refactor cholesky module usage, remove force_tilesize option and upda…

…te docs

.gitignore

@@ -12,6 +12,7 @@ Release/
 cmake-build-debug/
 cmake-build-release/
 build/
+ebuild/
 debug*
 build*
 .xlsx

ci/reference_output/ch4.def2-tzvp.ccsd_t.json

@@ -450,7 +450,6 @@
       "convd": 1e-06,
       "diis_hist": 20,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "scf_type": "unrestricted",
       "multiplicity": 3,
       "lambdas": [],

ci/reference_output/co.cc-pvdz.gfccsd.json

@@ -507,7 +507,6 @@
       "convd": 1e-11,
       "diis_hist": 10,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "damp": 100,
       "debug": "false",
       "restart": "false",
@@ -535,7 +534,6 @@
     "CCSD": {
       "threshold": 1e-06,
       "tilesize": 40,
-      "force_tilesize": "false",
       "lshift": 0.0,
       "ndiis": 5,
       "readt": "false",

ci/reference_output/cr2.def2-svp.scf.json

@@ -306,7 +306,6 @@
       "convd": 1e-07,
       "diis_hist": 10,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "damp": 100,
       "debug": "false",
       "restart": "false",

ci/reference_output/h2o_ducc.cc-pvdz.ducc.json

@@ -249,7 +249,6 @@
       "convd": 1e-08,
       "diis_hist": 10,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "scf_type": "restricted",
       "multiplicity": 1,
       "lambdas": [],

ci/reference_output/h2o_eom.cc-pvdz.eomccsd.json

@@ -611,7 +611,6 @@
       "convd": 1e-08,
       "diis_hist": 10,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "scf_type": "restricted",
       "multiplicity": 1,
       "lambdas": [],

ci/reference_output/ozone.sto-3g.rt-eomccsd.json

@@ -116,7 +116,6 @@
       "convd": 0.1,
       "diis_hist": 10,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "scf_type": "unrestricted",
       "multiplicity": 2
     },

ci/reference_output/ubiquitin_dgrtl/ubiquitin_dgrtl.6-31g.ccsd_t.json

@@ -303,7 +303,6 @@
       "convd": 1e-07,
       "diis_hist": 12,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "scf_type": "restricted",
       "multiplicity": 1,
       "lambdas": [],

ci/reference_output/ubiquitin_dgrtl/ubiquitin_dgrtl.cc-pvdz.ccsd_t.json

@@ -295,7 +295,6 @@
       "convd": 1e-07,
       "diis_hist": 12,
       "AO_tilesize": 37,
-      "force_tilesize": "false",
       "scf_type": "restricted",
       "multiplicity": 1,
       "lambdas": [],

ci/reference_output/ubiquitin_dgrtl/ubiquitin_dgrtl.sto-3g.ccsd_t.json

@@ -435,7 +435,6 @@
       "convd": 1e-09,
       "diis_hist": 12,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "scf_type": "restricted",
       "multiplicity": 1
     },

ci/reference_output/uracil.cc-pvdz.ccsd_t.json

@@ -320,7 +320,6 @@
       "convd": 1e-7,
       "diis_hist": 10,
       "AO_tilesize": 30,
-      "force_tilesize": "false",
       "scf_type": "restricted",
       "multiplicity": 1
     },

docs/schema/input_schema.json

@@ -93,9 +93,6 @@
         "diis_hist": {
           "type": "number"
         },
-        "force_tilesize": {
-          "type": "boolean"
-        },
         "tilesize": {
           "type": "number"
         },
@@ -287,9 +284,6 @@
         "threshold": {
           "type": "number"
         },
-        "force_tilesize": {
-          "type": "boolean"
-        },
         "tilesize": {
           "type": "number"
         },

docs/user_guide/basic_options.rst

@@ -52,7 +52,7 @@ Some common options supported are as follows:
 :file_prefix: A string indicating the prefix for the name of the workspace folder where the results of a run are stored.
    It also forms the prefix for the files written to the workspace folder. The *default prefix* is the name of the input file without the *.json* extension.
 
-:output_dir: A string indicating the path to the directory where the where the results of a run are stored. If not provided, the default is the current working directory.
+:output_dir: A string indicating the path to an existing directory where the where the results of a run are stored. If not provided, the default is the current working directory.
 
 .. _Basis:
 

docs/user_guide/coupledcluster.rst

@@ -19,7 +19,6 @@ The following CCSD options are supported. The remaining CC methods (CC2, Lambda,
 
  "CC": {
    "threshold": 1e-6,
-   "force_tilesize": false,
    "tilesize": 50,
    "lshift": 0,
    "ndiis": 5,
@@ -45,11 +44,14 @@ The following CCSD options are supported. The remaining CC methods (CC2, Lambda,
    }   
   }
 
+
 :threshold: ``[default=1e-6]`` Specifies the convergence threshold of iterative solutions of amplitude equations. The threshold refers to the norm of residual, namely, the deviation from the amplitude equations.
 
-:force_tilesize: ``[default=false]``
+:tilesize: ``[default=40]`` The tilesize for the MO or MSO space. An integer value that is automatically set unless specified explicitly by the user. It is recommended to let the CC module determine this value automatically.
+
+   .. warning::
 
-:tilesize: ``[default=40]`` The tilesize for the MO or MSO space. An integer value that is automatically set. If **force_tilesize=true**, the value specified by the user is respected. It is recommended to let the CC module automatically determine this value.
+      The automatically determined tilesize may vary depending on node count and the number of processes per node. When restarting a calculation with different node counts, the user is expected to explicitly set the tilesize to match the value used in the initial run.
 
 :lshift: ``[default=0]`` The level shift option that increases small orbital energy differences used in calculating the updates for cluster amplitudes. When calculating ground states with multi-configurational character or if convergence is slow, one may need to specify a typical values for lshift between 0.3 and 0.5.
 

docs/user_guide/scf.rst

@@ -28,7 +28,6 @@ The values listed below are the defaults where few options are automatically adj
    "conve": 1e-8,
    "convd": 1e-7,
    "diis_hist": 10,
-   "force_tilesize": false,
    "tilesize": 30,
    "damp": 100,
    "nnodes": 1,
@@ -97,10 +96,8 @@ The values listed below are the defaults where few options are automatically adj
 
 :diis_hist: ``[default=10]`` Specifies the number of DIIS history entries to store for the fock and error matrices.
 
-:force_tilesize: ``[default=false]``
-
-:tilesize: The tilesize for the AO dimension. An integer value that is automatically set to ``ceil(Nbf * 0.05)``. If **force_tilesize=true**, 
-   the value specified by the user is respected. It is recommended to let the SCF module automatically determine this value.
+:tilesize: The tilesize for the AO dimension. An integer value that is automatically set to ``ceil(Nbf * 0.05)``
+ if the user has not explicitly set it. It is recommended to let the SCF module automatically determine this value.
 
 :damp: damping (mixing) factor for the density matrix where :math:`0 \leq \alpha \leq 100`.  Specifies the percentage of the current iterations density mixed with the previous iterations density. ``default=100`` indicates no damping.
 

exachem/cc/cc2/cc2_canonical.cpp

@@ -281,7 +281,7 @@ cc2_v2_driver(ChemEnv& chem_env, ExecutionContext& ec, const TiledIndexSpace& MO
   }
 
   chem_env.cc_context.cc2_correlation_energy = energy;
-  chem_env.cc_context.cc2_total_energy       = chem_env.hf_energy + energy;
+  chem_env.cc_context.cc2_total_energy       = chem_env.scf_context.hf_energy + energy;
 
   if(ec.pg().rank() == 0) {
     sys_data.results["output"]["CC2"]["n_iterations"]                = niter + 1;
@@ -304,15 +304,15 @@ void cc2_canonical_driver(ExecutionContext& ec, ChemEnv& chem_env) {
 
   scf(ec, chem_env);
 
-  double              hf_energy      = chem_env.hf_energy;
+  double              hf_energy      = chem_env.scf_context.hf_energy;
   libint2::BasisSet   shells         = chem_env.shells;
-  Tensor<T>           C_AO           = chem_env.C_AO;
-  Tensor<T>           C_beta_AO      = chem_env.C_beta_AO;
-  Tensor<T>           F_AO           = chem_env.F_AO;
-  Tensor<T>           F_beta_AO      = chem_env.F_beta_AO;
-  TiledIndexSpace     AO_opt         = chem_env.AO_opt;
-  std::vector<size_t> shell_tile_map = chem_env.shell_tile_map;
-  bool                scf_conv       = chem_env.no_scf;
+  Tensor<T>           C_AO           = chem_env.scf_context.C_AO;
+  Tensor<T>           C_beta_AO      = chem_env.scf_context.C_beta_AO;
+  Tensor<T>           F_AO           = chem_env.scf_context.F_AO;
+  Tensor<T>           F_beta_AO      = chem_env.scf_context.F_beta_AO;
+  TiledIndexSpace     AO_opt         = chem_env.is_context.AO_opt;
+  std::vector<size_t> shell_tile_map = chem_env.scf_context.shell_tile_map;
+  bool                scf_conv       = chem_env.scf_context.no_scf;
 
   SystemData&  sys_data     = chem_env.sys_data;
   CCSDOptions& ccsd_options = chem_env.ioptions.ccsd_options;

exachem/cc/cc2/cd_cc2.cpp

@@ -13,58 +13,37 @@ namespace fs = std::filesystem;
 
 namespace exachem::cc2 {
 void cd_cc2_driver(ExecutionContext& ec, ChemEnv& chem_env) {
-  using T = double;
-
+  using T   = double;
   auto rank = ec.pg().rank();
 
-  scf::scf_driver(ec, chem_env);
-
-  double              hf_energy      = chem_env.hf_energy;
-  libint2::BasisSet   shells         = chem_env.shells;
-  Tensor<T>           C_AO           = chem_env.C_AO;
-  Tensor<T>           C_beta_AO      = chem_env.C_beta_AO;
-  Tensor<T>           F_AO           = chem_env.F_AO;
-  Tensor<T>           F_beta_AO      = chem_env.F_beta_AO;
-  TiledIndexSpace     AO_opt         = chem_env.AO_opt;
-  TiledIndexSpace     AO_tis         = chem_env.AO_tis;
-  std::vector<size_t> shell_tile_map = chem_env.shell_tile_map;
-  bool                scf_conv       = chem_env.no_scf;
-
-  SystemData   sys_data     = chem_env.sys_data;
-  CCSDOptions& ccsd_options = chem_env.ioptions.ccsd_options;
-  auto         debug        = ccsd_options.debug;
-  if(rank == 0) ccsd_options.print();
+  cholesky_2e::cholesky_2e_driver(ec, chem_env);
 
-  if(rank == 0)
-    cout << endl << "#occupied, #virtual = " << sys_data.nocc << ", " << sys_data.nvir << endl;
+  const bool is_rhf = chem_env.sys_data.is_restricted;
 
-  auto [MO, total_orbitals] = cholesky_2e::setupMOIS(ec, chem_env);
+  std::string files_prefix = chem_env.get_files_prefix();
 
-  std::string out_fp       = chem_env.workspace_dir;
-  std::string files_dir    = out_fp + chem_env.ioptions.scf_options.scf_type + "/cc2";
-  std::string files_prefix = /*out_fp;*/ files_dir + "/" + sys_data.output_file_prefix;
-  std::string f1file       = files_prefix + ".f1_mo";
-  std::string t1file       = files_prefix + ".t1amp";
-  std::string t2file       = files_prefix + ".t2amp";
-  std::string v2file       = files_prefix + ".cholv2";
-  std::string cholfile     = files_prefix + ".cholcount";
-  std::string ccsdstatus   = files_prefix + ".ccsdstatus";
-
-  const bool is_rhf = sys_data.is_restricted;
+  CDContext& cd_context = chem_env.cd_context;
+  CCContext& cc_context = chem_env.cc_context;
+  cc_context.init_filenames(files_prefix);
+  CCSDOptions& ccsd_options = chem_env.ioptions.ccsd_options;
 
-  bool ccsd_restart = ccsd_options.readt || ((fs::exists(t1file) && fs::exists(t2file) &&
-                                              fs::exists(f1file) && fs::exists(v2file)));
+  auto              debug      = ccsd_options.debug;
+  bool              scf_conv   = chem_env.scf_context.no_scf;
+  std::string       t1file     = cc_context.t1file;
+  std::string       t2file     = cc_context.t2file;
+  const std::string ccsdstatus = cc_context.ccsdstatus;
 
-  // deallocates F_AO, C_AO
-  auto [cholVpr, d_f1, lcao, chol_count, max_cvecs, CI] =
-    exachem::cholesky_2e::cholesky_2e_driver<T>(chem_env, ec, MO, AO_opt, C_AO, F_AO, C_beta_AO,
-                                                F_beta_AO, shells, shell_tile_map, ccsd_restart,
-                                                cholfile);
-  free_tensors(lcao);
+  bool ccsd_restart = ccsd_options.readt ||
+                      ((fs::exists(t1file) && fs::exists(t2file) && fs::exists(cd_context.f1file) &&
+                        fs::exists(cd_context.v2file)));
 
   if(ccsd_options.writev) ccsd_options.writet = true;
 
-  TiledIndexSpace N = MO("all");
+  TiledIndexSpace& MO      = chem_env.is_context.MSO;
+  TiledIndexSpace& CI      = chem_env.is_context.CI;
+  TiledIndexSpace  N       = MO("all");
+  Tensor<T>        d_f1    = chem_env.cd_context.d_f1;
+  Tensor<T>        cholVpr = chem_env.cd_context.cholV2;
 
   std::vector<T>         p_evl_sorted;
   Tensor<T>              d_r1, d_r2, d_t1, d_t2;
@@ -78,31 +57,13 @@ void cd_cc2_driver(ExecutionContext& ec, ChemEnv& chem_env) {
       ec, MO, d_f1, ccsd_options.ndiis, ccsd_restart && fs::exists(ccsdstatus) && scf_conv);
 
   if(ccsd_restart) {
-    read_from_disk(d_f1, f1file);
     if(fs::exists(t1file) && fs::exists(t2file)) {
       read_from_disk(d_t1, t1file);
       read_from_disk(d_t2, t2file);
     }
-    read_from_disk(cholVpr, v2file);
-    ec.pg().barrier();
     p_evl_sorted = tamm::diagonal(d_f1);
   }
 
-  else if(ccsd_options.writet) {
-    // fs::remove_all(files_dir);
-    if(!fs::exists(files_dir)) fs::create_directories(files_dir);
-
-    write_to_disk(d_f1, f1file);
-    write_to_disk(cholVpr, v2file);
-
-    if(rank == 0) {
-      std::ofstream out(cholfile, std::ios::out);
-      if(!out) cerr << "Error opening file " << cholfile << endl;
-      out << chol_count << std::endl;
-      out.close();
-    }
-  }
-
   if(rank == 0 && debug) {
     print_vector(p_evl_sorted, files_prefix + ".eigen_values.txt");
     cout << "Eigen values written to file: " << files_prefix + ".eigen_values.txt" << endl << endl;
@@ -114,7 +75,7 @@ void cd_cc2_driver(ExecutionContext& ec, ChemEnv& chem_env) {
 
   ccsd_restart = ccsd_restart && fs::exists(ccsdstatus) && scf_conv;
 
-  std::string fullV2file = files_prefix + ".fullV2";
+  std::string fullV2file = cd_context.fullV2file;
   t1file                 = files_prefix + ".fullT1amp";
   t2file                 = files_prefix + ".fullT2amp";
 
@@ -145,7 +106,7 @@ void cd_cc2_driver(ExecutionContext& ec, ChemEnv& chem_env) {
     }
   }
 
-  ccsd_stats(ec, hf_energy, residual, corr_energy, ccsd_options.threshold);
+  ccsd_stats(ec, chem_env.scf_context.hf_energy, residual, corr_energy, ccsd_options.threshold);
 
   if(ccsd_options.writet && !fs::exists(ccsdstatus)) {
     // write_to_disk(d_t1,t1file);

exachem/cc/cc2/cd_cc2_cs.cpp

@@ -537,7 +537,7 @@ std::tuple<double, double> cc2_cs::cd_cc2_cs_driver(
   }
 
   chem_env.cc_context.cc2_correlation_energy = energy;
-  chem_env.cc_context.cc2_total_energy       = chem_env.hf_energy + energy;
+  chem_env.cc_context.cc2_total_energy       = chem_env.scf_context.hf_energy + energy;
 
   if(ec.pg().rank() == 0) {
     sys_data.results["output"]["CC2"]["n_iterations"]                = niter + 1;

exachem/cc/cc2/cd_cc2_os.cpp

@@ -719,7 +719,7 @@ std::tuple<double, double> cc2_os::cd_cc2_os_driver(
   }
 
   chem_env.cc_context.cc2_correlation_energy = energy;
-  chem_env.cc_context.cc2_total_energy       = chem_env.hf_energy + energy;
+  chem_env.cc_context.cc2_total_energy       = chem_env.scf_context.hf_energy + energy;
 
   if(ec.pg().rank() == 0) {
     sys_data.results["output"]["CC2"]["n_iterations"]                = niter + 1;