Use results to create th uncertainties with pdf errors.

validphys2/src/validphys/covmats.py

@@ -1,29 +1,25 @@
 """Module for handling logic and manipulation of covariance and correlation
 matrices on different levels of abstraction
 """
-import logging
+
 import functools
+import logging
 
 import numpy as np
 import pandas as pd
 import scipy.linalg as la
 
 from reportengine import collect
 from reportengine.table import table
-from validphys.calcutils import get_df_block, regularize_covmat
+from validphys.calcutils import regularize_covmat
 from validphys.checks import (
     check_cuts_considered,
-    check_data_cuts_match_theorycovmat,
-    check_dataset_cuts_match_theorycovmat,
     check_norm_threshold,
     check_pdf_is_montecarlo_or_hessian,
     check_speclabels_different,
 )
-from validphys.commondata import loaded_commondata_with_cuts
 from validphys.convolution import central_predictions
-from validphys.core import PDF, DataGroupSpec, DataSetSpec
 from validphys.covmats_utils import construct_covmat, systematics_matrix
-from validphys.results import ThPredictionsResult
 
 log = logging.getLogger(__name__)
 
@@ -670,7 +666,7 @@ def groups_corrmat(groups_covmat):
 
 
 @check_pdf_is_montecarlo_or_hessian
-def pdferr_plus_covmat(dataset, pdf, covmat_t0_considered):
+def pdferr_plus_covmat(results_without_covmat, pdf, covmat_t0_considered):
     """For a given `dataset`, returns the sum of the covariance matrix given by
     `covmat_t0_considered` and the PDF error:
     - If the PDF error_type is 'replicas', a covariance matrix is estimated from
@@ -701,19 +697,22 @@ def pdferr_plus_covmat(dataset, pdf, covmat_t0_considered):
     `use_pdferr` makes this action be used for `covariance_matrix`
 
     >>> from validphys.api import API
-    >>> from import numpy as np
+    >>> import numpy as np
     >>> inp = {
-            'dataset_input': {'dataset' : 'ATLASTTBARTOT'},
-            'theoryid': 53,
-            'pdf': 'NNPDF31_nlo_as_0118',
-            'use_cuts': 'nocuts'
+            'dataset_input': {
+                'dataset': 'ATLAS_TTBAR_8TEV_LJ_DIF_YTTBAR-NORM',
+                'variant': 'legacy',
+            },
+            'theoryid': 700,
+            'pdf': 'NNPDF40_nlo_as_01180',
+            'use_cuts': 'internal',
         }
     >>> a = API.covariance_matrix(**inp, use_pdferr=True)
     >>> b = API.pdferr_plus_covmat(**inp)
-    >>> np.allclose(a == b)
+    >>> (a == b).all()
     True
     """
-    th = ThPredictionsResult.from_convolution(pdf, dataset)
+    _, th = results_without_covmat
 
     if pdf.error_type == 'replicas':
         pdf_cov = np.cov(th.error_members, rowvar=True)
@@ -754,18 +753,26 @@ def reorder_thcovmat_as_expcovmat(fitthcovmat, data):
     return tmp.reindex(index=bb, columns=bb, level=0)
 
 
-def pdferr_plus_dataset_inputs_covmat(data, pdf, dataset_inputs_covmat_t0_considered, fitthcovmat):
+def pdferr_plus_dataset_inputs_covmat(
+    dataset_inputs_results_without_covmat,
+    data,
+    pdf,
+    dataset_inputs_covmat_t0_considered,
+    fitthcovmat,
+):
     """Like `pdferr_plus_covmat` except for an experiment"""
     # do checks get performed here?
     if fitthcovmat is not None:
         # change ordering according to exp_covmat (so according to runcard order)
         return pdferr_plus_covmat(
-            data,
+            dataset_inputs_results_without_covmat,
             pdf,
             dataset_inputs_covmat_t0_considered
             + reorder_thcovmat_as_expcovmat(fitthcovmat, data).values,
         )
-    return pdferr_plus_covmat(data, pdf, dataset_inputs_covmat_t0_considered)
+    return pdferr_plus_covmat(
+        dataset_inputs_results_without_covmat, pdf, dataset_inputs_covmat_t0_considered
+    )
 
 
 def dataset_inputs_sqrt_covmat(dataset_inputs_covariance_matrix):

validphys2/src/validphys/results.py

@@ -3,6 +3,7 @@
 
 Tools to obtain theory predictions and basic statistical estimators.
 """
+
 from __future__ import generator_stop
 
 from collections import OrderedDict, namedtuple
@@ -536,7 +537,7 @@ def procs_corrmat(procs_covmat):
     return groups_corrmat(procs_covmat)
 
 
-def results(dataset: (DataSetSpec), pdf: PDF, covariance_matrix, sqrt_covmat):
+def results(dataset: DataSetSpec, pdf: PDF, covariance_matrix, sqrt_covmat):
     """Tuple of data and theory results for a single pdf. The data will have an associated
     covariance matrix, which can include a contribution from the theory covariance matrix which
     is constructed from scale variation. The inclusion of this covariance matrix by default is used
@@ -553,14 +554,26 @@ def results(dataset: (DataSetSpec), pdf: PDF, covariance_matrix, sqrt_covmat):
     )
 
 
-def results_central(dataset: (DataSetSpec), pdf: PDF, covariance_matrix, sqrt_covmat):
+def results_central(dataset: DataSetSpec, pdf: PDF, covariance_matrix, sqrt_covmat):
     """Same as :py:func:`results` but only calculates the prediction for replica0."""
     return (
         DataResult(dataset, covariance_matrix, sqrt_covmat),
         ThPredictionsResult.from_convolution(pdf, dataset, central_only=True),
     )
 
 
+def results_without_covmat(dataset: DataSetSpec, pdf: PDF):
+    """Return a results object with a diagonal covmat so that it can be used to generate
+    results-depending covmats elsewhere. Uses :py:funct:`results` under the hook"""
+    loaded_cd = dataset.load_commondata()
+    if isinstance(loaded_cd, list):
+        ndata = np.sum([cd.ndata for cd in loaded_cd])
+    else:
+        ndata = loaded_cd.ndata
+    diag = np.eye(ndata)
+    return results(dataset, pdf, diag, diag)
+
+
 def results_with_theory_covmat(dataset, results, theory_covmat_dataset):
     """Returns results with a modfy ``DataResult`` such that the covariance matrix includes
     also the theory covmat.
@@ -602,6 +615,11 @@ def results_with_scale_variations(results, theory_covmat_dataset):
     return (data_result, theory_error_result)
 
 
+def dataset_inputs_results_without_covmat(data, pdf: PDF):
+    """Like `dataset_inputs_results` but skipping the computation of the covmat"""
+    return results_without_covmat(data, pdf)
+
+
 def dataset_inputs_results_central(
     data, pdf: PDF, dataset_inputs_covariance_matrix, dataset_inputs_sqrt_covmat
 ):