hello everyone, I am a beginner at Pymc3. I built a nonlinear regression model y=a(x-b)^c to fit the observed data (x,y). This model assumes that the observation of x is equal to the true value while the observation of y is not. I have constructed prior distribution of 5 parameters and the likeli…

jitter+adapt_diag makes a jitter on the starting value, so that different chain started at a different point in the parameter space. It is the recommended option, as it helps you discover bad parameterization. For example, in your case there are indication that some parameter space is difficult to s…

Difference between 'jitter+adapt_diag' and 'adapt_diag'?

Questions

Jiangchao1 April 2, 2019, 8:47am 3

thanks junpeng very much! I solved the problem using a HalfNormal distribution to describe the prior of my error parameters. the iteration speed is also good, thanks very much!

Topic		Replies	Views
Initialization energy is NaN or Inf with jitter Questions	4	1338	December 9, 2020
What exactly is `jitter+adapt_diag` and why is it the default now? Questions	3	3497	November 3, 2017
Derivatives are zero for jitter but not ADVI initialization Questions	2	556	September 28, 2018
Sampler occasionally gets static in pymc3.1, but not in pymc3.0 Questions development	19	2271	July 5, 2018
Converting Model from Stan Questions	15	3176	June 27, 2017

Difference between 'jitter+adapt_diag' and 'adapt_diag'?

Related topics