Implement type consistent ancestor callback#125
Conversation
|
SSMProblems.jl/SSMProblems documentation for PR #125 is available at: |
|
SSMProblems.jl/GeneralisedFilters documentation for PR #125 is available at: |
Codecov Report✅ All modified and coverable lines are covered by tests.
Additional details and impacted files@@ Coverage Diff @@
## main #125 +/- ##
===========================================
- Coverage 65.26% 43.90% -21.37%
===========================================
Files 30 5 -25
Lines 1212 82 -1130
===========================================
- Hits 791 36 -755
+ Misses 421 46 -375
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
An attempt at solving #124.
I think this is probably the cleanest approach without forcing the initial and following states to be of the same type.
I've only implemented the dense ancestor callback for now. Will also do sparse if @charlesknipp likes these changes.
I'm not sure this is a good idea to make all callbacks behave like this in general. For example, the GPU version of this should really store everything a big N x (K+1) array so we would need to force constant types in that case.
For the CPU case though, I think it's reasonable to allow different types.