fix bug in _hqp_yq_xq2map process

validphys2/src/validphys/process_options.py

@@ -159,11 +159,11 @@ def _hqp_yq_xq2map(kin_info):
     # Theory predictions computed with HT/4 ~ mt/2 for rapidity distr.
     # see section 3 from 1906.06535
     # HT defined in Eqn. (1) of 1611.08609
-    rapidity = kin_info.get_one_of(_Vars.y_t, _Vars.y_ttBar)
-    ratio = np.sqrt(mass2) / kin_info[_Vars.sqrts]
+    rapidity = kin_info.get_one_of(_Vars.y_t, _Vars.y_ttBar, "k1")
+    q2 = kin_info.get_one_of(_Vars.m_t2, "k2")
+    ratio = np.sqrt(q2) / kin_info[_Vars.sqrts]
     x1 = ratio * np.exp(rapidity)
     x2 = ratio * np.exp(-rapidity)
-    q2 = kin_info[_Vars.m_t2]
     x = np.concatenate((x1, x2))
     return np.clip(x, a_min=None, a_max=1, out=x), np.concatenate((q2, q2)) / 4
 

validphys2/src/validphys/tests/test_datafiles.py

@@ -1,13 +1,14 @@
 """
     Test all datafiles
 
-    The checks in ``test_all_datasets`` are run for each dataset independently so that one gets one 
+    The checks in ``test_all_datasets`` are run for each dataset independently so that one gets one
     failure per dataset in case of problems
 """
 
 import pytest
 
 from validphys.covmats import INTRA_DATASET_SYS_NAME
+from validphys.kinematics import xq2map_with_cuts
 from validphys.loader import FallbackLoader
 from validphys.plotoptions.kintransforms import identity as kintransform_identity
 
@@ -24,9 +25,21 @@ def _load_main_and_variants(dataset_name):
 
 
 @pytest.mark.parametrize("dataset_name", all_datasets)
-def test_all_datasets(dataset_name):
-    """checks that a dataset can be loaded (together with its variants),
-    that the kinematics, uncertainties and data can be read
+def test_all_datasets(dataset_name, data_internal_cuts_new_theory_config):
+    """Checks that a dataset can be loaded (together with its variants),
+    that the kinematics, uncertainties and data can be read.
+
+    All checks pertaining to a given dataset are done together in this function
+    since a broken dataset will likely fail more than one check.
+    This avoids polluting the output with many errors for a single dataset.
+
+    Checks:
+        1. Loading of data, kinematics, uncertainties
+        2. Kinematic coverage is included in the dataframe
+        3. A process type is being used (if not legacy)
+        4. All variants can be loaded
+        5. Uncertainties are either ADD or MULT
+        6. The kinematic coverage coverage can be generated
     """
     # Load the data and all its variants
     cds = _load_main_and_variants(dataset_name)
@@ -65,11 +78,15 @@ def test_all_datasets(dataset_name):
     # check the uncertainties can be loaded
     # note that due to legacy data there might be datasets without data_uncertainties
     # but that would only happen for non-variant (or member 0 of the list)
-    all_unc = [cd.metadata.load_uncertainties() for cd in cds[1:]]
+    # Separate member 0 in that case
+    valid_cds = []
     if main_cd.metadata.data_uncertainties:
-        all_unc.insert(0, main_cd.metadata.load_uncertainties())
+        valid_cds.append(main_cd)
+    valid_cds += cds[1:]
+
+    for cd in valid_cds:
+        unc = cd.metadata.load_uncertainties()
 
-    for unc in all_unc:
         # Check that, if present, the special `stat` key is ADD and UNCORR
         if "stat" in unc:
             stat = unc["stat"]
@@ -90,3 +107,12 @@ def test_all_datasets(dataset_name):
 
         # Check that all treatments are either MULT or ADD
         assert set(unc.columns.get_level_values("treatment").unique()) <= {"MULT", "ADD"}
+
+    # Extra checks for non-polarized datasets
+    if str(process_type).endswith("_POL"):
+        return
+
+    # Legacy datasets with no "new implementation" are skipped
+    for cd in valid_cds:
+        # check without cuts
+        xq2map_with_cuts(cd, False)