changes for integration into proxySEM

.gitmodules

@@ -1,6 +1,6 @@
 [submodule "cmake/blt"]
 	path = cmake/blt
-	url = ../../LLNL/blt.git
+	url = https://github.com/LLNL/blt.git
 [submodule "tpl/camp"]
 	path = tpl/camp
-	url = ../../LLNL/camp.git
+	url = https://github.com/LLNL/camp.git

CMakeLists.txt

@@ -15,12 +15,12 @@ set( SHIVA_VERSION_PATCHLEVEL 0 )
 # check if Shiva is build as a submodule or a separate project
 get_directory_property( parent_dir PARENT_DIRECTORY )
 if(parent_dir)
-    set( is_submodule ON )
+    set( SHIVA_IS_SUBMODULE ON )
 else()
-    set( is_submodule OFF )
+    set( SHIVA_IS_SUBMODULE OFF )
 endif()
 
-if( NOT is_submodule )
+if( NOT SHIVA_IS_SUBMODULE )
     message( "not a submodule")
     project( Shiva LANGUAGES CXX C )
 
@@ -66,12 +66,22 @@ include( cmake/Macros.cmake )
 include( cmake/Config.cmake )
 
 
+set(SHIVA_SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR} )
+set(SHIVA_BINARY_DIR ${CMAKE_CURRENT_BINARY_DIR} )
+
+message( STATUS "SHIVA_BINARY_DIR: ${SHIVA_BINARY_DIR}" )
+message( STATUS "SHIVA_SOURCE_DIR: ${SHIVA_SOURCE_DIR}" )
+
+
 add_subdirectory( src )
-add_subdirectory( tpl/camp )
-target_compile_options( camp PRIVATE "-Wno-shadow") 
 
-configure_file(tpl/camp/include/camp/config.in.hpp
-  ${PROJECT_BINARY_DIR}/include/camp/config.hpp)
+if( SHIVA_ENABLE_CAMP )
+    add_subdirectory( tpl/camp )
+    target_compile_options( camp PRIVATE "-Wno-shadow") 
+
+    configure_file(tpl/camp/include/camp/config.in.hpp
+      ${PROJECT_BINARY_DIR}/include/camp/config.hpp)
+endif()
 
 
 if( SHIVA_ENABLE_DOCS )

cmake/CMakeBasics.cmake

@@ -27,5 +27,6 @@ blt_append_custom_compiler_flag( FLAGS_VAR CMAKE_CXX_FLAGS_DEBUG
                                  CLANG "-fstandalone-debug"
                                 )
 
+set( SHIVA_ENABLE_CAMP OFF CACHE BOOL "")
 
 set( CAMP_ENABLE_TESTS OFF CACHE BOOL "")

cmake/Config.cmake

@@ -1,6 +1,7 @@
 #
 set( PREPROCESSOR_DEFINES CUDA
                           HIP
+                          CAMP
                           BOUNDS_CHECK
                         )
 

docs/doxygen/ShivaConfig.hpp

@@ -14,4 +14,6 @@
 
 /* #undef SHIVA_USE_CALIPER */
 
+/* #undef SHIVA_USE_CAMP */
+
 #define SHIVA_USE_BOUNDS_CHECK

scripts/aggregateOrSplit.py

@@ -20,11 +20,13 @@ def create_dependency_graph(self, header, include_paths=None):
         if include_paths is None:
             include_paths = []
 
+        header = os.path.abspath(header)  # Normalize here
+
         if header in self.dependencies:
             return  # Already processed
 
         self.dependencies[header] = set()
-        base_path = os.path.dirname(os.path.abspath(header))  # Base directory of the current header
+        base_path = os.path.dirname(header)  # Base directory of the current header
 
         try:
             with open(header, 'r') as file:
@@ -34,10 +36,10 @@ def create_dependency_graph(self, header, include_paths=None):
                         included_file = include_match.group(1)
 
                         if included_file != self.config_file:
-                            resolved_path = self.resolve_path(
-                                included_file, base_path, include_paths)
+                            resolved_path = self.resolve_path( included_file, base_path, include_paths)
 
                             if resolved_path:
+                                resolved_path = os.path.abspath(resolved_path)
                                 self.dependencies[header].add(resolved_path)
 
                                 if os.path.exists(resolved_path):
@@ -82,16 +84,21 @@ def resolve_path(self, included_file, base_path, include_paths):
 
         return None  # Return None if no resolution was possible
 
+
     def generate_header_list(self):
         remaining_dependencies = self.dependencies.copy()
         size_of_remaining_dependencies = len(remaining_dependencies)
+        unique_files = set()  # Track unique files by absolute path
 
         while size_of_remaining_dependencies > 0:
             local_included = []
 
             for key in remaining_dependencies:
                 if len(remaining_dependencies[key]) == 0:
-                    self.included_list.append(key)
+                    abs_key = os.path.abspath(key)
+                    if abs_key not in unique_files:
+                        self.included_list.append(abs_key)
+                        unique_files.add(abs_key)
                     local_included.append(key)
 
             for included_key in local_included:
@@ -111,6 +118,7 @@ def process_header(header_path, output):
             """
             Processes a single header file, commenting out includes and pragmas.
             """
+            header_path = os.path.abspath(header_path)
             if header_path in self.included:
                 return  # Avoid duplicate processing
             self.included.add(header_path)
@@ -133,6 +141,7 @@ def process_header(header_path, output):
 
         with open(output_file, 'w') as output:
             for header in headers:
+                header = os.path.abspath(header)
                 self.create_dependency_graph(header, include_paths)
 
             for header in self.dependencies:

src/CMakeLists.txt

@@ -23,7 +23,8 @@ blt_add_library( NAME             shiva
 
 target_include_directories( shiva
                             INTERFACE
-                            $<BUILD_INTERFACE:${CMAKE_BINARY_DIR}/include>
+                            $<BUILD_INTERFACE:${SHIVA_BINARY_DIR}/include>
+                            $<BUILD_INTERFACE:${SHIVA_SOURCE_DIR}/src>
                             $<INSTALL_INTERFACE:include> )
 
 install( FILES ${shiva_headers} 

src/ShivaConfig.hpp.in

@@ -14,4 +14,6 @@
 
 #cmakedefine SHIVA_USE_CALIPER
 
+#cmakedefine SHIVA_USE_CAMP
+
 #cmakedefine SHIVA_USE_BOUNDS_CHECK

src/common/CMakeLists.txt

@@ -35,14 +35,14 @@ blt_add_library( NAME             common
 
 target_include_directories( common 
                             INTERFACE 
-                            $<BUILD_INTERFACE:${CMAKE_BINARY_DIR}/include>
-                            $<BUILD_INTERFACE:${CMAKE_SOURCE_DIR}/src>
+                            $<BUILD_INTERFACE:${SHIVA_BINARY_DIR}/include>
+                            $<BUILD_INTERFACE:${SHIVA_SOURCE_DIR}/src>
                             $<INSTALL_INTERFACE:include> )
 
 target_include_directories( common 
                             SYSTEM INTERFACE
-                            $<BUILD_INTERFACE:${CMAKE_SOURCE_DIR}/tpl/camp/include>
-                            $<BUILD_INTERFACE:${CMAKE_BINARY_DIR}/tpl/camp/include> )
+                            $<BUILD_INTERFACE:${SHIVA_SOURCE_DIR}/tpl/camp/include>
+                            $<BUILD_INTERFACE:${SHIVA_BINARY_DIR}/tpl/camp/include> )
 
 install( FILES ${common_headers} 
          DESTINATION include/common )

src/common/pmpl.hpp

@@ -100,7 +100,7 @@ void genericKernelWrapper( LAMBDA && func, bool const abortOnError = true )
 #if defined(SHIVA_USE_DEVICE)
   // UNCRUSTIFY-OFF
   genericKernel <<< 1, 1 >>> ( std::forward< LAMBDA >( func ) );
-  //UNCRUSTIFY-ON
+  // UNCRUSTIFY-ON
   deviceError_t err = deviceDeviceSynchronize();
   if ( err != cudaSuccess )
   {
@@ -157,6 +157,7 @@ void genericKernelWrapper( int const N, DATA_TYPE * const hostData, LAMBDA && fu
 #if defined(SHIVA_USE_DEVICE)
   DATA_TYPE * deviceData;
   deviceMalloc( &deviceData, N * sizeof(DATA_TYPE) );
+  deviceMemCpy( deviceData, hostData, N * sizeof(DATA_TYPE), cudaMemcpyHostToDevice );
   // UNCRUSTIFY-OFF
   genericKernel <<< 1, 1 >>> ( std::forward< LAMBDA >( func ), deviceData );
   // UNCRUSTIFY-ON

src/common/types.hpp

@@ -20,7 +20,6 @@
 #include "common/ShivaMacros.hpp"
 
 /// @brief Macro to define whether or not to use camp.
-#define SHIVA_USE_CAMP
 #if defined(SHIVA_USE_CAMP)
 #include <camp/camp.hpp>
 #else

src/functions/quadrature/Quadrature.hpp

@@ -53,13 +53,13 @@ struct QuadratureGaussLegendre : public GaussLegendreSpacing< REAL_TYPE, N >
     }
     else if constexpr ( N == 3 )
     {
-      assert( index >= 0 && index < 3 );
+      //assert( index >= 0 && index < 3 );
       return 0.5555555555555555555555555555555556 +
              0.3333333333333333333333333333333333 * ( index & 1 );
     }
     else if constexpr ( N == 4 )
     {
-      assert( index >= 0 && index < 4 );
+      //assert( index >= 0 && index < 4 );
       return 0.5 + ( -1 + ( ( ( index + 1 ) & 2 ) ) ) * 0.15214515486254614262693605077800059277;
     }
     return std::numeric_limits< REAL_TYPE >::max();
@@ -145,20 +145,20 @@ struct QuadratureGaussLobatto : public GaussLobattoSpacing< REAL_TYPE, N >
     }
     else if constexpr ( N == 3 )
     {
-      assert( index >= 0 && index < 3 );
+      //assert( index >= 0 && index < 3 );
       return 0.3333333333333333333333333333333333 + ( index & 1 );
     }
     else if constexpr ( N == 4 )
     {
-      assert( index >= 0 && index < 4 );
+      //assert( index >= 0 && index < 4 );
       return 0.1666666666666666666666666666666667 + ( ((index + 1) & 2) >> 1 ) * 0.6666666666666666666666666666666667;
     }
     else if constexpr ( N == 5 )
     {
-      assert( index >= 0 && index < 5 );
+      //assert( index >= 0 && index < 5 );
       return 0.1 + (index & 1) * 0.4444444444444444444444444444444444 + !( index - 2 ) * 0.6111111111111111111111111111111111;
     }
-    return std::numeric_limits< REAL_TYPE >::max();
+    return 0;//std::numeric_limits< REAL_TYPE >::max();
   }
 
   /**

src/geometry/mapping/LinearTransform.hpp

@@ -208,19 +208,19 @@ SHIVA_STATIC_CONSTEXPR_HOSTDEVICE_FORCEINLINE void
 jacobian( LinearTransform< REAL_TYPE, INTERPOLATED_SHAPE > const & transform,
           typename LinearTransform< REAL_TYPE, INTERPOLATED_SHAPE >::JacobianType & J )
 {
-  using Transform = std::remove_reference_t< decltype(transform) >;
+  using Transform = LinearTransform< REAL_TYPE, INTERPOLATED_SHAPE >;
   using InterpolatedShape = typename Transform::InterpolatedShape;
   constexpr int DIMS = Transform::numDims;
 
   auto const & nodeCoords = transform.getData();
-  constexpr double qcoords[3] = { ( QUADRATURE::template coordinate< QA >() )... };
+  constexpr REAL_TYPE qcoords[3] = { ( QUADRATURE::template coordinate< QA >() )... };
 
   InterpolatedShape::supportLoop( [&] ( auto const ... ic_spIndices ) constexpr
         {
           constexpr CArrayNd< REAL_TYPE, DIMS > dNadXi = InterpolatedShape::template gradient< decltype(ic_spIndices)::value ... >( qcoords );
 
           // dimensional loop from domain to codomain
-    #if 0
+    #if 1
           forNestedSequence< DIMS, DIMS >( [&] ( auto const ici, auto const icj ) constexpr
           {
             constexpr int i = decltype(ici)::value;