Tracer plots for LDhat run

[Homap]

Hello,

I ran interval on 2000 SNPs of a region about 300 Kb. I have 10 diploid individuals. I ran interval with block penalty of 5, 10 million iterations and sampled every 2000 steps. I use a subset of likelihood table for 20 haplotypes and a theta of 0.001. The pi in my samples is about 0.002 and theta is very close to it.

The effective sample sizes (ESS) are very low for likelihood and blocks and there is a large difference between their ESS and the one of map.

Sorry as this is a very general questions but I wonder if you have any suggestions on how I could improve this? Is it the matter of running the chain longer or some other potential issue? Could the likelihood table be the problem? Is there a difference in the likelihood table when using genotypes vs. haplotypes?

Thank you!
Homa

[Homap]

Very interesting, I just converted the length of region to Kb, keeping everything else constant. The result changed drastically:

Why would this change the output? It's really interesting.