multiple solvers and toml input parser (#74)

This PR changes the config file of solid_dmft to the toml format and allows to use multiple solvers within a single calculation. Requires dft_tools from commit b355173 on the unstable branch or later.

- The following input parameters can now be a list per impurity:
    - `general_params`: U, J, U_prime, ratio_F4_F2, h_int_type, enforce_off_diag, dc_type
    - `advanced_params`: dc_U, dc_J, dc_fixed_occ, map_solver_struct, pick_solver_struct, mapped_solver_struct_degeneracies
- Multiple solvers can be used, which only solve the impurity problems specified in `idx_impurities`
    - general parameter `solver_type` moved to solver section and renamed to `type`
    - general parameter `n_l` moved to solver section
    - general parameter `measure_chi` moved to solver section
    - general parameter `delta_interface` moved to solver section
- All possible input parameters are defined in the `python/solid_dmft/io_tools/default.toml`
- according to toml format the config file is now called .toml (instead of .ini), and boolean are not capitalized, strings are given with quotes and lists are given with brackets.
- Documentation of the input is now generated from `python/solid_dmft/io_tools/documentation.txt`
- For an example, refer to the new integration test (see below)
- Updated interface to python scripts wrapping solid_dmft: new routine `main.run_dmft` that expects the params as python dictionaries, which are then supplemented with the defaults etc equivalent to what happens when reading in a toml file

- the existence of the parameter `general_params['beta']` now determines if a imaginary- or real-frequency grid is used within solid_dmft
- Bug fix: Slater interaction for p orbitals can now be constructed
- Renaming of solver parameters for the different solvers is now moved to `solver.py`. The idea is that every other part of solid_dmft should care as little as possible what solvers are used, with the details abstracted by the SolverStructure class
    - In `solver.py`, all solver parameters that are passed to the triqs solver are transferred to a dict `triqs_solver_params`. When adding new triqs solver parameter to solid_dmft in the future, they also need to be added within solver.py.
- In the determination of the block structure, the largely unused parameter `general_params['block_suppress_orbital_symm']` removed. Its behavior can be replaced by using `advanced_params['mapped_solver_struct_degeneracies']`
- Integration tests: previously existing tests updated, new tests added. One with ftps solver (requires installation of ftps, otherwise just passes without doing anything) and one with a combination of CT-HYB and Hartree solver
- Unit tests: added test for toml-related functionality
- `read_config.py` removed and the functionality for dealing with the dicts from reading a toml file moved to `postproc_toml_dict.py`
- `io_tools/verify_input_params.py` contains all checks of the input params that the code performs before starting the DMFT calculations
- Updated the documentation of the input parameters

---------

Co-authored-by: Alexander Hampel <[email protected]>

doc/documentation.rst

@@ -37,10 +37,10 @@ Input/Output
 
 Further details for running
 ===========================
-   
+
 .. toctree::
     :maxdepth: 1
-    
+
     md_notes/docker.md
     md_notes/run_locally.md
     md_notes/run_cluster.md
@@ -57,10 +57,10 @@ Module reference manual
     dft_managers
     dmft_cycle
     dmft_tools
+    io_tools
     postprocessing
-    read_config
     util
-
 
-
+
+
 

doc/input_output/DMFT_input/generate_doc_from_comments.sh

@@ -7,25 +7,23 @@
 green='\033[1;32m'
 wipe="\033[1m\033[0m"
 
-echo -e "${green}Extracting the comments from read_config.py file${wipe}"
+echo -e "${green}Extracting the comments from io_tools/documentation.txt file${wipe}"
 buildfolder="../../.."
-docfile="$buildfolder/python/solid_dmft/read_config.py"
+docfile="$buildfolder/python/solid_dmft/io_tools/documentation.txt"
 
-awk '/---XXX---start/{flag=1; c=0} flag; /---XXX/&& ++c==2{flag=0}' $docfile | tail -n +2 | head -n -2 > python_comments.txt
-
-#from the python comments 
+#from the documentation file
 echo "Generating .rst syntax from the python syntax"
-sed 's/\(.*\) :\s*\(.*\)/\n.. admonition:: \1 \n 	:class: intag  \n \n            \*\*type=\*\* \2\n/g' python_comments.txt > matches_comments.txt
+sed 's/\(.*\) :\s*\(.*\)/\n.. admonition:: \1 \n 	:class: intag  \n \n            \*\*type\*\* = \2\n/g' $docfile > matches_comments.txt
 
 #add blank line after type
 #sed -i 's/\(\*type=.*\),/\1 \n\n           /g'  matches_comments.txt
 #sed -i 's/\(\*type=.*\),/\1 \n\n           /g'  matches_comments.txt
 
-# make 'optional' and 'default' bold
-sed -i 's/,.*\(\optional\)/;  \*\*\1\*\*/g'  matches_comments.txt
-sed -i 's/,.*\(\default=\)/;  \*\*\1\*\* /g'  matches_comments.txt
+# make 'mandatory' and 'default' bold
+sed -i 's/,.*\(\mandatory\)/;  \*\*\1\*\*/g'  matches_comments.txt
+sed -i 's/,.*\(\default\)/;  \*\*\1\*\* /g'  matches_comments.txt
 
-# grep all admonitions and store them in a file 
+# grep all admonitions and store them in a file
 
 echo "Generating input page"
 cat > input.rst << EOF
@@ -51,7 +49,7 @@ EOF
 #
 # awk '/\[  SECTION  \]/{flag=1; c=0} flag; /\[ /&& ++c==2{flag=0}'
 # matches from the pattern '[  SECTION  ]'(note, two spaces are important) up to either the next occurence of
-# '[  ', which is the next group, or the end of the file 
+# '[  ', which is the next group, or the end of the file
 #
 #
 # the second part takes the divided section and extracts a list out of it:
@@ -99,7 +97,7 @@ cat > general.rst << EOF
 [general]: General parameters
 -----------------------------
 
-Includes the majority of the parameters
+Frequently used parameters that apply to the whole simulation.
 
 
 EOF
@@ -113,10 +111,9 @@ cat > solver.rst << EOF
 [solver]: solver specific parameters
 ------------------------------------
 
-Here are the parameters that are uniquely dependent on the solver chosen. Below a list of the supported solvers:
-
-
-
+Here are the parameters that are uniquely dependent on the solver chosen. Some parameters are used within solid_dmft and some are passed directly into the triqs solver.
+To see which parameters were passed to the solver for a given calculation, look at the triqs_solver_params in DMFT_input/solver in the h5 archive.
+Solver-specific parameters are listed in the respective sections.
 
 
 EOF
@@ -125,16 +122,14 @@ echo -e "\n"  >> solver.rst
 cat solver.tmp >> solver.rst
 ##############
 
-
-
 ###############
 cat > dft.rst << EOF
 
 [dft]: DFT related inputs
 -------------------------
 
-List of parameters that relate to the DFT calculation, useful mostly when doing CSC.
-
+List of parameters for the DFT calculation in charge-self-consistent calculations.
+The parameters are ignored in one-shot calculations.
 
 
 EOF
@@ -149,7 +144,7 @@ cat > advanced.rst << EOF
 [advanced]: Advanced inputs
 ---------------------------
 
-Advanced parameters, do not modify default value unless you know what you are doing
+Advanced parameters, do not modify the default value unless you know what you are doing.
 
 
 EOF

doc/tutorials/Ce2O3_csc_w90/dmft_config.toml

@@ -0,0 +1,44 @@
+[general]
+seedname = "ce2o3"
+jobname = "b10-U6.46-J0.46"
+csc = true
+
+eta = 0.5
+n_iw = 100
+n_tau = 5001
+
+n_iter_dmft_first = 2
+n_iter_dmft_per = 1
+n_iter_dmft = 5
+
+block_threshold = 1e-03
+
+h_int_type = "density_density"
+U = 6.46
+J = 0.46
+beta = 10
+prec_mu = 0.1
+
+sigma_mix = 1.0
+g0_mix = 1.0
+dc_type = 0
+dc = true
+dc_dmft = true
+calc_energies = true
+
+h5_save_freq = 1
+
+[solver]
+type = "hubbardI"
+n_l = 15
+store_solver = false
+measure_G_l = false
+measure_density_matrix = true
+
+[dft]
+dft_code = "qe"
+n_cores = 10
+mpi_env = "default"
+projector_type = "w90"
+dft_exec = "pw.x"
+w90_tolerance = 1e-1

doc/tutorials/Ce2O3_csc_w90/tutorial.ipynb

@@ -117,14 +117,14 @@
     "### DMFT\n",
     "\n",
     "1. Wannier90Converter: [ce2o3.inp](./dft_input/ce2o3.inp)\n",
-    "2. solid_dmft: [dmft_config.ini](./dmft_config.ini)\n",
+    "2. solid_dmft: [dmft_config.toml](./dmft_config.toml)\n",
     "\n",
     "Here we'll discuss the most important input flags for solid_dmft:"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 1,
    "id": "165c087b",
    "metadata": {},
    "outputs": [
@@ -133,12 +133,10 @@
      "output_type": "stream",
      "text": [
       "[general]\n",
-      "seedname = ce2o3\n",
-      "jobname = b10-U6.46-J0.46\n",
-      "csc = True\n",
-      "#dft_mu = 0.\n",
-      "solver_type = hubbardI\n",
-      "n_l = 15\n",
+      "seedname = \"ce2o3\"\n",
+      "jobname = \"b10-U6.46-J0.46\"\n",
+      "csc = true\n",
+      "\n",
       "eta = 0.5\n",
       "n_iw = 100\n",
       "n_tau = 5001\n",
@@ -149,7 +147,7 @@
       "\n",
       "block_threshold = 1e-03\n",
       "\n",
-      "h_int_type = density_density\n",
+      "h_int_type = \"density_density\"\n",
       "U = 6.46\n",
       "J = 0.46\n",
       "beta = 10\n",
@@ -158,30 +156,31 @@
       "sigma_mix = 1.0\n",
       "g0_mix = 1.0\n",
       "dc_type = 0\n",
-      "dc = True\n",
-      "dc_dmft = True\n",
-      "calc_energies = True\n",
+      "dc = true\n",
+      "dc_dmft = true\n",
+      "calc_energies = true\n",
       "\n",
       "h5_save_freq = 1\n",
       "\n",
       "[solver]\n",
-      "store_solver = False\n",
-      "measure_G_l = False\n",
-      "measure_density_matrix = True\n",
+      "type = \"hubbardI\"\n",
+      "n_l = 15\n",
+      "store_solver = false\n",
+      "measure_G_l = false\n",
+      "measure_density_matrix = true\n",
       "\n",
       "[dft]\n",
-      "dft_code = qe\n",
+      "dft_code = \"qe\"\n",
       "n_cores = 10\n",
-      "mpi_env = default\n",
-      "projector_type = w90\n",
-      "dft_exec = \n",
-      "w90_exec = wannier90.x\n",
+      "mpi_env = \"default\"\n",
+      "projector_type = \"w90\"\n",
+      "dft_exec = \"pw.x\"\n",
       "w90_tolerance = 1.e-1\n"
      ]
     }
    ],
    "source": [
-    "!cat ./dmft_config.ini"
+    "!cat ./dmft_config.toml"
    ]
   },
   {
@@ -509,7 +508,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "Python 3",
    "language": "python",
    "name": "python3"
   },
@@ -523,7 +522,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.12"
+   "version": "3.10.10"
   }
  },
  "nbformat": 4,

doc/tutorials/NNO_os_plo_mag/config.ini → doc/tutorials/NNO_os_plo_mag/config.toml

@@ -1,47 +1,47 @@
 [general]
-seedname = nno
-jobname = NNO_lowT
+seedname = "nno"
+jobname = "NNO_lowT"
 
-enforce_off_diag = False
+enforce_off_diag = false
 block_threshold = 0.001
 
-solver_type = cthyb
 n_iw = 2001
 n_tau = 20001
 
 prec_mu = 0.001
 
-h_int_type = density_density
+h_int_type = "density_density"
 U = 8.0
 J = 1.0
 
 # temperature ~290 K
 beta = 40
 
-magnetic = True
-magmom = -0.3, 0.3
-afm_order = True
+magnetic = true
+magmom = [-0.3, 0.3]
+afm_order = true
 
 n_iter_dmft = 14
 
 g0_mix = 0.9
 
 dc_type = 0
-dc = True
-dc_dmft = False
+dc = true
+dc_dmft = false
 
-load_sigma = False
-path_to_sigma = pre_AFM.h5
+load_sigma = false
+path_to_sigma = "pre_AFM.h5"
 
 [solver]
+type = "cthyb"
 length_cycle = 2000
 n_warmup_cycles = 5e+3
 n_cycles_tot = 1e+7
 imag_threshold = 1e-5
 
-perform_tail_fit = True
+perform_tail_fit = true
 fit_max_moment = 6
 fit_min_w = 10
 fit_max_w = 16
-measure_density_matrix = True
+measure_density_matrix = true