Pass igraph (per-event MLM graph) through Fortran bridge to SMATRIX_MULTI

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/Bridge.h

@@ -125,7 +125,7 @@ namespace mg5amcCpu
    * @param selcol the pointer to the output selected colors
    * @param goodHelOnly quit after computing good helicities?
    */
-    void gpu_sequence( const FORTRANFPTYPE* momenta, const FORTRANFPTYPE* gs, const FORTRANFPTYPE* rndhel, const FORTRANFPTYPE* rndcol, const unsigned int* channelIds, FORTRANFPTYPE* mes, int* selhel, int* selcol, const bool goodHelOnly = false );
+    void gpu_sequence( const FORTRANFPTYPE* momenta, const FORTRANFPTYPE* gs, const FORTRANFPTYPE* rndhel, const FORTRANFPTYPE* rndcol, const unsigned int* channelIds, const int* igraph, FORTRANFPTYPE* mes, int* selhel, int* selcol, const bool goodHelOnly = false );
 #else
     /**
    * Sequence to be executed for the vectorized CPU matrix element calculation
@@ -143,7 +143,7 @@ namespace mg5amcCpu
    * @param selcol the pointer to the output selected colors
    * @param goodHelOnly quit after computing good helicities?
    */
-    void cpu_sequence( const FORTRANFPTYPE* momenta, const FORTRANFPTYPE* gs, const FORTRANFPTYPE* rndhel, const FORTRANFPTYPE* rndcol, const unsigned int* channelIds, FORTRANFPTYPE* mes, int* selhel, int* selcol, const bool goodHelOnly = false );
+    void cpu_sequence( const FORTRANFPTYPE* momenta, const FORTRANFPTYPE* gs, const FORTRANFPTYPE* rndhel, const FORTRANFPTYPE* rndcol, const unsigned int* channelIds, const int* igraph, FORTRANFPTYPE* mes, int* selhel, int* selcol, const bool goodHelOnly = false );
 #endif
 
     // Return the number of good helicities (-1 initially when they have not yet
@@ -343,6 +343,7 @@ paramCard; #endif
                                             const FORTRANFPTYPE* rndhel,
                                             const FORTRANFPTYPE* rndcol,
                                             const unsigned int* channelIds,
+                                            const int* igraph,
                                             FORTRANFPTYPE* mes,
                                             int* selhel,
                                             int* selcol,
@@ -394,6 +395,7 @@ paramCard; #endif
           "Bridge gpu_sequence: computeGoodHelicities returned nGoodHel<0" );
     }
     if( goodHelOnly ) return;
+    m_pmek->setigraph( igraph );
     m_pmek->computeMatrixElements( useChannelIds );
     copyHostFromDevice( m_hstMEs, m_devMEs );
 #ifdef MGONGPUCPP_VERBOSE
@@ -423,6 +425,7 @@ paramCard; #endif
                                             const FORTRANFPTYPE* rndhel,
                                             const FORTRANFPTYPE* rndcol,
                                             const unsigned int* channelIds,
+                                            const int* igraph,
                                             FORTRANFPTYPE* mes,
                                             int* selhel,
                                             int* selcol,
@@ -454,6 +457,7 @@ paramCard; #endif
           "Bridge cpu_sequence: computeGoodHelicities returned nGoodHel<0" );
     }
     if( goodHelOnly ) return;
+    m_pmek->setigraph( igraph );
     m_pmek->computeMatrixElements( useChannelIds );
 #ifdef MGONGPUCPP_VERBOSE
     flagAbnormalMEs( m_hstMEs.data(), m_nevt );

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/BridgeKernels.cc

@@ -80,7 +80,7 @@ namespace mg5amcCpu
   {
     constexpr bool goodHelOnly = true;
     constexpr unsigned int* pChannelIds = nullptr; // disable multi-channel for helicity filtering
-    m_bridge.cpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
+    m_bridge.cpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, nullptr, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
     return m_bridge.nGoodHel();
   }
 
@@ -90,7 +90,7 @@ namespace mg5amcCpu
   {
     constexpr bool goodHelOnly = false;
     const unsigned int* pChannelIds = ( useChannelIds ? m_channelIds.data() : nullptr );
-    m_bridge.cpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
+    m_bridge.cpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_igraph, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
   }
 
   //--------------------------------------------------------------------------
@@ -139,7 +139,7 @@ namespace mg5amcGpu
   {
     constexpr bool goodHelOnly = true;
     constexpr unsigned int* pChannelIds = nullptr; // disable multi-channel for helicity filtering
-    m_bridge.gpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
+    m_bridge.gpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, nullptr, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
     return m_bridge.nGoodHel();
   }
 
@@ -149,7 +149,7 @@ namespace mg5amcGpu
   {
     constexpr bool goodHelOnly = false;
     const unsigned int* pChannelIds = ( useChannelIds ? m_channelIds.data() : nullptr );
-    m_bridge.gpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
+    m_bridge.gpu_sequence( m_fortranMomenta.data(), m_gs.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_igraph, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), goodHelOnly );
   }
 
   //--------------------------------------------------------------------------

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/MatrixElementKernels.cc

@@ -220,7 +220,7 @@ namespace mg5amcCpu
     computeDependentCouplings( m_gs.data(), m_couplings.data(), m_gs.size() );
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
     const unsigned int* pChannelIds = ( useChannelIds ? m_channelIds.data() : nullptr );
-    sigmaKin( m_momenta.data(), m_couplings.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), m_numerators.data(), m_denominators.data(), nevt() );
+    sigmaKin( m_momenta.data(), m_couplings.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_igraph, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), m_numerators.data(), m_denominators.data(), nevt() );
 #else
     assert( useChannelIds == false );
     sigmaKin( m_momenta.data(), m_couplings.data(), m_rndhel.data(), m_matrixElements.data(), m_selhel.data(), nevt() );
@@ -504,7 +504,7 @@ namespace mg5amcGpu
 #endif
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
     const unsigned int* pChannelIds = ( useChannelIds ? m_channelIds.data() : nullptr );
-    sigmaKin( m_momenta.data(), m_couplings.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), m_colJamp2s.data(), m_pHelNumerators->data(), m_pHelDenominators->data(), m_pHelMEs->data(), m_pHelJamps->data(), ghelAllBlasTmp, pBlasHandle, m_helStreams, m_gpublocks, m_gputhreads );
+    sigmaKin( m_momenta.data(), m_couplings.data(), m_rndhel.data(), m_rndcol.data(), pChannelIds, m_igraph, m_matrixElements.data(), m_selhel.data(), m_selcol.data(), m_colJamp2s.data(), m_pHelNumerators->data(), m_pHelDenominators->data(), m_pHelMEs->data(), m_pHelJamps->data(), ghelAllBlasTmp, pBlasHandle, m_helStreams, m_gpublocks, m_gputhreads );
 #else
     assert( useChannelIds == false );
     sigmaKin( m_momenta.data(), m_couplings.data(), m_rndhel.data(), m_matrixElements.data(), m_selhel.data(), m_pHelMEs->data(), m_pHelJamps->data(), ghelAllBlasTmp, pBlasHandle, m_helStreams, m_gpublocks, m_gputhreads );

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/MatrixElementKernels.h

@@ -46,6 +46,9 @@ namespace mg5amcCpu
     // Compute good helicities (returns nGoodHel, the number of good helicity combinations out of ncomb)
     virtual int computeGoodHelicities() = 0;
 
+    // Set the per-event MLM graph array (nullptr = no MLM matching; must be called before computeMatrixElements if needed)
+    void setigraph( const int* igraph ) { m_igraph = igraph; }
+
     // Compute matrix elements
     virtual void computeMatrixElements( const bool useChannelIds ) = 0;
 
@@ -84,6 +87,9 @@ namespace mg5amcCpu
     // The buffer for the channel ids for single-diagram enhancement
     const BufferChannelIds& m_channelIds;
 
+    // The per-event MLM graph array (nullptr = no MLM; set via setigraph before computeMatrixElements)
+    const int* m_igraph = nullptr;
+
     // The buffer for the output matrix elements
     BufferMatrixElements& m_matrixElements;
 

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/cudacpp.mk

@@ -89,7 +89,7 @@ endif
 # Create file with the resolved backend in case user chooses 'cppauto'
 BACKEND_LOG ?= .resolved-backend
 ifneq ($(BACKEND_ORIG),$(BACKEND))
-  $(file >$(BACKEND_LOG),$(BACKEND))
+  $(shell echo '$(BACKEND)' >> $(BACKEND_LOG))
 endif
 #-------------------------------------------------------------------------------
 

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/fbridge.cc

@@ -91,6 +91,7 @@ extern "C"
                          const FORTRANFPTYPE* rndhel,
                          const FORTRANFPTYPE* rndcol,
                          const unsigned int* channelIds,
+                         const int* igraph,
                          FORTRANFPTYPE* mes,
                          int* selhel,
                          int* selcol,
@@ -102,11 +103,11 @@ extern "C"
 #ifdef MGONGPUCPP_GPUIMPL
     // Use the device/GPU implementation in the CUDA library
     // (there is also a host implementation in this library)
-    pbridge->gpu_sequence( momenta, gs, rndhel, rndcol, channelIds, mes, selhel, selcol, *pgoodHelOnly );
+    pbridge->gpu_sequence( momenta, gs, rndhel, rndcol, channelIds, igraph, mes, selhel, selcol, *pgoodHelOnly );
 #else
     // Use the host/CPU implementation in the C++ library
     // (there is no device implementation in this library)
-    pbridge->cpu_sequence( momenta, gs, rndhel, rndcol, channelIds, mes, selhel, selcol, *pgoodHelOnly );
+    pbridge->cpu_sequence( momenta, gs, rndhel, rndcol, channelIds, igraph, mes, selhel, selcol, *pgoodHelOnly );
 #endif
   }
 
@@ -129,13 +130,14 @@ extern "C"
                                         const FORTRANFPTYPE* gs,
                                         const FORTRANFPTYPE* rndhel,
                                         const FORTRANFPTYPE* rndcol,
+                                        const int* igraph,
                                         FORTRANFPTYPE* mes,
                                         int* selhel,
                                         int* selcol,
                                         const bool* pgoodHelOnly )
   {
     //printf("fbridgesequence_nomultichannel_ goodHelOnly=%d\n", ( *pgoodHelOnly ? 1 : 0 ) );
-    fbridgesequence_( ppbridge, momenta, gs, rndhel, rndcol, nullptr, mes, selhel, selcol, pgoodHelOnly );
+    fbridgesequence_( ppbridge, momenta, gs, rndhel, rndcol, nullptr, igraph, mes, selhel, selcol, pgoodHelOnly );
   }
 
   /**

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/fbridge.h

@@ -29,6 +29,7 @@ extern "C"
                          const FORTRANFPTYPE* rndhel,
                          const FORTRANFPTYPE* rndcol,
                          const unsigned int* channelIds,
+                         const int* igraph,
                          FORTRANFPTYPE* mes,
                          int* selhel,
                          int* selcol,
@@ -39,11 +40,12 @@ extern "C"
                                         const FORTRANFPTYPE* gs,
                                         const FORTRANFPTYPE* rndhel,
                                         const FORTRANFPTYPE* rndcol,
+                                        const int* igraph,
                                         FORTRANFPTYPE* mes,
                                         int* selhel,
                                         int* selcol,
                                         const bool* pgoodHelOnly );
 
   void fbridgegetngoodhel_( CppObjectInFortran** ppbridge, unsigned int* pngoodhel, unsigned int* pntothel );
 }
-#endif // _FBRIDGE_H_
+#endif // _FBRIDGE_H_

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/fbridge.inc

@@ -37,20 +37,23 @@ C - GS:      the input Gs (running QCD coupling constant alphas) Fortran array
 C - RNDHEL:  the input random number Fortran array for helicity selection
 C - RNDCOL:  the input random number Fortran array for color selection
 C - CHANID:  the input array of channels (Feynman diagrams) to enhance
+C - IGRAPH: the input per-event MLM graph array (0 = no MLM graph)
 C - MES:     the output matrix element Fortran array
 C - SELHEL:  the output selected helicity Fortran array
 C - SELCOL:  the output selected color Fortran array
 C - HELONLY: input flag, quit after computing good helicities?
 C
       INTERFACE
          SUBROUTINE FBRIDGESEQUENCE(PBRIDGE, MOMENTA, GS,
-     &     RNDHEL, RNDCOL, CHANID, MES, SELHEL, SELCOL, HELONLY)
+     &     RNDHEL, RNDCOL, CHANID, IGRAPH, MES, SELHEL, SELCOL,
+     &     HELONLY)
          INTEGER*8 PBRIDGE
          DOUBLE PRECISION MOMENTA(*)
          DOUBLE PRECISION GS(*)
          DOUBLE PRECISION RNDHEL(*)
          DOUBLE PRECISION RNDCOL(*)
          INTEGER*4 CHANID(*)
+         INTEGER*4 IGRAPH(*)
          DOUBLE PRECISION MES(*)
          INTEGER*4 SELHEL(*)
          INTEGER*4 SELCOL(*)
@@ -65,19 +68,21 @@ C - MOMENTA: the input 4-momenta Fortran array
 C - GS:      the input Gs (running QCD coupling constant alphas) Fortran array
 C - RNDHEL:  the input random number Fortran array for helicity selection
 C - RNDCOL:  the input random number Fortran array for color selection
+C - IGRAPH: the input per-event MLM graph array (0 = no MLM graph)
 C - MES:     the output matrix element Fortran array
 C - SELHEL:  the output selected helicity Fortran array
 C - SELCOL:  the output selected color Fortran array
 C - HELONLY: input flag, quit after computing good helicities?
 C
       INTERFACE
          SUBROUTINE FBRIDGESEQUENCE_NOMULTICHANNEL(PBRIDGE, MOMENTA, GS,
-     &     RNDHEL, RNDCOL, MES, SELHEL, SELCOL, HELONLY)
+     &     RNDHEL, RNDCOL, IGRAPH, MES, SELHEL, SELCOL, HELONLY)
          INTEGER*8 PBRIDGE
          DOUBLE PRECISION MOMENTA(*)
          DOUBLE PRECISION GS(*)
          DOUBLE PRECISION RNDHEL(*)
          DOUBLE PRECISION RNDCOL(*)
+         INTEGER*4 IGRAPH(*)
          DOUBLE PRECISION MES(*)
          INTEGER*4 SELHEL(*)
          INTEGER*4 SELCOL(*)

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/fcheck_sa.f

@@ -20,6 +20,7 @@ PROGRAM FCHECK_SA
       DOUBLE PRECISION RNDHEL(NEVTMAX) ! not yet used
       DOUBLE PRECISION RNDCOL(NEVTMAX) ! not yet used
       DOUBLE PRECISION MES(NEVTMAX)
+      INTEGER*4 IGRAPH(NEVTMAX) ! per-event MLM graph (0 = no MLM)
       INTEGER*4 SELHEL(NEVTMAX) ! not yet used
       INTEGER*4 SELCOL(NEVTMAX) ! not yet used
       DOUBLE PRECISION MES_SUM ! use REAL*16 for quadruple precision
@@ -62,8 +63,9 @@ PROGRAM FCHECK_SA
         DO IEVT = 1, NEVT
           GS(IEVT) = 1.2177157847767195 ! fixed G for aS=0.118 (hardcoded for now in check_sa.cc, fcheck_sa.f, runTest.cc)
         END DO
+        IGRAPH(:) = 0 ! no MLM graph matching in standalone check
         CALL FBRIDGESEQUENCE_NOMULTICHANNEL(BRIDGE, MOMENTA, GS, ! TEMPORARY? disable multi-channel in fcheck.exe and fgcheck.exe #466
-     &    RNDHEL, RNDCOL, MES, SELHEL, SELCOL, .FALSE.) ! do not quit after computing helicities
+     &    RNDHEL, RNDCOL, IGRAPH, MES, SELHEL, SELCOL, .FALSE.) ! do not quit after computing helicities
         DO IEVT = 1, NEVT
 c         DO IEXTERNAL = 1, NEXTERNAL
 c           WRITE(6,*) 'MOMENTA', IEVT, IEXTERNAL,