I don’t understand whether his model here refers to the code above or rffw.rffw?
My goal is to obtain the dimensions and probability distribution of vs and depth from the observed rf_matrix and amp data. My current method of building speed and depth models makes the convergence very slow. I hope to directly obtain the posterior distribution of velocity-depth by sampling directly from the velocity-depth space. Do you have any methods and suggestions? Can pymc.Mixture set the dimension (For example, the number of layers) of vs and depth for sampling via the broadcasting rules ? Thanks so much.