It seems to deliver concrete speed improvements, most other projects use it, we should also.
However, the tests/test-suitability-report.R fails, in a combination of:
Using g3_suitability_andersen() twice (replacing one instance with g3_suitability_exponentiall50() is enough to get things going again)or maybe not, replacing the prey_a instance causes things to work, prey_b not so much.g3_suitability_andersen(p0 = 0, ...) Using pred_a with g3_suitability_andersenfleet() is enough to trigger, fleet_a is fine. The obvious difference being catchability being used more than once.
1885e40
auto suitability instead of vector<Type> suitability (can also cast suitability). Assigning a type to catchability doesn't help.
The TMB hint is "Variable not initialized?", so presumably it's expecting to share some part of the tape, only to find it's not there.
lentinj
commented
1 week ago
It seems to deliver concrete speed improvements, most other projects use it, we should also.
However, the
tests/test-suitability-report.Rfails, in a combination of:Usingg3_suitability_andersen()twice (replacing one instance withg3_suitability_exponentiall50()is enough to get things going again)or maybe not, replacing theprey_ainstance causes things to work,prey_bnot so much.g3_suitability_andersen(p0 = 0, ...)pred_awithg3_suitability_andersenfleet()is enough to trigger,fleet_ais fine. The obvious difference beingcatchabilitybeing used more than once.auto suitabilityinstead ofvector<Type> suitability(can also castsuitability). Assigning a type tocatchabilitydoesn't help.The TMB hint is "Variable not initialized?", so presumably it's expecting to share some part of the tape, only to find it's not there.