Merge branch 'develop' into fix-issue-125

CMake/macros.cmake

@@ -408,7 +408,7 @@ else(QMC_NO_SLOW_CUSTOM_TESTING_COMMANDS)
               #MESSAGE("check command = ${CHECK_CMD}")
               add_test(
                 NAME ${TEST_NAME}
-                COMMAND ${CHECK_CMD}
+                COMMAND ${Python3_EXECUTABLE} ${CHECK_CMD}
                 WORKING_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}/${FULL_NAME}")
               set_property(TEST ${TEST_NAME} APPEND PROPERTY DEPENDS ${FULL_NAME})
               set_property(TEST ${TEST_NAME} APPEND PROPERTY LABELS "QMCPACK-checking-results")
@@ -536,7 +536,7 @@ else(QMC_NO_SLOW_CUSTOM_TESTING_COMMANDS)
             ${SCALAR_ERROR})
         add_test(
           NAME ${TEST_NAME}
-          COMMAND ${CHECK_CMD}
+          COMMAND ${Python3_EXECUTABLE} ${CHECK_CMD}
           WORKING_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}/${FULL_NAME}")
         set_property(TEST ${TEST_NAME} APPEND PROPERTY DEPENDS ${FULL_NAME})
         set_property(TEST ${TEST_NAME} APPEND PROPERTY LABELS "QMCPACK-checking-results")
@@ -595,7 +595,7 @@ else(QMC_NO_SLOW_CUSTOM_TESTING_COMMANDS)
 
     add_test(
       NAME "${test_name}"
-      COMMAND ${check_cmd} ${ARGN}
+      COMMAND ${Python3_EXECUTABLE} ${check_cmd} ${ARGN}
       WORKING_DIRECTORY "${work_dir}")
 
     # make test depend on the run

CMake/test_labels.cmake

@@ -6,7 +6,7 @@ function(ADD_TEST_LABELS TEST_NAME TEST_LABELS)
   else()
     set(SUCCESS FALSE)
     execute_process(
-      COMMAND ${qmcpack_SOURCE_DIR}/tests/scripts/test_labels.py ${TEST_NAME} ${QMC_CUDA} ${QMC_COMPLEX}
+      COMMAND ${Python3_EXECUTABLE} ${qmcpack_SOURCE_DIR}/tests/scripts/test_labels.py ${TEST_NAME} ${QMC_CUDA} ${QMC_COMPLEX}
               ${QMC_MIXED_PRECISION}
       OUTPUT_VARIABLE TEST_LABELS_TEMP
       RESULT_VARIABLE SUCCESS)

CMake/unit_test.cmake

@@ -36,3 +36,21 @@ function(ADD_UNIT_TEST TESTNAME PROCS THREADS TEST_BINARY)
     APPEND
     PROPERTY LABELS "unit")
 endfunction()
+
+# Add a test to see if the target output exists in the desired location in the build directory.
+function(add_test_target_in_output_location TARGET_NAME_TO_TEST EXE_DIR_RELATIVE_TO_BUILD)
+
+  # obtain BASE_NAME
+  get_target_property(BASE_NAME ${TARGET_NAME_TO_TEST} OUTPUT_NAME)
+  if(NOT BASE_NAME)
+    set(BASE_NAME ${TARGET_NAME_TO_TEST})
+  endif()
+
+  set(TESTNAME build_output_${TARGET_NAME_TO_TEST}_exists)
+  add_test(NAME ${TESTNAME} COMMAND ls ${qmcpack_BINARY_DIR}/bin/${BASE_NAME})
+
+  set_property(
+    TEST ${TESTNAME}
+    APPEND
+    PROPERTY LABELS "unit;deterministic")
+endfunction()

docs/hamiltonianobservable.rst

@@ -327,7 +327,7 @@ attributes:
   +-----------------------------+--------------+-----------------------+------------------------+--------------------------------------------------+
   | ``format``:math:`^r`        | text         | xml/table             | table                  | Select file format                               |
   +-----------------------------+--------------+-----------------------+------------------------+--------------------------------------------------+
-  | ``algorithm``:math:`^o`     | text         | batched/non-batched   | depends                | Choose NLPP algorithm                            |
+  | ``algorithm``:math:`^o`     | text         | batched/non-batched   | batched                | Choose NLPP algorithm                            |
   +-----------------------------+--------------+-----------------------+------------------------+--------------------------------------------------+
   | ``DLA``:math:`^o`           | text         | yes/no                | no                     | Use determinant localization approximation       |
   +-----------------------------+--------------+-----------------------+------------------------+--------------------------------------------------+
@@ -359,9 +359,7 @@ Additional information:
    These elements specify individual file names and formats (both the
    FSAtom XML and CASINO tabular data formats are supported).
 
--  **algorithm** The default value is ``batched`` when OpenMP offload
-   is enabled and``non-batched`` otherwise.
-   The ``non-batched`` algorithm evaluates the ratios of
+-  **algorithm** The ``non-batched`` algorithm evaluates the ratios of
    wavefunction components together for each quadrature point and then
    one point after another. The ``batched`` algorithm evaluates the ratios
    of quadrature points together for each wavefunction component and

src/QMCApp/CMakeLists.txt

@@ -56,4 +56,5 @@ if(USE_OBJECT_TARGET)
     platform_omptarget_LA)
 endif()
 
+add_test_target_in_output_location(qmcpack bin)
 install(TARGETS qmcpack RUNTIME DESTINATION bin)

src/QMCHamiltonians/ECPotentialBuilder.cpp

@@ -71,20 +71,13 @@ bool ECPotentialBuilder::put(xmlNodePtr cur)
 
   OhmmsAttributeSet pAttrib;
   pAttrib.add(ecpFormat, "format", {"table", "xml"});
-  pAttrib.add(NLPP_algo, "algorithm", {"", "batched", "non-batched"});
+  pAttrib.add(NLPP_algo, "algorithm", {"batched", "non-batched"});
   pAttrib.add(use_DLA, "DLA", {"no", "yes"});
   pAttrib.add(pbc, "pbc", {"yes", "no"});
   pAttrib.add(forces, "forces", {"no", "yes"});
   pAttrib.add(physicalSO, "physicalSO", {"yes", "no"});
   pAttrib.put(cur);
 
-  if (NLPP_algo.empty())
-#ifdef ENABLE_OFFLOAD
-    NLPP_algo = "batched";
-#else
-    NLPP_algo = "non-batched";
-#endif
-
   bool doForces = (forces == "yes") || (forces == "true");
   if (use_DLA == "yes")
     app_log() << "    Using determinant localization approximation (DLA)" << std::endl;

src/QMCHamiltonians/tests/test_ion_derivs.cpp

@@ -20,6 +20,7 @@
 #include "QMCHamiltonians/tests/MinimalHamiltonianPool.h"
 #include "ParticleIO/XMLParticleIO.h"
 #include "Utilities/RandomGenerator.h"
+#include "QMCWaveFunctions/TWFFastDerivWrapper.h"
 
 namespace qmcplusplus
 {
@@ -766,9 +767,9 @@ TEST_CASE("Eloc_Derivatives:proto_sd_noj", "[hamiltonian]")
   REQUIRE(psi != nullptr);
   //end incantation
 
-//  TWFPrototype twf;
+  TWFFastDerivWrapper twf;
 
-//  psi->initialize_TWF_Prototype(elec, twf);
+  psi->initializeTWFFastDerivWrapper(elec, twf);
   SPOSet::ValueVector_t values;
   SPOSet::GradVector_t dpsi;
   SPOSet::ValueVector_t d2psi;
@@ -833,7 +834,7 @@ TEST_CASE("Eloc_Derivatives:proto_sd_noj", "[hamiltonian]")
   X.push_back(upmat);
   X.push_back(dnmat);
 
-//  twf.get_M(elec, matlist);
+  twf.getM(elec, matlist);
 
   OperatorBase* kinop = ham.getHamiltonian(KINETIC);
 
@@ -851,17 +852,17 @@ TEST_CASE("Eloc_Derivatives:proto_sd_noj", "[hamiltonian]")
   minv.push_back(dnmat);
 
 
-//  twf.get_M(elec, matlist);
+//  twf.getM(elec, matlist);
   std::vector<std::vector<ValueMatrix_t>> dB_gs;
   std::vector<std::vector<ValueMatrix_t>> dM_gs;
   std::vector<ValueMatrix_t> tmp_gs;
-//  twf.get_gs_matrix(B, B_gs);
-//  twf.get_gs_matrix(matlist, M_gs);
-//  twf.invert_M(M_gs, minv);
-//  twf.build_X(minv, B_gs, X);
+  twf.getGSMatrices(B, B_gs);
+  twf.getGSMatrices(matlist, M_gs);
+  twf.invertMatrices(M_gs, minv);
+  twf.buildX(minv, B_gs, X);
   for (int id = 0; id < matlist.size(); id++)
   {
-//    int ptclnum = twf.num_particles(id);
+//    int ptclnum = twf.numParticles(id);
     int ptclnum = (id==0 ? Nup : Ndn); //hard coded until twf interface comes online.  
     ValueMatrix_t gs_m;
     gs_m.resize(ptclnum, ptclnum);
@@ -887,26 +888,26 @@ TEST_CASE("Eloc_Derivatives:proto_sd_noj", "[hamiltonian]")
   {
     for (int idim = 0; idim < OHMMS_DIM; idim++)
     {
-//      twf.wipe_matrix(dB[idim]);
-//      twf.wipe_matrix(dM[idim]);
+      twf.wipeMatrices(dB[idim]);
+      twf.wipeMatrices(dM[idim]);
     }
 
-//    twf.get_igrad_M(elec, ions, ionid, dM);
+    twf.getIonGradM(elec, ions, ionid, dM);
 //    kinop->evaluateOneBodyOpMatrixForceDeriv(elec, ions, twf, ionid, dB);
 
     for (int idim = 0; idim < OHMMS_DIM; idim++)
     {
-//      twf.get_gs_matrix(dB[idim], dB_gs[idim]);
-//      twf.get_gs_matrix(dM[idim], dM_gs[idim]);
-//      fkin_complex[ionid][idim] = twf.compute_gs_derivative(minv, X, dM_gs[idim], dB_gs[idim]);
+      twf.getGSMatrices(dB[idim], dB_gs[idim]);
+      twf.getGSMatrices(dM[idim], dM_gs[idim]);
+      fkin_complex[ionid][idim] = twf.computeGSDerivative(minv, X, dM_gs[idim], dB_gs[idim]);
     }
     convertToReal(fkin_complex[ionid], fkin[ionid]);
   }
 
 
   ValueType keval = 0.0;
   RealType keobs  = 0.0;
-//  keval           = twf.trAB(minv, B_gs);
+  keval           = twf.trAB(minv, B_gs);
   convertToReal(keval, keobs);
 //  CHECK(keobs == Approx(9.1821937928e+00));
 #if defined(MIXED_PRECISION)
@@ -933,12 +934,12 @@ TEST_CASE("Eloc_Derivatives:proto_sd_noj", "[hamiltonian]")
   app_log() << "  Evaluated.  Calling evaluteOneBodyOpMatrix\n";
 
 
-//  twf.wipe_matrix(B);
-//  twf.wipe_matrix(B_gs);
-//  twf.wipe_matrix(X);
+//  twf.wipeMatrices(B);
+//  twf.wipeMatrices(B_gs);
+//  twf.wipeMatrices(X);
 //  nlppop->evaluateOneBodyOpMatrix(elec, twf, B);
-//  twf.get_gs_matrix(B, B_gs);
-//  twf.build_X(minv, B_gs, X);
+//  twf.getGSMatrices(B, B_gs);
+//  twf.buildX(minv, B_gs, X);
 
   ValueType nlpp    = 0.0;
   RealType nlpp_obs = 0.0;
@@ -955,18 +956,18 @@ TEST_CASE("Eloc_Derivatives:proto_sd_noj", "[hamiltonian]")
   {
     for (int idim = 0; idim < OHMMS_DIM; idim++)
     {
-//      twf.wipe_matrix(dB[idim]);
-//      twf.wipe_matrix(dM[idim]);
+//      twf.wipeMatrices(dB[idim]);
+//      twf.wipeMatrices(dM[idim]);
     }
 
-//    twf.get_igrad_M(elec, ions, ionid, dM);
+//    twf.getIonGradM(elec, ions, ionid, dM);
 //    nlppop->evaluateOneBodyOpMatrixForceDeriv(elec, ions, twf, ionid, dB);
 
     for (int idim = 0; idim < OHMMS_DIM; idim++)
     {
-//      twf.get_gs_matrix(dB[idim], dB_gs[idim]);
-//      twf.get_gs_matrix(dM[idim], dM_gs[idim]);
-//      fnlpp_complex[ionid][idim] = twf.compute_gs_derivative(minv, X, dM_gs[idim], dB_gs[idim]);
+//      twf.getGSMatrices(dB[idim], dB_gs[idim]);
+//      twf.getGSMatrices(dM[idim], dM_gs[idim]);
+//      fnlpp_complex[ionid][idim] = twf.computeGSDerivative(minv, X, dM_gs[idim], dB_gs[idim]);
     }
     convertToReal(fnlpp_complex[ionid], fnlpp[ionid]);
   }

src/QMCTools/CMakeLists.txt

@@ -53,8 +53,10 @@ if(USE_OBJECT_TARGET)
   target_link_libraries(qmcfinitesize qmcparticle qmcutil)
 endif()
 
-install(TARGETS convert4qmc qmc-extract-eshdf-kvectors qmc-get-supercell qmc-check-affinity convertpw4qmc qmcfinitesize
-        RUNTIME DESTINATION bin)
+foreach(EXE_TARGET convert4qmc qmc-extract-eshdf-kvectors qmc-get-supercell qmc-check-affinity convertpw4qmc qmcfinitesize)
+  add_test_target_in_output_location(${EXE_TARGET} bin)
+  install(TARGETS ${EXE_TARGET} RUNTIME DESTINATION bin)
+endforeach()
 
 add_subdirectory(ppconvert)
 

src/QMCTools/ppconvert/src/CMakeLists.txt

@@ -9,3 +9,6 @@ target_sources(
          "${CMAKE_CURRENT_SOURCE_DIR}/ParserClass.cc")
 
 target_link_libraries(ppconvert common platform_LA)
+
+add_test_target_in_output_location(ppconvert bin)
+install(TARGETS ppconvert RUNTIME DESTINATION bin)

src/QMCWaveFunctions/CMakeLists.txt

@@ -169,6 +169,7 @@ set(FERMION_SRCS
     SPOSetBuilderFactory.cpp
     TrialWaveFunction.cpp
     TWFdispatcher.cpp
+    TWFFastDerivWrapper.cpp
     WaveFunctionFactory.cpp)
 
 if(ENABLE_CUDA)

src/QMCWaveFunctions/Fermion/DiracDeterminant.cpp

@@ -21,6 +21,7 @@
 #include "CPU/SIMD/simd.hpp"
 #include "Numerics/DeterminantOperators.h"
 #include "Numerics/MatrixOperators.h"
+#include "QMCWaveFunctions/TWFFastDerivWrapper.h"
 
 namespace qmcplusplus
 {
@@ -344,6 +345,12 @@ void DiracDeterminant<DU_TYPE>::copyFromBuffer(ParticleSet& P, WFBufferType& buf
   updateEng.initializeInv(psiM);
 }
 
+template<typename DU_TYPE>
+void DiracDeterminant<DU_TYPE>::registerTWFFastDerivWrapper(const ParticleSet& P, TWFFastDerivWrapper& twf) const
+{
+  twf.addGroup(P, P.getGroupID(FirstIndex), Phi.get());
+}
+
 /** return the ratio only for the  iat-th partcle move
  * @param P current configuration
  * @param iat the particle thas is being moved

src/QMCWaveFunctions/Fermion/DiracDeterminant.h

@@ -76,6 +76,10 @@ class DiracDeterminant : public DiracDeterminantBase
 
   void copyFromBuffer(ParticleSet& P, WFBufferType& buf) override;
 
+  /** Finds the SPOSet associated with this determinant, and registers it with WFN wrapper
+   */ 
+  void registerTWFFastDerivWrapper(const ParticleSet& P, TWFFastDerivWrapper& twf) const final;
+
   /** return the ratio only for the  iat-th partcle move
    * @param P current configuration
    * @param iat the particle thas is being moved

src/QMCWaveFunctions/Fermion/DiracDeterminantBase.h

@@ -85,6 +85,10 @@ class DiracDeterminantBase : public WaveFunctionComponent
 
   inline void reportStatus(std::ostream& os) final {}
 
+  virtual void registerTWFFastDerivWrapper(const ParticleSet& P, TWFFastDerivWrapper& twf) const override
+  {
+    APP_ABORT("DiracDeterminantBase::registerTWFFastDerivWrapper must be overridden\n");
+  }
   // expose CPU interfaces
   using WaveFunctionComponent::evaluateDerivatives;
   using WaveFunctionComponent::evaluateGL;

src/QMCWaveFunctions/Fermion/DiracDeterminantBatched.cpp

@@ -16,6 +16,7 @@
 #include "CPU/BLAS.hpp"
 #include "OhmmsPETE/OhmmsMatrix.h"
 #include "Numerics/MatrixOperators.h"
+#include "QMCWaveFunctions/TWFFastDerivWrapper.h"
 #include "CPU/SIMD/simd.hpp"
 #include <cassert>
 
@@ -956,6 +957,12 @@ void DiracDeterminantBatched<DET_ENGINE>::evaluateDerivatives(ParticleSet& P,
   Phi->evaluateDerivatives(P, active, dlogpsi, dhpsioverpsi, FirstIndex, LastIndex);
 }
 
+template<typename DET_ENGINE>
+void DiracDeterminantBatched<DET_ENGINE>::registerTWFFastDerivWrapper(const ParticleSet& P, TWFFastDerivWrapper& twf) const
+{  
+  twf.addGroup(P, P.getGroupID(FirstIndex), Phi.get());
+}
+
 template<typename DET_ENGINE>
 std::unique_ptr<DiracDeterminantBase> DiracDeterminantBatched<DET_ENGINE>::makeCopy(std::shared_ptr<SPOSet>&& spo) const
 {

src/QMCWaveFunctions/Fermion/DiracDeterminantBatched.h

@@ -29,6 +29,9 @@
 namespace qmcplusplus
 {
 
+//forward declaration
+class TWFFastDerivWrapper;
+
 template<typename DET_ENGINE = MatrixUpdateOMPTarget<QMCTraits::ValueType, QMCTraits::QTFull::ValueType>>
 class DiracDeterminantBatched : public DiracDeterminantBase
 {
@@ -209,6 +212,7 @@ class DiracDeterminantBatched : public DiracDeterminantBase
   void releaseResource(ResourceCollection& collection,
                        const RefVectorWithLeader<WaveFunctionComponent>& wfc_list) const override;
 
+  void registerTWFFastDerivWrapper(const ParticleSet& P, TWFFastDerivWrapper& twf) const override;
   /** cloning function
    * @param tqp target particleset
    * @param spo spo set

src/QMCWaveFunctions/Fermion/SlaterDet.cpp

@@ -253,4 +253,12 @@ std::unique_ptr<WaveFunctionComponent> SlaterDet::makeClone(ParticleSet& tqp) co
   return myclone;
 }
 
+void SlaterDet::registerTWFFastDerivWrapper(const ParticleSet& P, TWFFastDerivWrapper& twf) const
+{
+  for (int i = 0; i < Dets.size(); ++i)
+  {
+    Dets[i]->registerTWFFastDerivWrapper(P, twf);
+  }
+}
+
 } // namespace qmcplusplus

src/QMCWaveFunctions/Fermion/SlaterDet.h

@@ -18,6 +18,7 @@
 #ifndef QMCPLUSPLUS_SLATERDETERMINANT_WITHBASE_H
 #define QMCPLUSPLUS_SLATERDETERMINANT_WITHBASE_H
 #include "QMCWaveFunctions/Fermion/DiracDeterminantBase.h"
+
 #include <map>
 
 namespace qmcplusplus
@@ -28,6 +29,7 @@ namespace qmcplusplus
 //     then change SlaterDet to SlaterDet<false>
 //     and SlaterDeterminantWithBackflow to SlaterDet<true>
 //     and remove all virtuals and inline them
+class TWFFastDerivWrapper;
 
 class SlaterDet : public WaveFunctionComponent
 {
@@ -53,6 +55,8 @@ class SlaterDet : public WaveFunctionComponent
 
   void reportStatus(std::ostream& os) override;
 
+  void registerTWFFastDerivWrapper(const ParticleSet& P, TWFFastDerivWrapper& twf) const override;
+
   LogValueType evaluateLog(const ParticleSet& P,
                            ParticleSet::ParticleGradient_t& G,
                            ParticleSet::ParticleLaplacian_t& L) override;