I am attempting to implement bayesian model calibration under the classical Kennedy-O’Hagan framework using PyMC3. Briefly, the setup is as follows: I have n observations z_i of the form z_i = f(x_i, \theta) + g(x_i) + e_i, where x_i are known imputs and \theta are unknown calibration parameters …

Yep, the key is to stack the theano scaler with a numpy array. For example another way to do is import theano import theano.tensor as tt x1shared = theano.shared(x1[:, None]) with pm.Model(): ... input_1 = tt.concatenate([x1shared, tt.tile(theta, len(x1), ndim=2).T], axis=1) ...

Bayesian model calibration with Gaussian Process

Questions

junpenglao March 14, 2018, 6:18am 5

You can run some simulation to see if it can recover the parameters. But it sounds like a valid use case to me.

Topic		Replies	Views
Finding posterior for calibration using saved Gaussian model in pymc3 Questions theano	2	1352	November 15, 2017
Multidimensional gaussian process Questions	55	10241	October 22, 2020
Computer model calibration using a Gaussian Process: From stan to pymc v5 gaussian_process	10	725	June 22, 2023
How to calibrate the GP model with various data? Questions	2	798	September 22, 2018
Bayesian calibration on GP model v3 gaussian_process	7	1146	July 23, 2022

Bayesian model calibration with Gaussian Process

Related topics