search

jmaih / RISE_toolbox

Solution and estimation of Markov Switching Rational Expectations / DSGE Models
BSD 3-Clause "New" or "Revised" License
106 stars 78 forks source link

Specification of chanis in Markov-chain #167

Closed narmindavoudi closed 2 years ago

narmindavoudi commented 3 years ago

Dear Maih, At firs, thank you for your help about working on rise-toolbox. I have a question about acceptance rate in Metropolis Hasting algorithm, 1)after estimating and simulating the posterior distributions, in set-upping the options for this algorithm, I want to be able to change the number of chins in Markov chain so that acceptance rate to stand between 0.3 and 0.5 as the same to dynare ,could you please guide me how I can do that. actually, results of plot of posteriors are not in the standard shape, and I guess it is because of acceptance rate which can be controlled by changing of chain number. 2) I chose100.000 for "ndraws_mcmc" and I wonder if is there any reasonable number of drawing parameters through MCMC to reach a standard shape of posteriors density? Best regard Narmin

jmaih commented 2 years ago

Dear Narmin,

If you run "help mh_sampler" in the command window you should be able to see the various options you can set in order to change. In your specific case, the option is called "nchain".

The option for controlling the acceptance rate is called "alpha", which you should set as an interval. For instance, options.alpha=[.25,.45] . The sampler will then target the middle of the interval

With respect to results of the posterior plots, they will depend on the sampler you use, the number of draws you consider and the various options you set.

There is no standard number of draws one should use. Various people use different numbers often depending on the time budget at hand and how hard it is to compute the likelihood.

If you are now doing any thinning I think 100 000 draws is not enough. The metropolis-hastings algorithm will typically give you highly correlated draws.

In summary, look up the various options you can set in the help for the mh_sampler.

Cheers,

J.

On Wed, Nov 24, 2021 at 12:51 PM Narmin davoudi @.***> wrote:

Dear Maih, At firs, thank you for your help about working on rise-toolbox. I have a question about acceptance rate in Metropolis Hasting algorithm, 1)after estimating and simulating the posterior distributions, in set-upping the options for this algorithm, I want to be able to change the number of chins in Markov chain so that acceptance rate to stand between 0.3 and 0.5 as the same to dynare ,could you please guide me how I can do that. actually, results of plot of posteriors are not in the standard shape, and I guess it is because of acceptance rate which can be controlled by changing of chain number. 2) I chose100.000 for "ndraws_mcmc" and I wonder if is there any reasonable number of drawing parameters through MCMC to reach a standard shape of posteriors density? Best regard Narmin

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jmaih/RISE_toolbox/issues/167, or unsubscribe https://github.com/notifications/unsubscribe-auth/AATKBTYBQFEKRUWLX4VW2I3UNTGTXANCNFSM5IV3VU6A . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

narmindavoudi commented 2 years ago

Thank you for your help. best regard Narmin