I think this can be accomplished with the pm.Potential method by adding a likelihood penalty when spots get very close to one another. This might be computationally expensive though, we'll need to implement it and test it. We can do something like:
where n = -2 or -3, for example. This will put a "repelling force" on the spot positions – the more negative the exponential index, the more sharply the repulsion increases with distance.
I think this can be accomplished with the
pm.Potentialmethod by adding a likelihood penalty when spots get very close to one another. This might be computationally expensive though, we'll need to implement it and test it. We can do something like:where n = -2 or -3, for example. This will put a "repelling force" on the spot positions – the more negative the exponential index, the more sharply the repulsion increases with distance.