Merge branch 'master' into nnpdfpol

validphys2/src/validphys/tests/conftest.py

@@ -35,8 +35,9 @@ def tmp(tmpdir):
 DATA = [
     {'dataset': 'NMC'},
     {'dataset': 'ATLASTTBARTOT', 'cfac': ['QCD']},
-    SINGLE_DATAPOINT,
     {'dataset': 'CMSZDIFF12', 'cfac': ('QCD', 'NRM'), 'sys': 10},
+    # Explicitly put a CMS dataset between the two ATLAS
+    SINGLE_DATAPOINT,
 ]
 
 
@@ -206,7 +207,7 @@ def pytest_runtest_setup(item):
     supported_platforms = ALL.intersection(mark.name for mark in item.iter_markers())
     plat = sys.platform
     if supported_platforms and plat not in supported_platforms:
-        pytest.skip("cannot run on platform {}".format(plat))
+        pytest.skip(f"cannot run on platform {plat}")
 
 
 def pytest_configure(config):

validphys2/src/validphys/tests/test_covmatreg.py

@@ -3,19 +3,22 @@
 
 functions to test the covariance matrix regularization
 """
+
 import numpy as np
 import pandas as pd
 
-from validphys.tests.test_regressions import make_table_comp
-from validphys.tableloader import parse_exp_mat
-from validphys.calcutils import regularize_covmat, regularize_l2
 from validphys.api import API
+from validphys.calcutils import regularize_covmat, regularize_l2
+from validphys.tableloader import parse_exp_mat
+from validphys.tests.test_regressions import make_table_comp
+
 
 def test_withidentity():
     identity_mat = np.eye(3)
     np.testing.assert_allclose(identity_mat, regularize_l2(identity_mat, 1))
     np.testing.assert_allclose(identity_mat, regularize_covmat(identity_mat, 1))
-    np.testing.assert_allclose(2*identity_mat, regularize_l2(identity_mat, 0.5))
+    np.testing.assert_allclose(2 * identity_mat, regularize_l2(identity_mat, 0.5))
+
 
 @make_table_comp(parse_exp_mat)
 def test_regularize_expcov(data_config):
@@ -27,35 +30,38 @@ def test_regularize_expcov(data_config):
     assert ~np.allclose(df1.values, df2.values)
     # check that square of sqrt matches
     sqrt_df1 = API.groups_sqrtcovmat(**inp)
-    np.testing.assert_allclose(df1.values, sqrt_df1.values@sqrt_df1.values.T)
+    np.testing.assert_allclose(df1.values, sqrt_df1.values @ sqrt_df1.values.T)
     # check that same result obtained
     return df1
 
+
 @make_table_comp(parse_exp_mat)
 def test_regularized_covmat_generation(data_config):
-    covmat = API.dataset_inputs_covariance_matrix(**data_config, norm_threshold=3)
+    covmat = API.groups_covmat(**data_config, norm_threshold=3)
     index = API.groups_index(**data_config)
     return pd.DataFrame(covmat, index=index, columns=index)
 
+
 def test_regularization_matches_sane():
     """Check that regularizing the sqrt cov produces same result as regularizing
     on the covariance matrix with a fairly sane matrix
     """
     # choose some sensible matrix to be regularized
-    a = np.ones((3, 3)) + 0.5*np.diag(np.ones(3))
-    cov = a@a.T
+    a = np.ones((3, 3)) + 0.5 * np.diag(np.ones(3))
+    cov = a @ a.T
     a_reg = regularize_l2(a, 3)
-    np.testing.assert_allclose(regularize_covmat(cov, 3), a_reg@a_reg.T)
+    np.testing.assert_allclose(regularize_covmat(cov, 3), a_reg @ a_reg.T)
+
 
 def test_regularization_matches_zero_eig():
     """Check that regularizing the sqrt cov produces the same result as regularizing
     on the covmat with a matrix with almost zero eigenvalue
     """
     # choose matrix with ~zero eigenvalue
     a = np.arange(9).reshape(3, 3)
-    cov = a@a.T
+    cov = a @ a.T
     a_reg = regularize_l2(a, 3)
-    np.testing.assert_allclose(regularize_covmat(cov, 3), a_reg@a_reg.T)
+    np.testing.assert_allclose(regularize_covmat(cov, 3), a_reg @ a_reg.T)
 
 
 @make_table_comp(parse_exp_mat)

validphys2/src/validphys/tests/test_covmats.py

@@ -3,16 +3,14 @@
 
 Tests related to the computation of the covariance matrix and its derivatives
 """
-import pytest
 
 import numpy as np
-
+import pytest
 
 from validphys.api import API
 from validphys.commondataparser import load_commondata
-from validphys.covmats import sqrt_covmat, dataset_t0_predictions
-from validphys.tests.conftest import THEORYID, PDF, HESSIAN_PDF, DATA
-
+from validphys.covmats import dataset_t0_predictions, reorder_thcovmat_as_expcovmat, sqrt_covmat
+from validphys.tests.conftest import DATA, HESSIAN_PDF, PDF, THEORYID
 
 # Experiments which have non trivial correlations between their datasets
 CORR_DATA = [
@@ -25,6 +23,15 @@
 ]
 
 
+def reorder_as_input(covmat_df, input_config):
+    """Reorder dataframe matrices as they come in the input"""
+    data = API.data(**input_config)
+    # Used to exploit reorder_thcovmat_as_expcovmat
+    fake_load = lambda: None
+    fake_load.load = lambda: covmat_df
+    return reorder_thcovmat_as_expcovmat(fake_load, data)
+
+
 def test_self_consistent_covmat_from_systematics(data_internal_cuts_config):
     """Test which checks that the single dataset implementation of
     ``covmat_from_systematics`` matches ``dataset_inputs_covmat_from_systematics``
@@ -35,10 +42,8 @@ def test_self_consistent_covmat_from_systematics(data_internal_cuts_config):
     dataset_inputs = base_config.pop("dataset_inputs")
 
     for dsinp in dataset_inputs:
-        covmat_a = API.covmat_from_systematics(
-            **base_config, dataset_input=dsinp)
-        covmat_b = API.dataset_inputs_covmat_from_systematics(
-            **base_config, dataset_inputs=[dsinp])
+        covmat_a = API.covmat_from_systematics(**base_config, dataset_input=dsinp)
+        covmat_b = API.dataset_inputs_covmat_from_systematics(**base_config, dataset_inputs=[dsinp])
         np.testing.assert_allclose(covmat_a, covmat_b)
 
 
@@ -55,10 +60,11 @@ def test_covmat_from_systematics(data_config, use_cuts, dataset_inputs):
     config["dataset_inputs"] = dataset_inputs
 
     covmat = API.dataset_inputs_covmat_from_systematics(**config)
-    another_covmat = API.groups_covmat(**config)
+    another_covmat = reorder_as_input(API.groups_covmat(**config), config)
 
     np.testing.assert_allclose(another_covmat, covmat)
 
+
 def test_covmat_with_one_systematic():
     """Test that a dataset with 1 systematic successfully builds covmat.
     This special case can break the covmat construction in python because of pandas indexing.
@@ -98,8 +104,7 @@ def test_sqrt_covmat(data_config):
 
 @pytest.mark.parametrize("t0pdfset", [PDF, HESSIAN_PDF])
 @pytest.mark.parametrize("dataset_inputs", [DATA, CORR_DATA])
-def test_python_t0_covmat_matches_variations(
-    data_internal_cuts_config, t0pdfset, dataset_inputs):
+def test_python_t0_covmat_matches_variations(data_internal_cuts_config, t0pdfset, dataset_inputs):
     """Test which checks the python computation of the t0 covmat relating to a
     collection of datasets
 
@@ -112,19 +117,16 @@ def test_python_t0_covmat_matches_variations(
     config["t0pdfset"] = t0pdfset
     config["use_t0"] = True
     covmat = API.dataset_inputs_t0_covmat_from_systematics(**config)
-    another_covmat = API.groups_covmat(**config)
+    another_covmat = reorder_as_input(API.groups_covmat(**config), config)
     # use allclose defaults or it fails
     np.testing.assert_allclose(another_covmat, covmat, rtol=1e-05, atol=1e-08)
     with pytest.raises(AssertionError):
-        np.testing.assert_allclose(
-            covmat, API.dataset_inputs_covmat_from_systematics(**config)
-        )
+        np.testing.assert_allclose(covmat, API.dataset_inputs_covmat_from_systematics(**config))
 
 
 @pytest.mark.parametrize("use_cuts", ["nocuts", "internal"])
 @pytest.mark.parametrize("dataset_input", DATA)
-def test_systematic_matrix(
-    data_config, use_cuts, dataset_input):
+def test_systematic_matrix(data_config, use_cuts, dataset_input):
     """Test which checks the python computation of the t0 covmat relating to a
     collection of datasets is equivalent using different functions
 

validphys2/src/validphys/tests/test_pythonmakereplica.py

@@ -4,6 +4,7 @@
 Module for testing the python implementation of make replica
 
 """
+
 from copy import deepcopy
 
 import numpy as np
@@ -14,10 +15,9 @@
 from validphys.tests.conftest import DATA
 from validphys.tests.test_covmats import CORR_DATA
 
-
 SEED = 123456
 
-#Datasets to be tested 
+# Datasets to be tested
 SINGLE_SYS_DATASETS = [
     {"dataset": "DYE886R"},
     {"dataset": "D0ZRAP", "cfac": ["QCD"]},
@@ -27,7 +27,7 @@
     {"dataset": "ATLASWZRAP36PB"},
     {"dataset": "ATLASZHIGHMASS49FB"},
     {"dataset": "CMSWEASY840PB"},
-    {"dataset": "CMSWMASY47FB"}
+    {"dataset": "CMSWMASY47FB"},
 ]
 
 
@@ -102,7 +102,12 @@ def test_genrep_off(data_config, dataset_inputs, use_cuts):
     config["use_cuts"] = use_cuts
     config["replica_mcseed"] = SEED
     config["genrep"] = False
-    ld_cds = API.dataset_inputs_loaded_cd_with_cuts(**config)
+
+    # `make_replica` reorders the data according to the groups in the exp covmat
     not_replica = API.make_replica(**config)
+
+    # And so it needs to be tested against the central values loaded by group
+    ld_cds = API.groups_dataset_inputs_loaded_cd_with_cuts(**config)
+
     central_data = np.concatenate([d.central_values for d in ld_cds])
     np.testing.assert_allclose(not_replica, central_data)