V8 — does the covariate path name the generating covariates at a fitted estimate?
| Status | registered 2026-10-31; runs on V7's trees once their dd fits are written |
| Series | the validation series (replaces E20, E21b, E22, E33) |
| Package build | emphasis 182055b (1.0.0.9000), R 4.5.3 |
| Driver | sim/V8.R |
| Data | results/V8/jobs/, results/V8/build.txt |
Question
The path selects covariates from one draw set without fitting each model. Earlier runs drew at the generating values (unavailable to a user) or read \(D\) from the old likelihood; E33 at a fitted estimate failed.
Design
| trees | V7's |
| draws | 3000 by thinning at V7's dd estimate, the 150 of largest weight kept |
| path | covariate_path() over \(N\), \(D\), ED, \(K\), speciation and extinction, BIC |
| jobs | 640, each run once V7's dd fit of its tree is written |
Pre-set rules
- On
crandddtrees no lineage-level covariate is active at BIC's optimum on at least 90%. - On each lineage-level generator's trees its covariate is active on at least 80%.
Results
Not yet.