search

dipetkov / eems

Estimating Effective Migration Surfaces
GNU General Public License v2.0
102 stars 28 forks source link

How to improve the acceptance rate of "mBirthDeath"? #68

Open ZheCai opened 2 years ago

ZheCai commented 2 years ago

Hi all,

When running eems on my data, the acceptance rate of "mBirthDeath" is always lower than 10%. Which parameters I should tune to improve the acceptance rate?

Thanks!

The end of the output:

Ending iteration 20000000 with acceptance proportions: (463182/1249153) = 37.08% for proposal type "qTileRate", with proposal variance "qEffctProposalS2" (418360/1251545) = 33.43% for proposal type "qTileMove", with proposal variance "qSeedsProposalS2" (284344/1251259) = 22.72% for proposal type "qBirthDeath" (1543607/3749139) = 41.17% for proposal type "mTileRate", with proposal variance "mEffctProposalS2" (655820/2500832) = 26.22% for proposal type "mMeanRate", with proposal variance "mrateMuProposalS2" (1325567/3748755) = 35.36% for proposal type "mTileMove", with proposal variance "mSeedsProposalS2" (310662/3748524) = 8.29% for proposal type "mBirthDeath" (1188935/2500793) = 47.54% for proposal type "degrees of freedom" and effective degrees of freedom = 9562.16 number of qVoronoi tiles = 61 number of mVoronoi tiles = 134 Log prior = 49.15 Log llike = 1148073.79 Final log prior: 49.15 Final log llike: 1148073.79