Replace "map do" with Ensemble Simulations for DeepBSDE solver. #44
Conversation
|
@vboussange can you have a look and merge? |
src/DeepBSDE.jl
Outdated
| (X,u) = (predict_ans[1:(end-1)], predict_ans[end]) | ||
| output_func(sol,i) = ((sol[end][1:end-1], sol[end][end]),false) | ||
| function prob_func(prob,i,repeat) | ||
| SDEProblem(prob.f , prob.g , init_cond, prob.tspan , prob.p ,noise_rate_prototype = prob.noise_rate_prototype) |
There was a problem hiding this comment.
| SDEProblem(prob.f , prob.g , init_cond, prob.tspan , prob.p ,noise_rate_prototype = prob.noise_rate_prototype) | |
| SDEProblem(prob.f , prob.g , init_cond, prob.tspan , prob.p ,noise_rate_prototype = copy(prob.noise_rate_prototype)) |
Make sure you aren't reusing caches amongst threads. Best would be to just create noise as a vector matching nthreads so you don't collide, but given the cost of a solve here I think it's fine just to copy.
|
CI didn't run? |
|
@ChrisRackauckas Yes, it looks like only the Documentation action is triggered with PR. Do you have any idea why? |
|
The CI was set to branch-specific PR checking. Fixed. |
|
Why are the MLP.jl tests failing? |
|
Looks fine, but should probably figure out why tests are failing, and if that was happening before this PR. |
|
@ashutosh-b-b, the MLP tests were failing on the original branch |
ashutosh-b-b
force-pushed
the
DeepBSDE_ensemble
branch
from
b3f281b to
937a9fc
Compare
|
@vboussange The non linear tests are failing due to a |
No description provided.