[TMVA] fix some inconsistencies and warnings found by doxygen

Fixes several doxygen warnings in TMVA.

Author: ferdymercury

tmva/sofie/README.md

@@ -18,57 +18,68 @@ Build ROOT with the cmake option tmva-sofie enabled.
 cmake ../root -Dtmva-sofie=ON
 make -j8
 ```
-    
+
 ## Usage
 SOFIE works in a parser-generator working architecture. With SOFIE, the user gets an [ONNX](https://github.com/root-project/root/tree/master/tmva/sofie_parsers), [Keras](https://github.com/root-project/root/blob/master/tmva/pymva/src/RModelParser_Keras.cxx) and a [PyTorch](https://github.com/root-project/root/blob/master/tmva/pymva/src/RModelParser_PyTorch.cxx) parser for translating models in respective formats into SOFIE's internal representation.
 
 From ROOT command line, or in a ROOT macro, we can proceed with an ONNX model:
 
-	using namespace TMVA::Experimental;
-	SOFIE::RModelParser_ONNX parser;
-	SOFIE::RModel model = parser.Parse(“./example_model.onnx”);
-	model.Generate();
-	model.OutputGenerated(“./example_output.hxx”);
+```c++
+using namespace TMVA::Experimental;
+SOFIE::RModelParser_ONNX parser;
+SOFIE::RModel model = parser.Parse(“./example_model.onnx”);
+model.Generate();
+model.OutputGenerated(“./example_output.hxx”);
+```
 
 And an C++ header file and a `.dat` file containing the model weights will be generated. You can also use
 
-	model.PrintRequiredInputTensors();
+```c++
+model.PrintRequiredInputTensors();
+```
 
 to check the required size and type of input tensor for that particular model, and use
 
-	model.PrintInitializedTensors();
+```c++
+model.PrintInitializedTensors();
+```
 
 to check the tensors (weights) already included in the model.
 
 To use the generated inference code:
 
-	#include "example_output.hxx"
-	float input[INPUT_SIZE];
+```c++
+#include "example_output.hxx"
+float input[INPUT_SIZE];
+std::vector<float> out = TMVA_SOFIE_example_model::infer(input);
 
-    // Generated header file shall contain a Session class which requires initialization to load the corresponding weights.
-    TMVA_SOFIE_example_model::Session s("example_model.dat")
+// Generated header file shall contain a Session class which requires initialization to load the corresponding weights.
+TMVA_SOFIE_example_model::Session s("example_model.dat")
 
-    // Once instantiated the session object's infer method can be used
-	std::vector<float> out = s.infer(input);
+// Once instantiated the session object's infer method can be used
+std::vector<float> out = s.infer(input);
+```
+
+With the default settings, the weights are contained in a separate binary file, but if the user instead wants them to be in the generated header file itself, they can use approproiate generation options.
 
-With the default settings, the weights are contained in a separate binary file, but if the user instead wants them to be in the generated header file itself, they can use approproiate generation options. 
-    
-    model.Generate(Options::kNoWeightFile);
+```c++
+model.Generate(Options::kNoWeightFile);
+```
 
-Other such options includes `Options::kNoSession` (for not generating the Session class, and instead keeping the infer function independent).  
+Other such options includes `Options::kNoSession` (for not generating the Session class, and instead keeping the infer function independent).
 SOFIE also supports generating inference code with RDataFrame as inputs, refer to the tutorials below for examples.
 
-  
+
 ## Additional Links
 
 - **Tutorials**
-	- [TMVA_SOFIE_Inference](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_Inference.py)
-	- [TMVA_SOFIE_Keras](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_Keras.C)
-	- [TMVA_SOFIE_Keras_HiggsModel](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_Keras_HiggsModel.C)
-	- [TMVA_SOFIE_ONNX](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_ONNX.C)
-	- [TMVA_SOFIE_PyTorch](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_PyTorch.C)
-	- [TMVA_SOFIE_RDataFrame](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RDataFrame.C)
-	- [TMVA_SOFIE_RDataFrame](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RDataFrame.py)
-	- [TMVA_SOFIE_RDataFrame_JIT](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RDataFrame_JIT.C)
-	- [TMVA_SOFIE_RSofieReader](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RSofieReader.C)
+    - [TMVA_SOFIE_Inference](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_Inference.py)
+    - [TMVA_SOFIE_Keras](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_Keras.C)
+    - [TMVA_SOFIE_Keras_HiggsModel](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_Keras_HiggsModel.C)
+    - [TMVA_SOFIE_ONNX](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_ONNX.C)
+    - [TMVA_SOFIE_PyTorch](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_PyTorch.C)
+    - [TMVA_SOFIE_RDataFrame](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RDataFrame.C)
+    - [TMVA_SOFIE_RDataFrame](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RDataFrame.py)
+    - [TMVA_SOFIE_RDataFrame_JIT](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RDataFrame_JIT.C)
+    - [TMVA_SOFIE_RSofieReader](https://github.com/root-project/root/blob/master/tutorials/tmva/TMVA_SOFIE_RSofieReader.C)
 

tmva/tmva/inc/TMVA/DNN/Architectures/Cuda.h

@@ -399,7 +399,7 @@ class TCuda
 
    /** @name Regularization
     * For each regularization type two functions are required, one named
-    * <tt><Type>Regularization</tt> that evaluates the corresponding
+    * <tt>`<Type>`Regularization</tt> that evaluates the corresponding
     * regularization functional for a given weight matrix and the
     * <tt>Add`<Type>`RegularizationGradients</tt>, that adds the regularization
     * component in the gradients to the provided matrix.
@@ -424,7 +424,7 @@ class TCuda
 
    /** @name Initialization
     * For each initialization method, one function in the low-level interface
-    * is provided. The naming scheme is <p>Initialize<Type></p> for a given
+    * is provided. The naming scheme is <p>Initialize`<Type>`</p> for a given
     * initialization method Type.
     */
       ///@{

tmva/tmva/inc/TMVA/DNN/Architectures/Reference.h

@@ -316,10 +316,10 @@ class TReference
    //____________________________________________________________________________
 
    /** @name Regularization
-    * For each regularization type two functions are required, one named
-    * <tt><Type>Regularization</tt> that evaluates the corresponding
+    * For each regularization type, two functions are required, one named
+    * `<Type>Regularization` that evaluates the corresponding
     * regularization functional for a given weight matrix and the
-    * <tt>Add`<Type>`RegularizationGradients</tt>, that adds the regularization
+    * `Add<Type>RegularizationGradients`, that adds the regularization
     * component in the gradients to the provided matrix.
     */
    ///@{
@@ -342,7 +342,7 @@ class TReference
 
    /** @name Initialization
     * For each initialization method, one function in the low-level interface
-    * is provided. The naming scheme is <p>Initialize<Type></p> for a given
+    * is provided. The naming scheme is `Initialize<Type>` for a given
     * initialization method Type.
     */
    ///@{

tmva/tmva/inc/TMVA/Executor.h

@@ -46,9 +46,15 @@ class Executor {
    //////////////////////////////////////
    /// Default constructor of TMVA Executor class
    /// if ROOT::EnableImplicitMT has not been called then by default a serial executor will be created
-   /// A user can create a thread pool and enable multi-thread execution by calling TMVA::Config::Instance()::EnableMT(nthreads)
-   /// For releasing the thread pool used by TMVA one can do it by calling  TMVA::Config::Instance()::DisableMT() or
-   /// calling TMVA::Config::Instance()::EnableMT() with only one thread
+   /// A user can create a thread pool and enable multi-thread excution by calling
+   /// 
+   /// ~~~{.cpp}
+   /// TMVA::Config::Instance()::%EnableMT(int nthreads);
+   /// ~~~
+   ///
+   /// For releasing the thread pool used by TMVA one can do it by calling
+   ///
+   ///     TMVA::Config::Instance()::%DisableMT();
    ////////////////////////////////////////////
    Executor() {
       // enable MT in TMVA if ROOT::IsImplicitMT is enabled

tmva/tmva/src/DNN/Architectures/Cpu/ActivationFunctions.hxx

@@ -79,7 +79,7 @@ void TCpu<AFloat>::ReluDerivative(TCpuTensor<AFloat> & B,
 
 //______________________________________________________________________________
 template<typename AFloat>
-void TCpu<AFloat>::Sigmoid(TCpuTensor<AFloat> & B)
+void TCpu<AFloat>::Sigmoid(TCpu<AFloat>::Tensor_t & B)
 {
    auto f = [](AFloat x) {return 1.0 / (1.0 + exp(-x));};
    B.Map(f);

tmva/tmva/src/DNN/Architectures/Cpu/Dropout.hxx

@@ -23,8 +23,8 @@ namespace DNN  {
 template<typename AFloat>
 void TCpu<AFloat>::DropoutForward(TCpuTensor<AFloat> & A, 
                                   TDescriptors * /*descriptors*/,
-                                  TWorkspace   * /*workspace*/, 
-                                  AFloat dropoutProbability)
+                                  TWorkspace   * /*workspace*/,
+                                  TCpu<AFloat>::Scalar_t dropoutProbability)
 {
    AFloat *data = A.GetData();
 

tmva/tmva/src/DNN/Architectures/Cpu/Initialization.hxx

@@ -150,7 +150,7 @@ void TCpu<AFloat>::InitializeIdentity(TCpuMatrix<AFloat> & A)
 
 //______________________________________________________________________________
 template<typename AFloat>
-void TCpu<AFloat>::InitializeZero(TCpuMatrix<AFloat> & A)
+void TCpu<AFloat>::InitializeZero(TCpu<AFloat>::Matrix_t & A)
 {
    size_t m,n;
    m = A.GetNrows();
@@ -164,7 +164,7 @@ void TCpu<AFloat>::InitializeZero(TCpuMatrix<AFloat> & A)
 }
 //______________________________________________________________________________
 template <typename AFloat>
-void TCpu<AFloat>::InitializeZero(TCpuTensor<AFloat> &A)
+void TCpu<AFloat>::InitializeZero(TCpu<AFloat>::Tensor_t &A)
 {
    size_t n = A.GetSize();
 

tmva/tmva/src/DNN/Architectures/Cpu/OutputFunctions.hxx

@@ -22,8 +22,8 @@ namespace DNN
 {
 
 template<typename AFloat>
-void TCpu<AFloat>::Sigmoid(TCpuMatrix<AFloat> & B,
-                           const TCpuMatrix<AFloat> & A)
+void TCpu<AFloat>::Sigmoid(TCpu<AFloat>::Matrix_t & B,
+                           const TCpu<AFloat>::Matrix_t & A)
 {
    auto f = [](AFloat x) {return 1.0 / (1.0 + exp(-x));};
    B.MapFrom(f, A);

tmva/tmva/src/DNN/Architectures/Cpu/Propagation.hxx

@@ -30,8 +30,8 @@ namespace DNN {
 
 
 template <typename AFloat>
-void TCpu<AFloat>::MultiplyTranspose(TCpuMatrix<AFloat> &output, const TCpuMatrix<AFloat> &input,
-                                     const TCpuMatrix<AFloat> &Weights)
+void TCpu<AFloat>::MultiplyTranspose(TCpu<AFloat>::Matrix_t &output, const TCpu<AFloat>::Matrix_t &input,
+                                     const TCpu<AFloat>::Matrix_t &Weights)
 {
 
    int m = (int)input.GetNrows();
@@ -72,7 +72,7 @@ void TCpu<AFloat>::MultiplyTranspose(TCpuMatrix<AFloat> &output, const TCpuMatri
 }
 
 template <typename AFloat>
-void TCpu<AFloat>::AddRowWise(TCpuMatrix<AFloat> &output, const TCpuMatrix<AFloat> &biases)
+void TCpu<AFloat>::AddRowWise(TCpu<AFloat>::Matrix_t &output, const TCpu<AFloat>::Matrix_t &biases)
 {
 #ifdef R__HAS_TMVACPU
    int m = (int)output.GetNrows();

tmva/tmva/src/DNN/Architectures/Reference/ActivationFunctions.hxx

@@ -73,16 +73,16 @@ inline void TReference<Real_t>::ReluDerivative(TMatrixT<Real_t> & B,
 
 //______________________________________________________________________________
 template<typename Real_t>
-void TReference<Real_t>::Sigmoid(TMatrixT<Real_t> & A)
+void TReference<Real_t>::Sigmoid(TMatrixT<Real_t> & B)
 {
    size_t m,n;
-   m = A.GetNrows();
-   n = A.GetNcols();
+   m = B.GetNrows();
+   n = B.GetNcols();
 
    for (size_t i = 0; i < m; i++) {
       for (size_t j = 0; j < n; j++) {
-         Real_t sig = 1.0 / (1.0 + std::exp(-A(i,j)));
-         A(i,j) = sig;
+         Real_t sig = 1.0 / (1.0 + std::exp(-B(i,j)));
+         B(i,j) = sig;
       }
    }
 }

tmva/tmva/src/DNN/Architectures/Reference/Dropout.hxx

@@ -26,19 +26,19 @@ namespace DNN
 //______________________________________________________________________________
 
 template<typename Real_t>
-void TReference<Real_t>::DropoutForward(TMatrixT<Real_t> & B, TDescriptors*, TWorkspace*, Real_t dropoutProbability)
+void TReference<Real_t>::DropoutForward(TReference<Real_t>::Tensor_t & A, TDescriptors*, TWorkspace*, Real_t dropoutProbability)
 {
    size_t m,n;
-   m = B.GetNrows();
-   n = B.GetNcols();
+   m = A.GetNrows();
+   n = A.GetNcols();
 
    for (size_t i = 0; i < m; i++) {
       for (size_t j = 0; j < n; j++) {
          Real_t r = gRandom->Uniform();
          if (r >= dropoutProbability) {
-            B(i,j) = 0.0;
+            A(i,j) = 0.0;
          } else {
-            B(i,j) /= dropoutProbability;
+            A(i,j) /= dropoutProbability;
          }
       }
    }

tmva/tmva/src/MethodPDERS.cxx

@@ -223,7 +223,7 @@ TMVA::MethodPDERS::~MethodPDERS( void )
 ///    - Unscaled
 ///    - RMS
 ///    - kNN
-///    - Adaptive <default>
+///    - Adaptive `<default>`
 ///
 ///  - KernelEstimator   `<string>`  Kernel estimation function
 ///    available values are: