La3 ni2 o7 bilayer model

applications/dca/main_dca.cpp

@@ -27,6 +27,10 @@
 #include "dca/io/writer.hpp"
 
 int dca_main(int argc, char** argv) {
+#ifdef DCA_HAVE_GPU
+  dca::linalg::util::initializeMagma();
+#endif  // DCA_HAVE_GPU
+
   Concurrency concurrency(argc, argv);
 
   try {
@@ -59,9 +63,6 @@ int dca_main(int argc, char** argv) {
           << std::endl;
     }
 
-#ifdef DCA_HAVE_GPU
-    dca::linalg::util::initializeMagma();
-#endif  // DCA_HAVE_GPU
 
     // Create the parameters object from the input file.
     ParametersType parameters(dca::util::GitVersion::string(), concurrency);

build-aux/frontier_rocm6_load_modules.sh

@@ -9,7 +9,6 @@
 
 module reset
 module load rocm/6.0.0
-module load libfabric/1.15.2.0
 module load ninja
 module load cmake
 

build-aux/local_CPU.cmake

@@ -16,14 +16,14 @@ set(gtest_DIR "${CMAKE_CURRENT_LIST_DIR}/../libs/gmock-1.7.0/gtest/"
 
 # HDF5
 # CMake 3.6 does not properly find HDF5 if HDF5_ROOT is not set.
-set(HDF5_ROOT "/usr/local/Cellar/hdf5/1.12.2_2/" CACHE PATH "Path to HDF5 installation directory.")
+set(HDF5_ROOT "/opt/homebrew/Cellar/hdf5/1.14.4.3/" CACHE PATH "Path to HDF5 installation directory.")
 # set(HDF5_ROOT "/usr/local/anaconda3/pkgs/hdf5-1.10.6-hdbbcd12_0/" CACHE PATH "Path to HDF5 installation directory.")
 
 # FFTW
 # set(FFTW_INCLUDE_DIR "/usr/local/Cellar/fftw/3.3.10/include" CACHE PATH "Path to fftw3.h.")
 # set(FFTW_LIBRARY "/usr/local/Cellar/fftw/3.3.10/lib/libfftw3.a" CACHE FILEPATH "Path to FFTW3 library.")
-set(FFTW_INCLUDE_DIR "/usr/local/anaconda3/pkgs/fftw-3.3.9-h9ed2024_1/include" CACHE PATH "Path to fftw3.h.")
-set(FFTW_LIBRARY "/usr/local/anaconda3/pkgs/fftw-3.3.9-h9ed2024_1/lib/libfftw3.a" CACHE FILEPATH "Path to FFTW3 library.")
+set(FFTW_INCLUDE_DIR "/opt/homebrew/Cellar/fftw/3.3.10_1/include" CACHE PATH "Path to fftw3.h.")
+set(FFTW_LIBRARY "/opt/homebrew/Cellar/fftw/3.3.10_1/lib/libfftw3.a" CACHE FILEPATH "Path to FFTW3 library.")
 
 mark_as_advanced(gtest_DIR SPRNG_DIR HDF5_ROOT)
 

cmake/dca_config.cmake

@@ -121,15 +121,25 @@ endif()
 
 # Lattice type
 set(DCA_LATTICE "square" CACHE STRING "Lattice type, options are: bilayer | square | triangular |
-    Kagome | hund | twoband_Cu | threeband | Rashba_Hubbard | Moire_Hubbard | FeAs | material_NiO | material_FeSn ")
+    Kagome | hund | twoband_Cu | threeband | Rashba_Hubbard | Moire_Hubbard | FeAs | material_NiO | material_FeSn | La3Ni2O7_bilayer")
 set_property(CACHE DCA_LATTICE PROPERTY STRINGS bilayer square triangular Kagome hund twoband_Cu threeband
-             Rashba_Hubbard Moire_Hubbard FeAs material_NiO material_FeSn)
+             Rashba_Hubbard Moire_Hubbard FeAs material_NiO material_FeSn La3Ni2O7_bilayer)
 
 if (DCA_LATTICE STREQUAL "bilayer")
   set(DCA_LATTICE_TYPE dca::phys::models::bilayer_lattice<PointGroup>)
   set(DCA_LATTICE_INCLUDE
     "dca/phys/models/analytic_hamiltonians/bilayer_lattice.hpp")
 
+elseif (DCA_LATTICE STREQUAL "La3Ni2O7_bilayer")
+  set(DCA_LATTICE_TYPE dca::phys::models::La3Ni2O7_bilayer<PointGroup>)
+  set(DCA_LATTICE_INCLUDE
+    "dca/phys/models/analytic_hamiltonians/La3Ni2O7_bilayer.hpp")
+
+elseif (DCA_LATTICE STREQUAL "twoOrbital")
+  set(DCA_LATTICE_TYPE dca::phys::models::twoOrbital<PointGroup>)
+  set(DCA_LATTICE_INCLUDE
+    "dca/phys/models/analytic_hamiltonians/twoOrbital.hpp")
+
 elseif (DCA_LATTICE STREQUAL "square")
   set(DCA_LATTICE_TYPE dca::phys::models::square_lattice<PointGroup>)
   set(DCA_LATTICE_INCLUDE
@@ -182,7 +192,7 @@ elseif (DCA_LATTICE STREQUAL "material_FeSn")
       "dca/phys/models/material_hamiltonians/material_lattice.hpp")
   set(DCA_MODEL_IS_MATERIAL_LATTICE ON CACHE BOOL "is the model a material lattice")
 else()
-  message(FATAL_ERROR "Please set DCA_LATTICE to a valid option: bilayer | square | triangular | Kagome | hund | twoband_Cu | threeband | Rashba_Hubbard | Moire_Hubbard | FeAs | material_NiO | material_FeSn.")
+  message(FATAL_ERROR "Please set DCA_LATTICE to a valid option: bilayer | La3Ni2O7_bilayer | square | triangular | Kagome | hund | twoband_Cu | threeband | Rashba_Hubbard | Moire_Hubbard | FeAs | material_NiO | material_FeSn.")
 endif()
 
 # Model type

include/dca/parallel/mpi_concurrency/mpi_collective_sum.hpp

@@ -626,7 +626,7 @@ inline void MPICollectiveSum::resolveSums() {
     return resolveSumsImplementation<float>();
   else if (current_type_ == MPI_CHAR)
     return resolveSumsImplementation<char>();
-  //else if (current_type_ == MPI_BYTE)
+  // else if (current_type_ == MPI_BYTE)
   else if (current_type_ == MPI_UNSIGNED_CHAR)
     return resolveSumsImplementation<std::uint8_t>();
   else if (current_type_ == MPI_SHORT)

include/dca/phys/dca_data/dca_data.hpp

@@ -348,14 +348,36 @@ void DcaData<Parameters, DT>::read(const std::string& filename) {
     read(reader);
     reader.close_file();
   }
+
   concurrency_.broadcast(parameters_.get_chemical_potential());
+#ifndef NDEBUG
+  if (concurrency_.id() == concurrency_.first()) {
+    std::cout << "broadcasted chemical potential: " << parameters_.get_chemical_potential() << '\n';
+  }
+#endif
   concurrency_.broadcast_object(Sigma);
+#ifndef NDEBUG
+  if (concurrency_.id() == concurrency_.first()) {
+    std::cout << "broadcasted Sigma \n";
+  }
+#endif
 
   if (parameters_.isAccumulatingG4()) {
     concurrency_.broadcast_object(G_k_w);
+#ifndef NDEBUG
+  if (concurrency_.id() == concurrency_.first()) {
+    std::cout << "broadcasted G_k_w \n";
+  }
+#endif
 
-    for (auto& G4_channel : G4_)
+  for (auto& G4_channel : G4_) {
       concurrency_.broadcast_object(G4_channel);
+#ifndef NDEBUG
+  if (concurrency_.id() == concurrency_.first()) {
+    std::cout << "broadcasted G4_channel \n";
+  }
+#endif
+  }
   }
 }
 
@@ -533,6 +555,7 @@ void DcaData<Parameters, DT>::initializeH0_and_H_i() {
     for (int nu2 = 0; nu2 < NuDmn::dmn_size(); ++nu2)
       for (int nu1 = 0; nu1 < NuDmn::dmn_size(); ++nu1) {
         if (std::abs(H_interactions(nu1, nu2, r) - H_interactions(nu2, nu1, minus_r)) > 1e-8) {
+          std::cout << r << " , " << minus_r << " , " << H_interactions(nu1, nu2, r) << " , " << H_interactions(nu2, nu1, minus_r) << "\n";
           throw(std::logic_error("Double counting is not consistent."));
         }
       }
@@ -636,7 +659,17 @@ void DcaData<Parameters, DT>::initializeSigma(const std::string& filename) {
     }
   }
   concurrency_.broadcast(parameters_.get_chemical_potential());
+  #ifndef NDEBUG
+  if (concurrency_.id() == concurrency_.first()) {
+    std::cout << "broadcasted chemical potential: " << parameters_.get_chemical_potential();
+  }
+#endif
   concurrency_.broadcast(Sigma);
+  #ifndef NDEBUG
+  if (concurrency_.id() == concurrency_.first()) {
+    std::cout << "broadcasted Sigma \n";
+  }
+#endif
 }
 
 template <class Parameters, DistType DT>

include/dca/phys/dca_step/cluster_solver/shared_tools/accumulation/tp/g4_helper.cuh

@@ -96,6 +96,10 @@ protected:
 // Global instance to be used in the tp accumulation kernel.
 extern __device__ __constant__ G4Helper g4_helper;
 
+extern __CONSTANT__ int* w_ex_indices_actual;
+extern __CONSTANT__ int* k_ex_indices_actual;
+
+
 inline __device__ int G4Helper::addWex(const int w_idx, const int w_ex_idx) const {
   return w_idx + w_ex_indices_[w_ex_idx];
 }

include/dca/phys/dca_step/cluster_solver/shared_tools/accumulation/tp/tp_accumulator_cpu.hpp

@@ -59,7 +59,7 @@ class TpAccumulator<Parameters, DT, linalg::CPU> : public TpAccumulatorBase<Para
   using Base = TpAccumulatorBase<Parameters, DT>;
 
   using Real = typename Parameters::Real;
-  
+
   using typename Base::NuDmn;
   using typename Base::TpPrecision;
   using typename Base::TpComplex;
@@ -647,7 +647,7 @@ double TpAccumulator<Parameters, DT, linalg::CPU>::updateG4(const int channel_id
                                   w_ex_minus_w(w1, w_ex), w_ex_minus_w(w2, w_ex), G_b_);
                     // getGMultiband(0, k2, k1, w2, w1, G_a_);
                     // getGMultiband(0, q_minus_k(k2, k_ex), q_minus_k(k1, k_ex),
-                    //               w_ex_minus_w(w2, w_ex), w_ex_minus_w(w1, w_ex), G_b_); 
+                    //               w_ex_minus_w(w2, w_ex), w_ex_minus_w(w1, w_ex), G_b_);
                    for (int b4 = 0; b4 < BDmn::dmn_size(); ++b4)
                       for (int b3 = 0; b3 < BDmn::dmn_size(); ++b3)
                         for (int b2 = 0; b2 < BDmn::dmn_size(); ++b2)
@@ -752,7 +752,7 @@ void TpAccumulator<Parameters, DT, linalg::CPU>::updateG4Atomic(
     // Assuming 0 = down 1 = up
     //G_a = G[s_a]
     //G_b = G[d_b]
-      
+
     if (!cross_legs)
       for (int b4 = 0; b4 < n_bands_; ++b4)
         for (int b3 = 0; b3 < n_bands_; ++b3)
@@ -810,7 +810,7 @@ void TpAccumulator<Parameters, DT, linalg::CPU>::updateG4SpinDifference(
     // So G_a_ and G_b_ are accumulating over both spins for each band combination
     getGMultiband(0, k1_a, k2_a, w1_a, w2_a, G_a_);
     getGMultiband(1, k1_a, k2_a, w1_a, w2_a, G_a_, sign);
-   
+
     getGMultiband(0, k1_b, k2_b, w1_b, w2_b, G_b_);
     getGMultiband(1, k1_b, k2_b, w1_b, w2_b, G_b_, sign);
     // Assuming 0 = down 1 = up
@@ -820,7 +820,7 @@ void TpAccumulator<Parameters, DT, linalg::CPU>::updateG4SpinDifference(
     // if sign == -1
     // G_a_ = G_down(k1_a, k2_a, w1_a, w2_a) - G_up(k1_a, k2_a, w1_a, w2_a)
     // G_b_ = G_down(k1_b, k2_b, w1_b, w2_b) - G_up(k1_b, k2_b, w1_b, w2_b)
-    
+
     if (!cross_legs)
       for (int b4 = 0; b4 < n_bands_; ++b4)
         for (int b3 = 0; b3 < n_bands_; ++b3)

include/dca/phys/dca_step/cluster_solver/shared_tools/cluster_helper.cuh

@@ -29,8 +29,10 @@ namespace details {
 
 class ClusterHelper {
 public:
-  // Initialize real reciprocal cluster if mometnum == true.
-  static void set(int nc, const int* add, int lda, const int* sub, int lds, bool momentum);
+  /// Initialize real Cluster
+  static void set(int nc, const int* add, int lda, const int* sub, int lds);
+  /// Initialize reciprocal cluster 
+  static void setMomentum(int nc, const int* add, int lda, const int* sub, int lds);
 
   // Returns the index of id_1 + id_2.
   __DEVICE__ inline int add(int id_1, int id_2) const;
@@ -39,12 +41,14 @@ public:
   // returns the index of -id
   __DEVICE__ inline int minus(int id) const;
 
-private:
   int nc_;
   const int* add_matrix_;
   const int* sub_matrix_;
 };
 
+extern __CONSTANT__ int* cluster_add_matrix;
+extern __CONSTANT__ int* cluster_sub_matrix;
+
 // Global instance for real space and momentum clusters.
 extern __DEVICE__ __CONSTANT__ ClusterHelper cluster_real_helper;
 extern __DEVICE__ __CONSTANT__ ClusterHelper cluster_momentum_helper;