I am trying to compute the complexity of my model using rStan. I have two nested integrals to compute which are \int_{m} P(m|x) (\int_{x} P(y|m,x) P(x)) dm. There are 2 for loops in my Stan model that are related to model complexity. In addition to that, Stan computes the parameters where they have…

[image] srtaheri: 1 parameter to learn or 20 parameters to learn, the complexity would be the same It means that doing leapfrog steps as part of HMC require some O(P) math. I’m assuming the gradient evaluations (defined by your model) are at least O(P), which would make them more important. O…

Computational complexity of sampling method in Stan

General

jonah August 26, 2020, 1:41am 12

In case it’s helpful, here’s a recent post that’s on a similar topic:

2 Likes

Topic		Replies	Views
Asymptotic computational complexity of HMC - for a simple GMM tutorial - is it - in the first approximation - O(P^3*N) ? where : P- number of parameters, N - number of samples Modeling performance	16	1591	January 13, 2019
Rstan much slower than self-coded MCMC Modeling performance	10	2506	January 14, 2022
Computationally expensive three component mixture model, suggestions for speed-up? Modeling	10	1323	August 22, 2017
Estimate time to run HMC Algorithms fitting-issues	5	57	February 19, 2026
Question about the HMC algorithm in rstan General rstan	14	1025	January 19, 2021

Computational complexity of sampling method in Stan

Related topics