Mixture models have been broken since 3020dbc with no errors triggered by the CI.
There are some tests for parsing composite models here, but no tests that they actually compute the correct result.
There are a couple of models that test structure factors (cylinder, sphere), but none to test, e.g., cylinder+sphere, sphere*hardsphere+cyinder.
Regarding the bug in mixture models, it is probably an offset issue when indexing into the parameter vector sent to the individual models. The changes to the magnetic model added an additional parameter into the parameter block.
Another broken model form is multiplicity*structure + form such as core_multi_shell*hardsphere + power_law. This breaks with the error "scale parameter duplicated".
Mixture models have been broken since 3020dbc with no errors triggered by the CI.
There are some tests for parsing composite models here, but no tests that they actually compute the correct result.
There are a couple of models that test structure factors (cylinder, sphere), but none to test, e.g.,
cylinder+sphere
,sphere*hardsphere+cyinder
.Regarding the bug in mixture models, it is probably an offset issue when indexing into the parameter vector sent to the individual models. The changes to the magnetic model added an additional parameter into the parameter block.