Harmonize scripts, command line options, common parsers

.github/workflows/main.yml

@@ -177,7 +177,7 @@ jobs:
           python tests/make_tensor.py -o $RABBIT_OUTDIR/ --histType boost --postfix boost
 
       - name: debug input
-        run: debug_inputdata.py $RABBIT_OUTDIR//test_tensor.hdf5
+        run: rabbit_debug_inputdata.py $RABBIT_OUTDIR//test_tensor.hdf5
 
       - name: plot input
         run: >-
@@ -241,11 +241,9 @@ jobs:
           rabbit_fit.py $RABBIT_OUTDIR/test_tensor.hdf5 -o $RABBIT_OUTDIR/ --postfix pseudodata
           -t 0 --pseudoData original --doImpacts --globalImpacts --unblind ^sig$
 
-
-  alternative-fits:
+  sparse-fits:
     runs-on: [self-hosted, linux, x64]
     needs: [setenv, make-tensor]
-
     steps:
       - env:
           RABBIT_OUTDIR: ${{ needs.setenv.outputs.RABBIT_OUTDIR }}
@@ -265,32 +263,96 @@ jobs:
           --saveHists --saveHistsPerProcess --computeHistErrors --computeHistErrorsPerProcess
           -m Project ch1 a -m Project ch1 b
 
-      - name: covariance fit
-        run: >- 
-          rabbit_fit.py $RABBIT_OUTDIR/test_tensor.hdf5 -o $RABBIT_OUTDIR/ --postfix covariance
-          -t 0 --unblind --covarianceFit --doImpacts --globalImpacts 
-          --saveHists --saveHistsPerProcess --computeHistErrors --computeHistErrorsPerProcess
-          -m Project ch1 a -m Project ch1 b
+  alternative-fits:
+    runs-on: [self-hosted, linux, x64]
+    needs: [setenv, make-tensor]
+    strategy:
+      fail-fast: false
+      matrix:
+        include:              
+          - name: covariance
+            tensor: test_tensor.hdf5
+            fit_args: >-
+              --covarianceFit
+          - name: linearized
+            tensor: test_tensor_symmetric.hdf5
+            fit_args: >-
+              --binByBinStatType normal-additive --allowNegativePOI
+          - name: bb_full_normal_additive
+            tensor: test_tensor.hdf5
+            fit_args: >-
+              --binByBinStatType normal-additive --binByBinStatMode full
+          - name: bb_full_normal_multiplicative
+            tensor: test_tensor.hdf5
+            fit_args: >-
+              --binByBinStatType normal-multiplicative --binByBinStatMode full
+          - name: bb_full_gamma
+            tensor: test_tensor.hdf5
+            fit_args: >-
+              --binByBinStatType gamma --binByBinStatMode full --globalImpactsDisableJVP
 
-      - name: linearized solution
-        run: >- 
-          rabbit_fit.py $RABBIT_OUTDIR/test_tensor_symmetric.hdf5 -o $RABBIT_OUTDIR/ --postfix linearized
-          -t 0 --chisqFit --doImpacts --globalImpacts --binByBinStatType normal-additive --allowNegativePOI
-          --saveHists --saveHistsPerProcess --computeHistErrors --computeHistErrorsPerProcess
-          --computeHistImpacts --computeHistCov -m Project ch1 a -m Project ch1 b 
+    steps:
+      - env:
+          RABBIT_OUTDIR: ${{ needs.setenv.outputs.RABBIT_OUTDIR }}
+          PYTHONPATH: ${{ needs.setenv.outputs.PYTHONPATH }}
+          PATH: ${{ needs.setenv.outputs.PATH }}
+          WEB_DIR: ${{ needs.setenv.outputs.WEB_DIR }}
+          PLOT_DIR: ${{ needs.setenv.outputs.PLOT_DIR }}
+        run: |
+          echo "RABBIT_OUTDIR=${RABBIT_OUTDIR}" >> $GITHUB_ENV
+          echo "PYTHONPATH=${PYTHONPATH}" >> $GITHUB_ENV
+          echo "PATH=${PATH}" >> $GITHUB_ENV
+          echo "WEB_DIR=${WEB_DIR}" >> $GITHUB_ENV
+          echo "PLOT_DIR=${PLOT_DIR}" >> $GITHUB_ENV
 
-      - name: Barlow--Beeston full (gamma)
-        run: >- 
-          rabbit_fit.py $RABBIT_OUTDIR/test_tensor_symmetric.hdf5 -o $RABBIT_OUTDIR/ --postfix nonlinear
-          -t 0 --doImpacts --globalImpacts --globalImpactsDisableJVP --binByBinStatType gamma --binByBinStatMode full
-          --saveHists --saveHistsPerProcess --computeHistErrors --computeHistErrorsPerProcess
-          --computeHistImpacts --computeHistCov -m Project ch1 a -m Project ch1 b 
+      - uses: actions/checkout@v5
 
-      - name: chi2 fit with Barlow--Beeston full (normal-multiplicative)
-        run: >- 
-          rabbit_fit.py $RABBIT_OUTDIR/test_tensor_symmetric.hdf5 -o $RABBIT_OUTDIR/ --postfix bb_full
-          -t 0 --globalImpacts --binByBinStatType normal-multiplicative --binByBinStatMode full
+      - name: fit ${{ matrix.name }}
+        run: >-
+          rabbit_fit.py $RABBIT_OUTDIR/${{ matrix.tensor }} -o $RABBIT_OUTDIR/ --postfix ${{ matrix.name }}
+          -t -1 0 --doImpacts --globalImpacts --gaussianGlobalImpacts ${{ matrix.fit_args }}
+          -m Project ch1 a -m Project ch1 b --saveHists --saveHistsPerProcess --computeHistErrors 
+          --computeHistErrorsPerProcess --computeHistImpacts --computeHistGaussianImpacts --computeHistCov
+
+      - name: print impacts ${{ matrix.name }}
+        run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 -u
+
+      - name: print gaussian global impacts ${{ matrix.name }}
+        run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 --impactType gaussian_global
 
+      - name: print global impacts ${{ matrix.name }}
+        run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 -s --impactType global
+
+      - name: plot impacts ${{ matrix.name }}
+        run: >-
+          rabbit_plot_pulls_and_impacts.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 -o $WEB_DIR/$PLOT_DIR/${{ matrix.name }}
+          --otherExtensions pdf png -n 50 --config tests/style_config.py -s absimpact --subtitle "Work in progress"
+          --grouping max --postfix ${{ matrix.name }}
+
+      - name: plot gaussian global impacts ${{ matrix.name }}
+        run: >-
+          rabbit_plot_pulls_and_impacts.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 -o $WEB_DIR/$PLOT_DIR/${{ matrix.name }}
+          --otherExtensions pdf png -n 50 --config tests/style_config.py -s absimpact --subtitle Preliminary
+          --grouping max --postfix ${{ matrix.name }} --impactType gaussian_global
+          --refResult asimov --refName "asimov" --name "data"
+
+      - name: plot global impacts ${{ matrix.name }}
+        run: >-
+          rabbit_plot_pulls_and_impacts.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 -o $WEB_DIR/$PLOT_DIR/${{ matrix.name }}
+          --otherExtensions pdf png -n 50 --config tests/style_config.py -s absimpact
+          --grouping max --postfix ${{ matrix.name }} --impactType global --oneSidedImpacts
+
+      - name: plot postfit distributions ${{ matrix.name }}
+        run: >-
+          rabbit_plot_hists.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 -o $WEB_DIR/$PLOT_DIR/${{ matrix.name }}
+          --extraTextLoc '0.05' '0.7' --legCols 1 -m Project ch1 a -m Project ch1 b --yscale '1.2'
+          --subtitle "Work in progress" --config tests/style_config.py
+
+      - name: plot histogram uncertainties ${{ matrix.name }}
+        run: >-
+          rabbit_plot_hists_uncertainties.py $RABBIT_OUTDIR/fitresults_${{ matrix.name }}.hdf5 -o $WEB_DIR/$PLOT_DIR/${{ matrix.name }}
+          --extraTextLoc '0.05' '0.7' --legCols 1 -m Project ch1 a -m Project ch1 b --yscale '1.2'
+          --subtitle "Work in progress" --config tests/style_config.py
 
   bsm:
     runs-on: [self-hosted, linux, x64]
@@ -348,10 +410,10 @@ jobs:
         run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults.hdf5 -s
 
       - name: print global impacts
-        run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults.hdf5 -s --globalImpacts
+        run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults.hdf5 -s --impactType global
 
       - name: print global impacts of observable
-        run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults.hdf5 --globalImpacts -m Project ch0_masked --ungroup --sort
+        run: rabbit_print_impacts.py $RABBIT_OUTDIR/fitresults.hdf5 --impactType global -m Project ch0_masked --ungroup --sort
 
       - name: plot impacts
         run: >-
@@ -362,14 +424,14 @@ jobs:
         run: >-
           rabbit_plot_pulls_and_impacts.py $RABBIT_OUTDIR/fitresults.hdf5 -o $WEB_DIR/$PLOT_DIR
           -r $RABBIT_OUTDIR/fitresults_pseudodata.hdf5 --refResult original --refName "Pseudo data" -s absimpact
-          --otherExtensions pdf png -n 50  --config tests/style_config.py --oneSidedImpacts --globalImpacts --diffPullAsym --showNumbers
+          --otherExtensions pdf png -n 50  --config tests/style_config.py --oneSidedImpacts --impactType global --diffPullAsym --showNumbers
           --grouping max --subtitle Preliminary --postfix data_vs_pseudodata
 
       - name: plot global impacts on observable
         run: >-
           rabbit_plot_pulls_and_impacts.py $RABBIT_OUTDIR/fitresults.hdf5 -o $WEB_DIR/$PLOT_DIR
           -s absimpact -m Project ch0_masked
-          --otherExtensions pdf png -n 50  --config tests/style_config.py --oneSidedImpacts --globalImpacts --showNumbers
+          --otherExtensions pdf png -n 50  --config tests/style_config.py --oneSidedImpacts --impactType global --showNumbers
           --grouping max --subtitle Preliminary --postfix ch0_masked
 
       - name: plot prefit distributions
@@ -411,7 +473,7 @@ jobs:
 
       - name: plot histogram covariance matrices
         run: >-
-          rabbit_plot_hists_cov.py $RABBIT_OUTDIR/fitresults.hdf5 -o $WEB_DIR/$PLOT_DIR 
+          rabbit_plot_cov.py $RABBIT_OUTDIR/fitresults.hdf5 -o $WEB_DIR/$PLOT_DIR 
           --subtitle Preliminary -m Project ch1 a -m Project ch1 b --subtitle Preliminary --config tests/style_config.py
 
 
@@ -456,7 +518,7 @@ jobs:
 
   copy-clean:
     runs-on: [self-hosted, linux, x64]
-    needs: [setenv, symmerizations, alternative-fits, bsm, plotting, likelihoodscans]
+    needs: [setenv, symmerizations, sparse-fits, alternative-fits, bsm, plotting, likelihoodscans]
     if: always() 
     steps:
       - env:

README.md

@@ -60,7 +60,7 @@ source env/bin/activate
 pip install wums[pickling,plotting] tensorflow tensorflow-probability tf_keras numpy h5py hist scipy matplotlib mplhep seaborn pandas plotly kaleido
 ```
 The packages `matplotlib`, `mplhep`, `seaborn`, `pandas`, `plotly`, and `kaleido` are only needed for the plotting scripts. 
-For the `text2hdf5.py` conversion also the `uproot` package is needed.
+For the `rabbit_text2hdf5.py` conversion also the `uproot` package is needed.
 In case you want to contribute to the development, please also install the linters `isort`, `flake8`, `autoflake`, `black`, and `pylint` used in the pre-commit hooks and the github CI
 Deactivate the environment with `deactivate`.
 
@@ -110,15 +110,15 @@ This is done by defining `masked=True` in the `tensorwriter` `add_channel` funct
 This is useful for example to compute unfolded (differential) cross sections and their uncertainties, including global impacts, taking into account all nuisance parameters that affect these channels.
 
 ### text2hdf5
-The input tensor can also be generated from the input used for the [Combine tool](https://link.springer.com/article/10.1007/s41781-024-00121-4)  using the `text2hdf5.py` command.
+The input tensor can also be generated from the input used for the [Combine tool](https://link.springer.com/article/10.1007/s41781-024-00121-4)  using the `rabbit_text2hdf5.py` command.
 This script is mainly intended for users that have these inputs already and want to perform some cross checks.
 Only basic functionality is supported and for complex models the conversion can take long, it is thus recommended to directly produce the input tensor using the provided interface as explained above. 
 
 ### Diagnostics
 Scripts for diagnosing the input tensor are available:
 Running some checks for empty bins etc.
 ```bash
-debug_inputdata.py test_tensor.hdf5
+rabbit_debug_inputdata.py test_tensor.hdf5
 ```
 Plotting the histograms that are actually used in the fit, supporting adding of systematic variations in the plot:
 ```bash

bin/debug_inputdata.py → bin/rabbit_debug_inputdata.py

@@ -1,15 +1,26 @@
 #!/usr/bin/env python
-import argparse
-
 import hist
 import numpy as np
 from wums import logging
 
-from rabbit import debugdata, inputdata
+from rabbit import debugdata, inputdata, parsing
 
 logger = None
 
 
+def make_parser():
+    parser = parsing.print_parser()
+    parser.description = "Debug input data for fitting"
+    parser.add_argument(
+        "--channels",
+        type=str,
+        help="Only check specified channels",
+        default=None,
+        nargs="+",
+    )
+    return parser
+
+
 def debug_input_data(input_file, output_dir=None, verbose=False, channels=None):
     """
     Debug input data file and report potential issues
@@ -254,32 +265,12 @@ def debug_input_data(input_file, output_dir=None, verbose=False, channels=None):
 
 
 def main():
-    parser = argparse.ArgumentParser(description="Debug input data for fitting")
-    parser.add_argument("inputFile", help="Path to input data file")
-    parser.add_argument(
-        "-v",
-        "--verbose",
-        type=int,
-        default=3,
-        choices=[0, 1, 2, 3, 4],
-        help="Set verbosity level with logging, the larger the more verbose",
-    )
-    parser.add_argument(
-        "--noColorLogger", action="store_true", help="Do not use logging with colors"
-    )
-    parser.add_argument(
-        "--channels",
-        type=str,
-        help="Only check specified channels",
-        default=None,
-        nargs="+",
-    )
-    args = parser.parse_args()
+    args = make_parser().parse_args()
 
     global logger
     logger = logging.setup_logger(__file__, args.verbose, args.noColorLogger)
 
-    debug_input_data(args.inputFile, args.verbose, channels=args.channels)
+    debug_input_data(args.infile, args.verbose, channels=args.channels)
 
 
 if __name__ == "__main__":