I have a forward simulation function forward_model that gives some parameters of type ndarray input and returns a result of type ndarray. I want to estimate the parameter epsilon_r given the synthetic data. Below is my script, but I don’t know how to include the function in MCMC model part. The prob…

So ideally you would rewrite it with PyTensor operators so that you can get gradients for free and use NUTS for sampling (and possibly higher performance). Mostly you can replace np.foo(a, b) by pytensor.tensor.foo(a, b), with special handling for conditionals/ loops and variable assignment: Tutori…

Inversion of parameter given forward simulation function

Questions version agnostic

ricardoV94 September 26, 2024, 12:18pm 8

You don’t need a CustomDist, just an Op to wrap around forward_model, so your case is just a subset of that blackbox likelihood tutorial

Topic		Replies	Views
Using pm.DensityDist and customized likelihood with a black-box function Questions	10	6853	August 29, 2018
Getting observations of distribution variables whilst sampling Questions from_github	2	1008	June 15, 2017
Problem model definition PyMC3 Questions	12	4749	September 21, 2017
How to set up a custom likelihood function for two variables Questions	8	13306	March 9, 2018
Problems with defining a custom likelihood that calls a Python wrapper function Questions	0	594	January 25, 2020

Inversion of parameter given forward simulation function

Related topics