class TMVA::MethodPDEFoam: public TMVA::MethodBase

 init PDEFoam objects

 constructor from weight file

 PDEFoam can handle classification with multiple classes and regression
 with one or more regression-targets

 default initialization called by all constructors

 Declare MethodPDEFoam options

 options that are used ONLY for the READER to ensure backward compatibility

 process user options

 destructor

 Determine foam range [fXmin, fXmax] for all dimensions, such
 that a fraction of 'fFrac' events lie outside the foam.

 Train PDE-Foam depending on the set options

 Creation of 2 separated foams: one for signal events, one for
 backgound events.  At the end the foam cells of fFoam[0] will
 contain the average number of signal events and fFoam[1] will
 contain the average number of background events.

 Create only one unified foam (fFoam[0]) whose cells contain the
 average discriminator (N_sig)/(N_sig + N_bg)

 Create one unified foam (see TrainUnifiedClassification()) for
 each class, where the cells of foam i (fFoam[i]) contain the
 average fraction of events of class i, i.e.

   D = number events of class i / total number of events

 Training one (mono target regression) foam, whose cells contain
 the average 0th target.  The dimension of the foam = number of
 non-targets (= number of variables).

 Training one (multi target regression) foam, whose cells contain
 the average event density.  The dimension of the foam = number
 of non-targets + number of targets.

 Return Mva-Value.

 In case of 'fSigBgSeparated==false' (one unifiend PDEFoam was
 trained) the function returns the content of the cell, which
 corresponds to the current TMVA::Event, i.e.  D =
 N_sig/(N_bg+N_sig).

 In case of 'fSigBgSeparated==true' (two separate PDEFoams were
 trained) the function returns

    D = Density_sig/(Density_sig+Density_bg)

 where 'Density_sig' is the content of the cell in the signal
 PDEFoam (fFoam[0]) and 'Density_bg' is the content of the cell
 in the background PDEFoam (fFoam[1]).

 In both cases the error on the discriminant is stored in 'err'
 and 'errUpper'.  (Of course err and errUpper must be non-zero
 and point to valid address to make this work.)

 Calculate the error on the Mva value

 If fSigBgSeparated == true the error is calculated from the
 number of events in the signal and background PDEFoam cells.

 If fSigBgSeparated == false, the error is taken directly from
 the PDEFoam cell.

 Get the multiclass MVA response for the PDEFoam classifier.  The
 returned MVA values are normalized, i.e. their sum equals 1.

 Compute ranking of input variables from the number of cuts made
 in each PDEFoam dimension.  The PDEFoam dimension (the variable)
 for which the most cuts were done is ranked highest.

 Fill in 'nCuts' the number of cuts made in every foam dimension,
 starting at the root cell 'cell'.

 Parameters:

 - cell - root cell to start the counting from

 - nCuts - the number of cuts are saved in this vector

 Set Xmin, Xmax for every dimension in the given pdefoam object

 Create a new PDEFoam, set the PDEFoam options (nCells, nBin,
 Xmin, Xmax, etc.) and initialize the PDEFoam by calling
 pdefoam->Initialize().

 Parameters:

 - foamcaption - name of PDEFoam object

 - ft - type of PDEFoam
   Candidates are:
     - kSeparate    - creates TMVA::PDEFoamEvent
     - kDiscr       - creates TMVA::PDEFoamDiscriminant
     - kMonoTarget  - creates TMVA::PDEFoamTarget
     - kMultiTarget - creates TMVA::MultiTarget
     - kMultiClass  - creates TMVA::PDEFoamDiscriminant

   If 'fDTSeparation != kFoam' then a TMVA::PDEFoamDecisionTree
   is created (the separation type depends on fDTSeparation).

 - cls - marked event class (optional, default value = 0)

 Return regression values for both multi- and mono-target regression

 create a pdefoam kernel estimator, depending on the current
 value of fKernel

 Deletes all trained foams

 reset MethodPDEFoam:
 - delete all PDEFoams
 - delete the kernel estimator

 create XML output of PDEFoam method variables

 Write PDEFoams to file

 read options and internal parameters

 read PDEFoam variables from xml weight file

 Reads a foam with name 'foamname' from file, and returns a clone
 of the foam.  The given ROOT file must be open.  (The ROOT file
 will not be closed in this function.)

 Parameters:

 - file - an open ROOT file

 - foamname - name of foam to load from the file

 Returns:

 If a foam with name 'foamname' exists in the file, then it is
 read from the file, cloned and returned.  If a foam with name
 'foamname' does not exist in the file or the clone operation
 does not succeed, then NULL is returned.

 read foams from file

 convert UInt_t to ETargetSelection (used for reading weight files)

 store the variable names in all foams

 write PDEFoam-specific classifier response
 NOT IMPLEMENTED YET!

 provide help message

 helper functions to convert enum types to UInt_t and back