tag:blogger.com,1999:blog-3240271627873788873.post8655331292502043905..comments2020-09-28T10:06:25.279-04:00Comments on Doing Bayesian Data Analysis: Thinning to reduce autocorrelation: Rarely useful!John K. Kruschkehttp://www.blogger.com/profile/17323153789716653784noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-3240271627873788873.post-43034817280016315452015-12-08T09:37:12.498-05:002015-12-08T09:37:12.498-05:00Here are more details regarding the previous comme...Here are more details regarding the previous comment. I have not read the paper, so am not endorsing it, merely reporting more info for interested readers.<br /><br />Active link: <a href="http://arxiv.org/abs/1510.07727" rel="nofollow">http://arxiv.org/abs/1510.07727</a><br /><br />Info:<br /><br />Statistically efficient thinning of a Markov chain sampler<br /><br />Art B. Owen<br /><br />(Submitted on 27 Oct 2015 (v1), last revised 11 Nov 2015 (this version, v3))<br /><br />It is common to subsample Markov chain samples to reduce the storage burden of the output. It is also well known that discarding k−1 out of every k observations will not improve statistical efficiency. It is less frequently remarked that subsampling a Markov chain allows one to omit some of the computation beyond that needed to simply advance the chain. When this reduced computation is accounted for, thinning the Markov chain by subsampling it can improve statistical efficiency. Given an autocorrelation parameter ρ and a cost ratio θ, this paper shows how to compute the most efficient subsampling frequency k. The optimal k grows rapidly as ρ increases towards 1. The resulting efficiency gain depends primarily on θ, not ρ. Taking k=1 (no thinning) is optimal when ρ≤0. For ρ>0 it is optimal if and only if θ≤(1−ρ)2/(2ρ). The efficiency gain never exceeds 1+θ. John K. Kruschkehttps://www.blogger.com/profile/17323153789716653784noreply@blogger.comtag:blogger.com,1999:blog-3240271627873788873.post-16030798306951901842015-12-08T05:56:24.416-05:002015-12-08T05:56:24.416-05:00I think you could find this paper interesting http...I think you could find this paper interesting http://arxiv.org/abs/1510.07727aloctavodiahttps://www.blogger.com/profile/09470130971106757391noreply@blogger.comtag:blogger.com,1999:blog-3240271627873788873.post-13797720787247790692014-01-24T16:40:58.152-05:002014-01-24T16:40:58.152-05:00Dear Dr. Kruschke,
Thank you for a very interesti...Dear Dr. Kruschke,<br /><br />Thank you for a very interesting post. Would you happen to know where I could get hold of the paper you mention ("On thinning of chains in MCMC")? I do not have access to it through my university library...<br /><br />Thanks again for another great post.<br /><br />PatrickPatrickhttps://www.blogger.com/profile/11301659025178029016noreply@blogger.com