Implementing smelli as an external likelihood

src/smefit/external_chi2_smelli.py

@@ -0,0 +1,50 @@
+from smefit.rge import RGE
+
+
+class smelli_chi2:
+    # coefficients are mandatory, taken from smefit runcard. rge_dict is the rge dictionary in the runcard,
+    # in this case we will only need it to read the initial scale, but we also need to build the RGE object.
+    # everything else is optional and can be specified in the external_chi2 runcard
+    def __init__(self, coefficients, rge_dict=None, dataset=[]):
+        # this gives you the scale
+        self.scale = rge_dict["init_scale"]
+
+        # this gives you the names of the coeff in smefit
+        self.names = coefficients.name
+
+        # construct the RGE object. We don't actually need the RGE from smefit, but
+        # it has inside of it a method to translate coefficients from smefit to wcxf.
+        # Maybe not super ideal that it's inside that but ok for now.
+        self.rge_obj = RGE(
+            self.names,
+            self.scale,
+            rge_dict["smeft_accuracy"],
+            rge_dict["adm_QCD"],
+            rge_dict["yukawa"],
+            # the last three are not actually needed
+        )
+
+        # This is a dictionary of dictionaries, associating to each of the smefit wc names, the wc_dict for smelli
+        self.translation_basis = self.rge_obj.RGEbasis
+
+    def compute_chi2(self, coefficient_values):
+        # this receives an array of numbers, which order corresponds
+        # to self.names
+
+        # you need to take those values and construct the wc_dict for smelli
+        # you have already the basis, which is telling you stuff like
+        # self.translation_basis["OpD"] -> {"phiD": 1e-6}
+        # so you need to take them all, multiply their values for the coefficien_values
+        # then create a single dictionary (possible repeated keys should be summed)
+        # something along these lines (adapted to the situation, this is copied from rge.RGEevolve)
+        wc_dict = {}
+        for op, values in self.RGEbasis.items():
+            for key in values:
+                if key not in wc_wilson:
+                    wc_dict[key] = values[key] * wcs[op]
+                else:
+                    wc_dict[key] += values[key] * wcs[op]
+
+        # now you should have the self.scale and the wc_dict needed for the smelli likelihood.
+        # make sure you are returning a "chi2"
+        return -2 smelli_loglikelihood