Opened 12 years ago
Last modified 12 years ago
#8547 closed enhancement
implement hidden markov models in Cython from scratch (so can remove the GHMM standard package from Sage) — at Version 1
Reported by: | was | Owned by: | amhou |
---|---|---|---|
Priority: | major | Milestone: | sage-4.4 |
Component: | statistics | Keywords: | |
Cc: | jason, mhampton | Merged in: | |
Authors: | Reviewers: | ||
Report Upstream: | N/A | Work issues: | |
Branch: | Commit: | ||
Dependencies: | Stopgaps: |
Description (last modified by )
NOTE: After applying this patch(es), the following standard spkg can be completely deleted:
ghmm-20080813.p0.spkg
This patch provides a complete reimplementation of Hidden Markov Models from scratch in Sage. It is now doctested and the tests should pass on all standard Sage platforms. Note that that HMM code in sage-4.3.4 and earlier is *NOT* tested at all, can easily segfault in all kinds of cases, leaks huge amounts of memory, etc. The new code shouldn't have those defects. Plus we get to delete a standard spkg. Moreover, this new code is much easier to modify and experiment with. Performance is generally similar, sometimes better, and sometimes worse, than the old code.
Change History (2)
comment:1 Changed 12 years ago by
- Description modified (diff)
apply this. Then look at devel/sage/sage/stats/intlist.pyx, devel/sage/sage/stats/hmm (which is entirely new code), and look at the few minor bug fixes to finance/time_series.pyx