Pymc3 capability for inference on every cell of a big grid?

I am newbie to pymc3 and sorry if my question is elementary. The problem is doing Bayesian inference for every cell of a 2D grid (e.g. 1000 * 800). Each cell has 4 target parameters which one of them is discrete and the others are continuous and conditional on the categorial parameter. It is desired to do Bayesian inference by MCMC columnwise ( solving inverse problem for each cell of a column and using the results as prior for other columns). Is it possible to solve this problem with pymc3? any codes to show me how to implement such a high dimensional problem?