to compute the quantiles in [0,1]
      double xq = double(i) / nEtaBins;
      // Double to contain the quantile
      double yq = 0;
      histo.GetQuantiles(1, &yq, &xq);
      //B2INFO("   Quantile at xq =" << xq << " is yq=" << yq);
      shapeClassifier->addEtaPercentile(shapeIndex, yq);
 
      string offsetname = str(format("offset_%1%_%2%") % name % i);
      TH2D offsetHisto(offsetname.c_str(), offsetname.c_str(), 1, 0, 1, 1, 0, 1);
      offsetHisto.SetDirectory(0);
      offsetHisto.StatOverflows();
      offsetHistos.push_back(offsetHisto);
 
    }
    offsetHistosVec.push_back(pair< string, vector<TH2D> >(name, offsetHistos));
  }
 
  // Loop over the tree is to fill offset histograms
 
  for (int i = 0; i < nEntries; ++i) {
    tree->GetEntry(i);
 
    string shapeName = m_shapeName;
    auto it = std::find_if(offsetHistosVec.begin(), offsetHistosVec.end(),
    [&](const pair<string, vector<TH2D>>& element) { return element.first == shapeName;});
    //Item exists in map
    if (it != offsetHistosVec.end()) {
      int shapeIndex = shapeIndexer->getShapeIndex(shapeName);
      int etaBin = shapeClassifier->getEtaIndex(shapeIndex, m_clusterEta);
      it->second.at(etaBin).Fill(m_positionOffsetU, m_positionOffsetV);
    }
  }
 
  // Compute the moments of the offset histograms and finalize the shape classifier object
 
  // Loop over shape names
  for (auto iter : offsetHistosVec) {
    auto name = iter.first;
    auto& histovec = iter.second;
 
    int shapeIndex = shapeIndexer->getShapeIndex(name);
 
    // Loop over eta bins
    for (auto& histo : histovec) {
      // Compute offset moments
      double etaLikelyhood = double(histo.GetEntries()) / nEntries;
      double offsetU = histo.GetMean(1);
      double offsetV = histo.GetMean(2);
      double covUV = histo.GetCovariance();
      double covU = pow(histo.GetRMS(1), 2);
      double covV = pow(histo.GetRMS(2), 2);
 
      B2INFO("Name " << name  << ", posU=" << offsetU << ", posV=" << offsetV << ", covU=" << covU << ", covV=" << covV << ", covUV=" <<
             covUV);
 
      TMatrixDSym HitCov(2);
      HitCov(0, 0) = covU;
      HitCov(1, 0) = covUV;
      HitCov(0, 1) = covUV;
      HitCov(1, 1) = covV;
 
      TMatrixDSymEigen HitCovE(HitCov);
      TVectorD eigenval = HitCovE.GetEigenValues();
      if (eigenval(0) <= 0 || eigenval(1) <= 0) {
        B2ERROR("Estimated covariance matrix not positive definite.");
      }
 
      auto offset = PXDClusterOffsetPar(offsetU, offsetV, covU, covV, covUV);
      shapeClassifier->addEtaLikelyhood(shapeIndex, etaLikelyhood);
      shapeClassifier->addEtaOffset(shapeIndex, offset);
    }
  }
 
  B2INFO("Added shape classifier with coverage " << 100 * coverage << "% on training data sample.");
 
  return;
}

dumpOutputJson()

const std::string dumpOutputJson ( ) const

inlineinherited

Dump the JSON string of the output JSON object.

Definition at line 223 of file CalibrationAlgorithm.h.

{return m_jsonExecutionOutput.dump();}

execute() [1/2]

CalibrationAlgorithm::EResult execute ( PyObject * runs, int iteration = 0, IntervalOfValidity iov = IntervalOfValidity() )

inherited

Runs calibration over Python list of runs. Converts to C++ and then calls the other execute() function.

Definition at line 83 of file CalibrationAlgorithm.cc.

{
  B2DEBUG(29, "Running execute() using Python Object as input argument");
  // Reset the execution specific data in case the algorithm was previously called
  m_data.reset();
  m_data.setIteration(iteration);
  vector<ExpRun> vecRuns;
  // Is it a list?
  if (PySequence_Check(runs)) {
    boost::python::handle<> handle(boost::python::borrowed(runs));
    boost::python::list listRuns(handle);
 
    int nList = boost::python::len(listRuns);
    for (int iList = 0; iList < nList; ++iList) {
      boost::python::object pyExpRun(listRuns[iList]);
      if (!checkPyExpRun(pyExpRun.ptr())) {
        B2ERROR("Received Python ExpRuns couldn't be converted to C++");
        m_data.setResult(c_Failure);
        return c_Failure;
      } else {
        vecRuns.push_back(convertPyExpRun(pyExpRun.ptr()));
      }
    }
  } else {
    B2ERROR("Tried to set the input runs but we didn't receive a Python sequence object (list,tuple).");
    m_data.setResult(c_Failure);
    return c_Failure;
  }
  return execute(vecRuns, iteration, iov);
}

execute() [2/2]

CalibrationAlgorithm::EResult execute ( std::vector< Calibration::ExpRun > runs = {}, int iteration = 0, IntervalOfValidity iov = IntervalOfValidity() )

inherited

Runs calibration over vector of runs for a given iteration.

You can also specify the IoV to save the database payload as. By default the Algorithm will create an IoV from your requested ExpRuns, or from the overall ExpRuns of the input data if you haven't specified ExpRuns in this function.

No checks are performed to make sure that a IoV you specify matches the data you ran over, it simply labels the IoV to commit to the database later.

Definition at line 114 of file CalibrationAlgorithm.cc.

{
  // Check if we are calling this function directly and need to reset, or through Python where it was already done.
  if (m_data.getResult() != c_Undefined) {
    m_data.reset();
    m_data.setIteration(iteration);
  }
 
  if (m_inputFileNames.empty()) {
    B2ERROR("There aren't any input files set. Please use CalibrationAlgorithm::setInputFiles()");
    m_data.setResult(c_Failure);
    return c_Failure;
  }
 
  // Did we receive runs to execute over explicitly?
  if (!(runs.empty())) {
    for (auto expRun : runs) {
      B2DEBUG(29, "ExpRun requested = (" << expRun.first << ", " << expRun.second << ")");
    }
    // We've asked explicitly for certain runs, but we should check if the data granularity is 'run'
    if (strcmp(getGranularity().c_str(), "all") == 0) {
      B2ERROR(("The data is collected with granularity=all (exp=-1,run=-1), but you seem to request calibration for specific runs."
               " We'll continue but using ALL the input data given instead of the specific runs requested."));
    }
  } else {
    // If no runs are provided, infer the runs from all collected data
    runs = getRunListFromAllData();
    // Let's check that we have some now
    if (runs.empty()) {
      B2ERROR("No collected data in input files.");
      m_data.setResult(c_Failure);
      return c_Failure;
    }
    for (auto expRun : runs) {
      B2DEBUG(29, "ExpRun requested = (" << expRun.first << ", " << expRun.second << ")");
    }
  }
 
  m_data.setRequestedRuns(runs);
  if (iov.empty()) {
    // If no user specified IoV we use the IoV from the executed run list
    iov = IntervalOfValidity(runs[0].first, runs[0].second, runs[runs.size() - 1].first, runs[runs.size() - 1].second);
  }
  m_data.setRequestedIov(iov);
  // After here, the getObject<...>(...) helpers start to work
 
  CalibrationAlgorithm::EResult result = calibrate();
  m_data.setResult(result);
  return result;
}

fillRunToInputFilesMap()

void fillRunToInputFilesMap ( )

inherited

Fill the mapping of ExpRun -> Files.

Definition at line 330 of file CalibrationAlgorithm.cc.

{
  m_runsToInputFiles.clear();
  // Save TDirectory to change back at the end
  TDirectory* dir = gDirectory;
  RunRange* runRange;
  // Construct the TDirectory name where we expect our objects to be
  string runRangeObjName(getPrefix() + "/" + RUN_RANGE_OBJ_NAME);
  for (const auto& fileName : m_inputFileNames) {
    //Open TFile to get the objects
    unique_ptr<TFile> f;
    f.reset(TFile::Open(fileName.c_str(), "READ"));
    runRange = dynamic_cast<RunRange*>(f->Get(runRangeObjName.c_str()));
    if (runRange) {
      // Insert or extend the run -> file mapping for this ExpRun
      auto expRuns = runRange->getExpRunSet();
      for (const auto& expRun : expRuns) {
        auto runFiles = m_runsToInputFiles.find(expRun);
        if (runFiles != m_runsToInputFiles.end()) {
          (runFiles->second).push_back(fileName);
        } else {
          m_runsToInputFiles.insert(std::make_pair(expRun, std::vector<std::string> {fileName}));
        }
      }
    } else {
      B2WARNING("Missing a RunRange object for file: " << fileName);
    }
  }
  dir->cd();
}

findPayloadBoundaries()

const std::vector< ExpRun > findPayloadBoundaries ( std::vector< Calibration::ExpRun > runs, int iteration = 0 )

inherited

Used to discover the ExpRun boundaries that you want the Python CAF to execute on. This is optional and only used in some.

Definition at line 520 of file CalibrationAlgorithm.cc.

{
  m_boundaries.clear();
  if (m_inputFileNames.empty()) {
    B2ERROR("There aren't any input files set. Please use CalibrationAlgorithm::setInputFiles()");
    return m_boundaries;
  }
  // Reset the internal execution data just in case something is hanging around
  m_data.reset();
  if (runs.empty()) {
    // Want to loop over all runs we could possibly know about
    runs = getRunListFromAllData();
  }
  // Let's check that we have some now
  if (runs.empty()) {
    B2ERROR("No collected data in input files.");
    return m_boundaries;
  }
  // In order to find run boundaries we must have collected with data granularity == 'run'
  if (strcmp(getGranularity().c_str(), "all") == 0) {
    B2ERROR("The data is collected with granularity='all' (exp=-1,run=-1), and we can't use that to find run boundaries.");
    return m_boundaries;
  }
  m_data.setIteration(iteration);
  // User defined setup function
  boundaryFindingSetup(runs, iteration);
  std::vector<ExpRun> runList;
  // Loop over run list and call derived class "isBoundaryRequired" member function
  for (auto currentRun : runs) {
    runList.push_back(currentRun);
    m_data.setRequestedRuns(runList);
    // After here, the getObject<...>(...) helpers start to work
    if (isBoundaryRequired(currentRun)) {
      m_boundaries.push_back(currentRun);
    }
    // Only want run-by-run
    runList.clear();
    // Don't want memory hanging around
    m_data.clearCalibrationData();
  }
  m_data.reset();
  boundaryFindingTearDown();
  return m_boundaries;
}

getAllGranularityExpRun()

Calibration::ExpRun getAllGranularityExpRun ( ) const

inlineprotectedinherited

Returns the Exp,Run pair that means 'Everything'. Currently unused.

Definition at line 327 of file CalibrationAlgorithm.h.

{return m_allExpRun;}

getCollectorName()

std::string getCollectorName ( ) const

inlineinherited

Alias for prefix.

For convenience and less writing, we say developers to set this to default collector module name in constructor of base class. One can however use the dublets of collector+algorithm multiple times with different settings. To bind these together correctly, the prefix has to be set the same for algo and collector. So we call the setter setPrefix rather than setModuleName or whatever. This getter will work out of the box for default cases -> return the name of module you have to add to your path to collect data for this algorithm.

Definition at line 164 of file CalibrationAlgorithm.h.

{return getPrefix();}

getDescription()

const std::string & getDescription ( ) const

inlineinherited

Get the description of the algorithm (set by developers in constructor)

Definition at line 216 of file CalibrationAlgorithm.h.

{return m_description;}

getExpRunString()

string getExpRunString ( Calibration::ExpRun & expRun ) const

privateinherited

Gets the "exp.run" string repr. of (exp,run)

Definition at line 254 of file CalibrationAlgorithm.cc.

{
  string expRunString;
  expRunString += to_string(expRun.first);
  expRunString += ".";
  expRunString += to_string(expRun.second);
  return expRunString;
}

getFullObjectPath()

string getFullObjectPath ( const std::string & name, Calibration::ExpRun expRun ) const

privateinherited

constructs the full TDirectory + Key name of an object in a TFile based on its name and exprun

Definition at line 263 of file CalibrationAlgorithm.cc.

{
  string dirName = getPrefix() + "/" + name;
  string objName = name + "_" + getExpRunString(expRun);
  return dirName + "/" + objName;
}

getGranularity()

std::string getGranularity ( ) const

inlineinherited

Get the granularity of collected data.

Definition at line 188 of file CalibrationAlgorithm.h.

{return m_granularityOfData;};

getGranularityFromData()

string getGranularityFromData ( ) const

protectedinherited

Get the granularity of collected data.

Definition at line 383 of file CalibrationAlgorithm.cc.

{
  // Save TDirectory to change back at the end
  TDirectory* dir = gDirectory;
  RunRange* runRange;
  string runRangeObjName(getPrefix() + "/" + RUN_RANGE_OBJ_NAME);
  // We only check the first file
  string fileName = m_inputFileNames[0];
  unique_ptr<TFile> f;
  f.reset(TFile::Open(fileName.c_str(), "READ"));
  runRange = dynamic_cast<RunRange*>(f->Get(runRangeObjName.c_str()));
  if (!runRange) {
    B2FATAL("The input file " << fileName << " does not contain a RunRange object at "
            << runRangeObjName << ". Please set your input files to exclude it.");
    return "";
  }
  string granularity = runRange->getGranularity();
  dir->cd();
  return granularity;
}

getInputFileNames()

PyObject * getInputFileNames ( )

inherited

Get the input file names used for this algorithm and pass them out as a Python list of unicode strings.

Definition at line 245 of file CalibrationAlgorithm.cc.

{
  PyObject* objInputFileNames = PyList_New(m_inputFileNames.size());
  for (size_t i = 0; i < m_inputFileNames.size(); ++i) {
    PyList_SetItem(objInputFileNames, i, Py_BuildValue("s", m_inputFileNames[i].c_str()));
  }
  return objInputFileNames;
}

getInputJsonObject()

const nlohmann::json & getInputJsonObject ( ) const

inlineprotectedinherited

Get the entire top level JSON object. We explicitly say this must be of object type so that we might pick.

Definition at line 357 of file CalibrationAlgorithm.h.

{return m_jsonExecutionInput;}

getInputJsonValue()

template<class T>

const T getInputJsonValue ( const std::string & key ) const

inlineprotectedinherited

Get an input JSON value using a key. The normal exceptions are raised when the key doesn't exist.

Definition at line 350 of file CalibrationAlgorithm.h.

    {
      return m_jsonExecutionInput.at(key);
    }

getIovFromAllData()

IntervalOfValidity getIovFromAllData ( ) const

inherited

Get the complete IoV from inspection of collected data.

Definition at line 325 of file CalibrationAlgorithm.cc.

{
  return getRunRangeFromAllData().getIntervalOfValidity();
}

getIteration()

int getIteration ( ) const

inlineprotectedinherited

Get current iteration.

Definition at line 269 of file CalibrationAlgorithm.h.

{ return m_data.getIteration(); }

getObjectPtr()

template<class T>

std::shared_ptr< T > getObjectPtr ( std::string name )

inlineprotectedinherited

Get calibration data object (for all runs the calibration is requested for) This function will only work during or after execute() has been called once.

Definition at line 285 of file CalibrationAlgorithm.h.

    {
      if (m_runsToInputFiles.size() == 0)
        fillRunToInputFilesMap();
      return getObjectPtr<T>(name, m_data.getRequestedRuns());
    }

getOutputJsonValue()

template<class T>

const T getOutputJsonValue ( const std::string & key ) const

inlineprotectedinherited

Get a value using a key from the JSON output object, not sure why you would want to do this.

Definition at line 342 of file CalibrationAlgorithm.h.

    {
      return m_jsonExecutionOutput.at(key);
    }

getPayloads()

std::list< Database::DBImportQuery > & getPayloads ( )

inlineinherited

Get constants (in TObjects) for database update from last execution.

Definition at line 204 of file CalibrationAlgorithm.h.

{return m_data.getPayloads();}

getPayloadValues()

std::list< Database::DBImportQuery > getPayloadValues ( )

inlineinherited

Get constants (in TObjects) for database update from last execution but passed by VALUE.

Definition at line 207 of file CalibrationAlgorithm.h.

{return m_data.getPayloadValues();}

getPrefix()

std::string getPrefix ( ) const

inlineinherited

Get the prefix used for getting calibration data.

Definition at line 146 of file CalibrationAlgorithm.h.

{return m_prefix;}

getRunList()

const std::vector< Calibration::ExpRun > & getRunList ( ) const

inlineprotectedinherited

Get the list of runs for which calibration is called.

Definition at line 266 of file CalibrationAlgorithm.h.

{return m_data.getRequestedRuns();}

getRunListFromAllData()

vector< ExpRun > getRunListFromAllData ( ) const

inherited

Get the complete list of runs from inspection of collected data.

Definition at line 318 of file CalibrationAlgorithm.cc.

{
  RunRange runRange = getRunRangeFromAllData();
  set<ExpRun> expRunSet = runRange.getExpRunSet();
  return vector<ExpRun>(expRunSet.begin(), expRunSet.end());
}

getRunRangeFromAllData()

RunRange getRunRangeFromAllData ( ) const

inherited

Get the complete RunRange from inspection of collected data.

Definition at line 361 of file CalibrationAlgorithm.cc.

{
  // Save TDirectory to change back at the end
  TDirectory* dir = gDirectory;
  RunRange runRange;
  // Construct the TDirectory name where we expect our objects to be
  string runRangeObjName(getPrefix() + "/" + RUN_RANGE_OBJ_NAME);
  for (const auto& fileName : m_inputFileNames) {
    //Open TFile to get the objects
    unique_ptr<TFile> f;
    f.reset(TFile::Open(fileName.c_str(), "READ"));
    RunRange* runRangeOther = dynamic_cast<RunRange*>(f->Get(runRangeObjName.c_str()));
    if (runRangeOther) {
      runRange.merge(runRangeOther);
    } else {
      B2WARNING("Missing a RunRange object for file: " << fileName);
    }
  }
  dir->cd();
  return runRange;
}

getVecInputFileNames()

std::vector< std::string > getVecInputFileNames ( ) const

inlineprotectedinherited

Get the input file names used for this algorithm as a STL vector.

Definition at line 275 of file CalibrationAlgorithm.h.

{return m_inputFileNames;}

inputJsonKeyExists()

bool inputJsonKeyExists ( const std::string & key ) const

inlineprotectedinherited

Test for a key in the input JSON object.

Definition at line 360 of file CalibrationAlgorithm.h.

{return m_jsonExecutionInput.count(key);}

isBoundaryRequired()

virtual bool isBoundaryRequired ( const Calibration::ExpRun & )

inlineprotectedvirtualinherited

Given the current collector data, make a decision about whether or not this run should be the start of a payload boundary.

Reimplemented in PXDAnalyticGainCalibrationAlgorithm, PXDValidationAlgorithm, SVD3SampleCoGTimeCalibrationAlgorithm, SVD3SampleELSTimeCalibrationAlgorithm, SVDCoGTimeCalibrationAlgorithm, TestBoundarySettingAlgorithm, and TestCalibrationAlgorithm.

Definition at line 243 of file CalibrationAlgorithm.h.

    {
      B2ERROR("You didn't implement a isBoundaryRequired() member function in your CalibrationAlgorithm but you are calling it!");
      return false;
    }

loadInputJson()

bool loadInputJson ( const std::string & jsonString )

inherited

Load the m_inputJson variable from a string (useful from Python interface). The return bool indicates success or failure.

Definition at line 502 of file CalibrationAlgorithm.cc.

{
  try {
    auto jsonInput = nlohmann::json::parse(jsonString);
    // Input string has an object (dict) as the top level object?
    if (jsonInput.is_object()) {
      m_jsonExecutionInput = jsonInput;
      return true;
    } else {
      B2ERROR("JSON input string isn't an object type i.e. not a '{}' at the top level.");
      return false;
    }
  } catch (nlohmann::json::parse_error&) {
    B2ERROR("Parsing of JSON input string failed");
    return false;
  }
}

localToGlobal()

PXDClusterShapeClassifierPar localToGlobal ( PXDClusterShapeClassifierPar * localShapeClassifier, PXDClusterShapeIndexPar * localShapeIndexer, PXDClusterShapeIndexPar * globalShapeIndexer )

private

Returns a shape classifier using global shape indices instead of local ones.

Definition at line 211 of file PXDClusterPositionCalibrationAlgorithm.cc.

{
  // Create a shape classifier using global shape indices
  auto globalShapeClassifier = PXDClusterShapeClassifierPar();
 
  // Re-index the the shape likelyhood map
  auto shapeLikelyhoodMap = shapeClassifier->getShapeLikelyhoodMap();
  for (auto indexAndValue : shapeLikelyhoodMap) {
    // Compute the global shape index
    auto shapeIndex = indexAndValue.first;
    auto shapeName = shapeIndexer->getShapeName(shapeIndex);
    auto globalIndex = globalShapeIndexer->getShapeIndex(shapeName);
    // Store the result
    globalShapeClassifier.addShapeLikelyhood(globalIndex, indexAndValue.second);
  }
 
  // Re-index the offset related maps
  auto offsetMap = shapeClassifier->getOffsetMap();
  auto percentileMap = shapeClassifier->getPercentilesMap();
  auto likelyhoodMap = shapeClassifier->getLikelyhoodMap();
  for (auto indexAndValue : offsetMap) {
    // Compute the global shape index
    auto shapeIndex = indexAndValue.first;
    auto shapeName = shapeIndexer->getShapeName(shapeIndex);
    auto globalIndex = globalShapeIndexer->getShapeIndex(shapeName);
 
    globalShapeClassifier.addShape(globalIndex);
 
    int etaBin = 0;
    for (auto offset : indexAndValue.second) {
      // Copy over percentile
      auto percentile = percentileMap[shapeIndex][etaBin];
      globalShapeClassifier.addEtaPercentile(globalIndex, percentile);
      // Copy over likelyhood
      auto likelyhood = likelyhoodMap[shapeIndex][etaBin];
      globalShapeClassifier.addEtaLikelyhood(globalIndex, likelyhood);
      // Copy over offset
      globalShapeClassifier.addEtaOffset(globalIndex, offset);
      etaBin++;
    }
  }
  return globalShapeClassifier;
}

mirrorShapeClassifier()

PXDClusterShapeClassifierPar mirrorShapeClassifier ( PXDClusterShapeClassifierPar * shapeClassifier, PXDClusterShapeIndexPar * shapeIndexer, int clusterKind )

private

Returns a mirrored version of shape classifier.

Definition at line 160 of file PXDClusterPositionCalibrationAlgorithm.cc.

{
  // Create a mirrored shape classifier
  auto mirroredShapeClassifier = PXDClusterShapeClassifierPar();
 
  // Mirror the shape likelyhood map
  auto shapeLikelyhoodMap = shapeClassifier->getShapeLikelyhoodMap();
  for (auto indexAndValue : shapeLikelyhoodMap) {
    // Compute the mirrored shape index
    auto shapeIndex = indexAndValue.first;
    auto shapeName = shapeIndexer->getShapeName(shapeIndex);
    auto mirroredName = m_mirrorMap[shapeName];
    auto mirroredIndex = shapeIndexer->getShapeIndex(mirroredName);
    // Store the result
    mirroredShapeClassifier.addShapeLikelyhood(mirroredIndex, indexAndValue.second);
  }
 
  // Mirror the offset related maps
  auto offsetMap = shapeClassifier->getOffsetMap();
  auto percentileMap = shapeClassifier->getPercentilesMap();
  auto likelyhoodMap = shapeClassifier->getLikelyhoodMap();
  for (auto indexAndValue : offsetMap) {
    // Compute the mirrored shape index
    auto shapeIndex = indexAndValue.first;
    auto shapeName = shapeIndexer->getShapeName(shapeIndex);
    auto mirroredName = m_mirrorMap[shapeName];
    auto mirroredIndex = shapeIndexer->getShapeIndex(mirroredName);
 
    mirroredShapeClassifier.addShape(mirroredIndex);
 
    int etaBin = 0;
    for (auto offset : indexAndValue.second) {
      // Copy over percentile
      auto percentile = percentileMap[shapeIndex][etaBin];
      mirroredShapeClassifier.addEtaPercentile(mirroredIndex, percentile);
      // Copy over likelyhood
      auto likelyhood = likelyhoodMap[shapeIndex][etaBin];
      mirroredShapeClassifier.addEtaLikelyhood(mirroredIndex, likelyhood);
      // Mirror the offset: v offset shifts and covariance swaps sign
      double shift = (m_sizeMap[shapeName] - 1) * m_pitchMap[clusterKind];
      auto mirroredOffset = PXDClusterOffsetPar(offset.getU(), shift - offset.getV(), offset.getUSigma2(), offset.getVSigma2(),
                                                -offset.getUVCovariance());
      mirroredShapeClassifier.addEtaOffset(mirroredIndex, mirroredOffset);
      etaBin++;
    }
  }
 
  return mirroredShapeClassifier;
}

resetInputJson()

void resetInputJson ( )

inlineprotectedinherited

Clears the m_inputJson member variable.

Definition at line 330 of file CalibrationAlgorithm.h.

{m_jsonExecutionInput.clear();}

resetOutputJson()

void resetOutputJson ( )

inlineprotectedinherited

Clears the m_outputJson member variable.

Definition at line 333 of file CalibrationAlgorithm.h.

{m_jsonExecutionOutput.clear();}

saveCalibration() [1/6]

void saveCalibration ( TClonesArray * data, const std::string & name )

protectedinherited

Store DBArray payload with given name with default IOV.

Definition at line 297 of file CalibrationAlgorithm.cc.

{
  saveCalibration(data, name, m_data.getRequestedIov());
}

saveCalibration() [2/6]

void saveCalibration ( TClonesArray * data, const std::string & name, const IntervalOfValidity & iov )

protectedinherited

Store DBArray with given name and custom IOV.

Definition at line 276 of file CalibrationAlgorithm.cc.

{
  B2DEBUG(29, "Saving calibration TClonesArray '" <<  name << "' to payloads list.");
  getPayloads().emplace_back(name, data, iov);
}

saveCalibration() [3/6]

void saveCalibration ( TObject * data )

protectedinherited

Store DB payload with default name and default IOV.

Definition at line 287 of file CalibrationAlgorithm.cc.

{
  saveCalibration(data, DataStore::objectName(data->IsA(), ""));
}

saveCalibration() [4/6]

void saveCalibration ( TObject * data, const IntervalOfValidity & iov )

protectedinherited

Store DB payload with default name and custom IOV.

Definition at line 282 of file CalibrationAlgorithm.cc.

{
  saveCalibration(data, DataStore::objectName(data->IsA(), ""), iov);
}

saveCalibration() [5/6]

void saveCalibration ( TObject * data, const std::string & name )

protectedinherited

Store DB payload with given name with default IOV.

Definition at line 292 of file CalibrationAlgorithm.cc.

{
  saveCalibration(data, name, m_data.getRequestedIov());
}

saveCalibration() [6/6]

void saveCalibration ( TObject * data, const std::string & name, const IntervalOfValidity & iov )

protectedinherited

Store DB payload with given name and custom IOV.

Definition at line 270 of file CalibrationAlgorithm.cc.

{
  B2DEBUG(29, "Saving calibration TObject = '" <<  name << "' to payloads list.");
  getPayloads().emplace_back(name, data, iov);
}

setDescription()

void setDescription ( const std::string & description )

inlineprotectedinherited

Set algorithm description (in constructor)

Definition at line 321 of file CalibrationAlgorithm.h.

{m_description = description;}

setInputFileNames() [1/2]

void setInputFileNames ( PyObject * inputFileNames )

inherited

Set the input file names used for this algorithm from a Python list.

Set the input file names used for this algorithm and resolve the wildcards.

Definition at line 166 of file CalibrationAlgorithm.cc.

{
  // The reasoning for this very 'manual' approach to extending the Python interface
  // (instead of using boost::python) is down to my fear of putting off final users with
  // complexity on their side.
  //
  // I didn't want users that inherit from this class to be forced to use boost and
  // to have to define a new python module just to use the CAF. A derived class from
  // from a boost exposed class would need to have its own boost python module definition
  // to allow access from a steering file and to the base class functions (I think).
  // I also couldn't be bothered to write a full framework to get around the issue in a similar
  // way to Module()...maybe there's an easy way.
  //
  // But this way we can allow people to continue using their ROOT implemented classes and inherit
  // easily from this one. But add in a few helper functions that work with Python objects
  // created in their steering file i.e. instead of being forced to use STL objects as input
  // to the algorithm.
  if (PyList_Check(inputFileNames)) {
    boost::python::handle<> handle(boost::python::borrowed(inputFileNames));
    boost::python::list listInputFileNames(handle);
    auto vecInputFileNames = PyObjConvUtils::convertPythonObject(listInputFileNames, vector<string>());
    setInputFileNames(vecInputFileNames);
  } else {
    B2ERROR("Tried to set the input files but we didn't receive a Python list.");
  }
}

setInputFileNames() [2/2]

void setInputFileNames ( std::vector< std::string > inputFileNames )

protectedinherited

Set the input file names used for this algorithm.

Set the input file names used for this algorithm and resolve the wildcards.

Definition at line 194 of file CalibrationAlgorithm.cc.

{
  // A lot of code below is tweaked from RootInputModule::initialize,
  // since we're basically copying the functionality anyway.
  if (inputFileNames.empty()) {
    B2WARNING("You have called setInputFileNames() with an empty list. Did you mean to do that?");
    return;
  }
  auto tmpInputFileNames = RootIOUtilities::expandWordExpansions(inputFileNames);
 
  // We'll use a set to enforce sorted unique file paths as we check them
  set<string> setInputFileNames;
  // Check that files exist and convert to absolute paths
  for (auto path : tmpInputFileNames) {
    string fullPath = fs::absolute(path).string();
    if (fs::exists(fullPath)) {
      setInputFileNames.insert(fs::canonical(fullPath).string());
    } else {
      B2WARNING("Couldn't find the file " << path);
    }
  }
 
  if (setInputFileNames.empty()) {
    B2WARNING("No valid files specified!");
    return;
  } else {
    // Reset the run -> files map as our files are likely different
    m_runsToInputFiles.clear();
  }
 
  // Open TFile to check they can be accessed by ROOT
  TDirectory* dir = gDirectory;
  for (const string& fileName : setInputFileNames) {
    unique_ptr<TFile> f;
    try {
      f.reset(TFile::Open(fileName.c_str(), "READ"));
    } catch (logic_error&) {
      //this might happen for ~invaliduser/foo.root
      //actually undefined behaviour per standard, reported as ROOT-8490 in JIRA
    }
    if (!f || !f->IsOpen()) {
      B2FATAL("Couldn't open input file " + fileName);
    }
  }
  dir->cd();
 
  // Copy the entries of the set to a vector
  m_inputFileNames = vector<string>(setInputFileNames.begin(), setInputFileNames.end());
  m_granularityOfData = getGranularityFromData();
}

setOutputJsonValue()

template<class T>

void setOutputJsonValue ( const std::string & key, const T & value )

inlineprotectedinherited

Set a key:value pair for the outputJson object, expected to used internally during calibrate()

Definition at line 337 of file CalibrationAlgorithm.h.

{m_jsonExecutionOutput[key] = value;}

setPrefix()

void setPrefix ( const std::string & prefix )

inlineinherited

Set the prefix used to identify datastore objects.

Definition at line 167 of file CalibrationAlgorithm.h.

{m_prefix = prefix;}

updateDBObjPtrs()

void updateDBObjPtrs ( const unsigned int event = 1, const int run = 0, const int experiment = 0 )

protectedinherited

Updates any DBObjPtrs by calling update(event) for DBStore.

Definition at line 404 of file CalibrationAlgorithm.cc.

{
  // Construct an EventMetaData object but NOT in the Datastore
  EventMetaData emd(event, run, experiment);
  // Explicitly update while avoiding registering a Datastore object
  DBStore::Instance().update(emd);
  // Also update the intra-run objects to the event at the same time (maybe unnecessary...)
  DBStore::Instance().updateEvent(event);
}

Member Data Documentation

clusterKinds

std::vector<int> clusterKinds

Vector of clusterkinds to calibrate.

Definition at line 42 of file PXDClusterPositionCalibrationAlgorithm.h.

m_allExpRun

const ExpRun m_allExpRun = make_pair(-1, -1)

staticprivateinherited

allExpRun

Definition at line 364 of file CalibrationAlgorithm.h.

m_boundaries

std::vector<Calibration::ExpRun> m_boundaries

protectedinherited

When using the boundaries functionality from isBoundaryRequired, this is used to store the boundaries. It is cleared when.

Definition at line 261 of file CalibrationAlgorithm.h.

m_clusterEta

float m_clusterEta

private

Eta value of cluster.

Definition at line 69 of file PXDClusterPositionCalibrationAlgorithm.h.

m_clusterKind

int m_clusterKind

private

Pitch in V.

Definition at line 81 of file PXDClusterPositionCalibrationAlgorithm.h.

m_data

ExecutionData m_data

privateinherited

Data specific to a SINGLE execution of the algorithm. Gets reset at the beginning of execution.

Definition at line 382 of file CalibrationAlgorithm.h.

m_description

std::string m_description {""}

privateinherited

Description of the algorithm.

Definition at line 385 of file CalibrationAlgorithm.h.

{""};

m_granularityOfData

std::string m_granularityOfData

privateinherited

Granularity of input data. This only changes when the input files change so it isn't specific to an execution.

Definition at line 379 of file CalibrationAlgorithm.h.

m_inputFileNames

std::vector<std::string> m_inputFileNames

privateinherited

List of input files to the Algorithm, will initially be user defined but then gets the wildcards expanded during execute()

Definition at line 373 of file CalibrationAlgorithm.h.

m_jsonExecutionInput

nlohmann::json m_jsonExecutionInput = nlohmann::json::object()

privateinherited

Optional input JSON object used to make decisions about how to execute the algorithm code.

Definition at line 397 of file CalibrationAlgorithm.h.

m_jsonExecutionOutput

nlohmann::json m_jsonExecutionOutput = nlohmann::json::object()

privateinherited

Optional output JSON object that can be set during the execution by the underlying algorithm code.

Definition at line 403 of file CalibrationAlgorithm.h.

m_mirroredShapeName

std::string m_mirroredShapeName

private

Name of mirrored cluster shape.

Definition at line 67 of file PXDClusterPositionCalibrationAlgorithm.h.

m_mirrorMap

std::map<std::string, std::string> m_mirrorMap

private

Helper needed to map the name of a shape to the name of the mirrored shape.

Definition at line 87 of file PXDClusterPositionCalibrationAlgorithm.h.

m_pitchMap

std::map<int, float> m_pitchMap

private

Helper needed to map the clusterkind to the V pitch of the sensor.

Definition at line 85 of file PXDClusterPositionCalibrationAlgorithm.h.

m_pitchV

float m_pitchV

private

Branches for pitchtree.

Pitch in V

Definition at line 79 of file PXDClusterPositionCalibrationAlgorithm.h.

m_positionOffsetU

float m_positionOffsetU

private

Position offset u of cluster.

Definition at line 71 of file PXDClusterPositionCalibrationAlgorithm.h.

m_positionOffsetV

float m_positionOffsetV

private

Position offset v of cluster.

Definition at line 73 of file PXDClusterPositionCalibrationAlgorithm.h.

m_prefix

std::string m_prefix {""}

privateinherited

The name of the TDirectory the collector objects are contained within.

Definition at line 388 of file CalibrationAlgorithm.h.

{""};

m_runsToInputFiles

std::map<Calibration::ExpRun, std::vector<std::string> > m_runsToInputFiles

privateinherited

Map of Runs to input files. Gets filled when you call getRunRangeFromAllData, gets cleared when setting input files again.

Definition at line 376 of file CalibrationAlgorithm.h.

m_shapeName

std::string m_shapeName

private

Branches for tree.

Name of cluster shape

Definition at line 65 of file PXDClusterPositionCalibrationAlgorithm.h.

m_shapeSet

std::set<std::string> m_shapeSet

private

Set of unique shape names.

Definition at line 91 of file PXDClusterPositionCalibrationAlgorithm.h.

m_sizeMap

std::map<std::string, int> m_sizeMap

private

Helper needed to map the name of a shape to the V size of the cluster.

Definition at line 89 of file PXDClusterPositionCalibrationAlgorithm.h.

m_sizeV

int m_sizeV

private

Size in V.

Definition at line 75 of file PXDClusterPositionCalibrationAlgorithm.h.

maxEtaBins

int maxEtaBins

Maximum number of eta bins for estimating cluster position offsets.

Definition at line 39 of file PXDClusterPositionCalibrationAlgorithm.h.

minClusterForPositionOffset

int minClusterForPositionOffset

Minimum number of collected clusters for estimating cluster position offsets.

Definition at line 36 of file PXDClusterPositionCalibrationAlgorithm.h.

minClusterForShapeLikelyhood

int minClusterForShapeLikelyhood

Minimum number of collected clusters for estimating shape likelyhood.

Definition at line 33 of file PXDClusterPositionCalibrationAlgorithm.h.

---

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

• pxd/calibration/include/PXDClusterPositionCalibrationAlgorithm.h
• pxd/calibration/src/PXDClusterPositionCalibrationAlgorithm.cc