ashander / ftprime

Forward-time simulation of the msprime data structure (for development)

2 stars 1 forks source link

Implement/alg with recomb #3

Closed ashander closed 7 years ago

ashander commented 7 years ago

Implements algorithm 1 w recomb.

See devel/test_alg.py for (approximately correct!) output corresponding to writeup in algorigthm-notes.md

todos:

[ ] get import to msprime working
[ ] get subset w msprime working
[ ] confirm trees are correct
[ ] update merge to concatenate adjacent records
[ ] finish general renumbering scheme
[ ] write testing functions (that compare expected trees to those we get)
[ ] test with more examples

ashander commented 7 years ago

so negative numbers in the initial state aren't counted and the endpoints of maxn are treated correctly in renumbering --- or at least the dict returned by Population.renumber() accurately describes the renumbering. the trees output you were looking at in devl/test_alg.py is unchanged from this fix but mapping no longer contain the extraneous 12:0 entry

ashander commented 7 years ago

Status

`merge_records`

we have much better tested ftprime.merge.merge_records that I think works on all fronts except one well defined one of not merging abutting records with same children -- this was merged in via #4

effects

(found in 34ee7c6 )

this changes the trees slightly (from python alg) vs the result we were getting this morning -- this was found in
the result prevents msprime version of alg T from working
cause / correctness of new trees is TBD

todos

see the top comment of this PR which I've edited to add some todo items

ashander commented 7 years ago

Implemented these goals in master and as module in #10