threadpoolexecutor for distributing MPI

.github/workflows/continous_integration.yml

@@ -120,8 +120,10 @@ jobs:
         run: |
           sudo apt-get update
           sudo apt-get install -y texlive texlive-latex-extra texlive-fonts-recommended dvipng cm-super
+
       - name: Install Python dependencies
         run: |
+          conda install mpi4py
           python -m pip install --upgrade git+https://github.com/bitranox/wrapt_timeout_decorator.git
           python -m pip install pytest pytest-cov flake8 pytest-aiohttp sqlparse freezegun PyJWT joblib coveralls
           python -m pip install -r  ml_requirements.txt -r grb_requirements.txt -r tf_requirements.txt -r sklearn_requirements.txt

example_files/multi_config_analysis/config.yaml

@@ -0,0 +1,75 @@
+Me2017_1:
+  model: Me2017
+  outdir: outdir_1
+  label: injection
+  prior: ../../priors/Me2017.prior
+  tmin: 0.1
+  tmax: 20
+  dt: 0.5
+  error-budget: 1
+  nlive: 40
+  Ebv-max: 0
+  injection: ./injection.json
+  injection-num: 0
+  injection-outfile: outdir_1/lc.csv
+  generation-seed: 42
+  filters: ztfg,ztfr,ztfi
+  remove-nondetections: True
+  #process-per-config: 4
+
+Me2017_2:
+  model: Me2017
+  outdir: outdir_2
+  label: injection
+  prior: ../../priors/Me2017.prior
+  tmin: 0.1
+  tmax: 20
+  dt: 0.5
+  error-budget: 1
+  nlive: 40
+  Ebv-max: 0
+  injection: ./injection.json
+  injection-num: 0
+  injection-outfile: outdir_2/lc.csv
+  generation-seed: 42
+  filters: ztfg,ztfr,ztfi
+  remove-nondetections: True
+  #process-per-config: 5
+
+Me2017_3:
+  model: Me2017
+  outdir: outdir_3
+  label: injection
+  prior: ../../priors/Me2017.prior
+  tmin: 0.1
+  tmax: 20
+  dt: 0.5
+  error-budget: 1
+  nlive: 40
+  Ebv-max: 0
+  injection: ./injection.json
+  injection-num: 0
+  injection-outfile: outdir_3/lc.csv
+  generation-seed: 42
+  filters: ztfg,ztfr,ztfi
+  remove-nondetections: True
+  #process-per-config: 6
+
+Me2017_4:
+  model: Me2017
+  outdir: outdir_4
+  label: injection
+  prior: ../../priors/Me2017.prior
+  tmin: 0.1
+  tmax: 20
+  dt: 0.5
+  error-budget: 1
+  nlive: 40
+  Ebv-max: 0
+  injection: ./injection.json
+  injection-num: 0
+  injection-outfile: outdir_4/lc.csv
+  generation-seed: 42
+  filters: ztfg,ztfr,ztfi
+  remove-nondetections: True
+  #process-per-config: 7

example_files/multi_config_analysis/injection.json

@@ -0,0 +1,31 @@
+{
+    "injections": {
+        "__dataframe__": true,
+        "content": {
+            "simulation_id": [
+                0
+            ],
+            "luminosity_distance": [
+                152.52372678074587
+            ],
+            "beta": [
+                3.1674793717549354
+            ],
+            "log10_kappa_r": [
+                1.9033736926068743
+            ],
+            "timeshift": [
+                -1.640734750742769
+            ],
+            "log10_vej": [
+                -1.0471069387482033
+            ],
+            "log10_mej": [
+                -2.9158814967880233
+            ],
+            "geocent_time": [
+                0.09662692256248354
+            ]
+        }
+    }
+}

nmma/em/multi_config_analysis.py

@@ -0,0 +1,107 @@
+import yaml
+import subprocess
+from .analysis import get_parser
+from pathlib import Path
+import argparse
+from concurrent.futures import ThreadPoolExecutor
+import sys
+import os
+
+
+def get_parser_here():
+    parser = argparse.ArgumentParser(description="Multi config analysis script for NMMA.")
+
+    parser.add_argument(
+        "--config",
+        type=str,
+        help="Name of the configuration file containing parameter values.",
+    )
+
+    parser.add_argument(
+        "--parallel",
+        action="store_true",
+        default=False,
+        help="To run multiple configs in parallel",
+    )
+
+    parser.add_argument(
+        "-p",
+        "--process",
+        type=int,
+        help=(
+            "No of processess each configuration should have, if --parallel is set then process will be divided equally among"
+            " all configs, else each config will run sequentially with given no of process. Strictly required if"
+            " --process-per-config is not given"
+        ),
+    )
+    parser.add_argument(
+        "--process-per-config",
+        type=int,
+        help=(
+            "If multiple configurations are given, how many MPI process should be assigned to each configuration. In the yaml"
+            " file, indicate the number of process for each configuration with the key 'process-per-config'. If not given, all"
+            " configurations will be run depending on the state and value of --parallel and --process. This takes precedence"
+            " over --process"
+        ),
+    )
+
+    return parser
+
+
+def run_cmd_in_subprocess(cmd):
+    subprocess.run(cmd)
+
+
+def main(args=None):
+    parser = get_parser_here()
+    args, _ = parser.parse_known_args(namespace=args)
+
+    _parser = get_parser()
+    main_args = _parser.parse_args([])
+
+    yaml_dict = yaml.safe_load(os.path.expandvars(Path(args.config).read_text()))
+
+    total_configs = len(list(yaml_dict.keys()))
+    futures = []
+
+    with ThreadPoolExecutor() as executor:
+        for analysis_set in yaml_dict.keys():
+            params = yaml_dict[analysis_set]
+
+            if "process-per-config" in params or args.process is None:
+                processes = params["process-per-config"]
+            elif args.parallel and args.process is not None:
+                processes = args.process // total_configs
+            else:
+                processes = args.process
+
+            cmd = ["mpiexec", "-np", str(processes), "lightcurve-analysis"]
+
+            for key, value in params.items():
+                key = key.replace("-", "_")
+
+                if key == "process_per_config":
+                    continue
+
+                if key not in main_args:
+                    print(f"{key} not a known argument... please remove")
+                    exit()
+                key = key.replace("_", "-")
+
+                if isinstance(value, bool) and value:
+                    cmd.append(f"--{key}")
+                else:
+                    cmd.append(f"--{key}")
+                    cmd.append(str(value))
+
+            if not args.parallel:
+                print(f"{'#'*100}")
+                print(f"Running analysis set:  {analysis_set} with {processes} processes")
+                run_cmd_in_subprocess(cmd)
+                print(f"{'#'*100}")
+            else:
+                future = executor.submit(run_cmd_in_subprocess, cmd)
+                futures.append(future)
+    if args.parallel:
+        for future in futures:
+            future.result()

nmma/tests/data/multi_config_analysis/Me2017.prior

@@ -0,0 +1,6 @@
+luminosity_distance = Uniform(minimum=0.0, maximum=200., name='luminosity_distance',latex_label='$D_L$')
+beta = Uniform(name='beta', minimum=1., maximum=5., latex_label='$\\beta$')
+log10_kappa_r = Uniform(name='log10_kappa_r', minimum=-1., maximum=2., latex_label='$\\log_{10}\\kappa_r$')
+timeshift = Uniform(minimum=-2.0, maximum=1.0, name='trigger_time',latex_label='$t_0$')
+log10_vej  = Uniform(name='log10_vej', minimum=-2., maximum=-0.5, latex_label='$\\log_{10}v^{\\rm{ej}}$')
+log10_mej = Uniform(name='log10_mej', minimum=-3., maximum=-0.5, latex_label='$\\log_{10}M^{\\rm{ej}}$')

nmma/tests/data/multi_config_analysis/config.yaml

@@ -0,0 +1,35 @@
+Me2017_1:
+  model: Me2017
+  outdir: ${WORKING_DIR}/outdir_1
+  label: injection
+  prior: ${WORKING_DIR}/data/multi_config_analysis/Me2017.prior
+  tmin: 0.1
+  tmax: 5
+  dt: 0.5
+  error-budget: 1
+  nlive: 40
+  Ebv-max: 0
+  injection: ${WORKING_DIR}/data/multi_config_analysis/injection.json
+  injection-num: 0
+  injection-outfile: ${WORKING_DIR}/outdir_1/lc.csv
+  generation-seed: 42
+  filters: ztfg,ztfr,ztfi
+  remove-nondetections: True
+
+Me2017_2:
+  model: Me2017
+  outdir: ${WORKING_DIR}/outdir_2
+  label: injection
+  prior: ${WORKING_DIR}/data/multi_config_analysis/Me2017.prior
+  tmin: 0.1
+  tmax: 5
+  dt: 0.5
+  error-budget: 1
+  nlive: 40
+  Ebv-max: 0
+  injection: ${WORKING_DIR}/data/multi_config_analysis/injection.json
+  injection-num: 0
+  injection-outfile: ${WORKING_DIR}/outdir_2/lc.csv
+  generation-seed: 42
+  filters: ztfg,ztfr,ztfi
+  remove-nondetections: True

nmma/tests/data/multi_config_analysis/injection.json

@@ -0,0 +1,31 @@
+{
+    "injections": {
+        "__dataframe__": true,
+        "content": {
+            "simulation_id": [
+                0
+            ],
+            "luminosity_distance": [
+                152.52372678074587
+            ],
+            "beta": [
+                3.1674793717549354
+            ],
+            "log10_kappa_r": [
+                1.9033736926068743
+            ],
+            "timeshift": [
+                -1.640734750742769
+            ],
+            "log10_vej": [
+                -1.0471069387482033
+            ],
+            "log10_mej": [
+                -2.9158814967880233
+            ],
+            "geocent_time": [
+                0.09662692256248354
+            ]
+        }
+    }
+}

nmma/tests/multi_config_analysis.py

@@ -0,0 +1,30 @@
+from argparse import Namespace
+import os
+import pytest
+import shutil
+
+from ..em import multi_config_analysis
+
+
+WORKING_DIR = os.path.dirname(__file__)
+
+
+
+@pytest.fixture(scope="module")
+def args():
+    WORKING_DIR = os.path.dirname(__file__)
+    os.environ["WORKING_DIR"] = WORKING_DIR
+    config = os.path.join(WORKING_DIR, "data/multi_config_analysis/config.yaml")
+
+    args = Namespace(config=config, process=2, parallel=True)
+
+    return args
+
+
+def test_analysis_multi(args):
+    multi_config_analysis.main(args)
+
+
+def test_analysis_parallel(args):
+    args.parallel = False
+    multi_config_analysis.main(args)

pyproject.toml

@@ -73,6 +73,7 @@ gwem-Hubble-estimate = "nmma.em.gwem_Hubble_estimate:main"
 maximum-mass-constraint = "nmma.joint.maximum_mass_constraint:main"
 convert-skyportal-lcs = "tools.convert_skyportal_lcs:main"
 resample-grid = "tools.resample_grid:main"
+multi-config-analysis = "nmma.em.multi_config_analysis:main"
 
 [project.urls]
 Homepage = "https://github.com/nuclear-multimessenger-astronomy/nmma"