GaussianRandomWalk with varying stepsizes?

Hi all,

For a problem I am currently working on, I am trying to define a GaussianRandomWalk that in its initial phase is quite limited in stepsizes for the initial n steps, and then has a larger stepsize in subsequent steps.

I tried to do this by setting sigma as a numpy array with length GRW steps, but it looks like the GRW prior just uses the first stepsize value in that np array as the general stepsize for the GRW.

Is this possible in some way?

I’m pretty sure your approach should work - can you post a minimal code sample?