I have been looking for solutions to my model because didn’t give good convergence. When I was searching for information I saw some people using
start = pm.find_MAP()
train_trace = pm.sample( train_samp, tune= train_tune, start=start, njobs=4 )
and on the documentation it states
map : Use the MAP as starting point. This is discouraged.
My questions are
- Is pm.sample( …,start=start, …) equivalent to pm.sample( …,init= ‘map’, …)?
- Why is it discouraged?
I tried with and without start= pm.find_MAP(), and the convergence with the later one is improved a lot. But then I saw the documentation that it’s not suggested… I am wondering if there was just luck or something I missed or problems I didn’t know…