update the povm.convert(...) free function to work with CPTPLND parameterizations on non-qubit spaces (for example, qutrit spaces) by changing the default error generator basis.

Author: rileyjmurray

pygsti/modelmembers/povms/__init__.py

@@ -547,10 +547,14 @@ def convert(povm, to_type, basis, ideal_povm=None, flatten_structure=False, cp_p
                 #It is often the case that there are more error generators than physical degrees of freedom in the POVM
                 #We define a function which finds linear comb. of errgens that span these degrees of freedom.
                 #This has been called "the trivial gauge", and this function is meant to avoid it
+                eegb = 'PP' if basis.name == 'pp' else basis
                 def calc_physical_subspace(dense_ideal_povm, epsilon = 1e-4):
 
                     degrees_of_freedom = (dense_ideal_povm.shape[0] - 1) * dense_ideal_povm.shape[1]
-                    errgen = _LindbladErrorgen.from_error_generator(povm.state_space.dim, parameterization=to_type)
+                    errgen = _LindbladErrorgen.from_error_generator(
+                        povm.state_space.dim, parameterization=to_type,
+                        elementary_errorgen_basis=eegb, mx_basis=basis
+                    )
 
                     if degrees_of_freedom > errgen.num_params:
                         warnings.warn("POVM has more degrees of freedom than the available number of parameters, representation in this parameterization is not guaranteed")