Skip to content

Optimization based solvers #350

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 59 commits into
base: master
Choose a base branch
from
Open

Optimization based solvers #350

wants to merge 59 commits into from

Conversation

ErikQQY
Copy link
Member

@ErikQQY ErikQQY commented Jul 14, 2025 •
edited
Loading

Part of #336

Some TODOs:

  • Multi-points BVP are encountering some DI issues, but two-point BVP are fine, need to dig deeper to see what's going wrong.
  • Need some patches in SciMLBase.jl and DiffEqBase.jl for better interface.
  • More docstrings and documentations

@github-actions GitHub Actions
Copy link
Contributor

github-actions bot commented Jul 14, 2025 •
edited
Loading

Benchmark Results

Click to check benchmark results
master 8a6fc74... master / 8a6fc74...
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK2() 0.582 ± 0.015 s 0.587 ± 0.015 s 0.992 ± 0.037
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK3() 13.5 ± 0.57 ms 13.6 ± 0.54 ms 0.991 ± 0.057
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK4() 3.11 ± 0.16 ms 3.13 ± 0.19 ms 0.993 ± 0.078
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK5() 9.19 ± 1.3 ms 9.28 ± 1.3 ms 0.99 ± 0.2
Simple Pendulum/IIP/BoundaryValueDiffEqMIRK.MIRK6() 1.6 ± 0.2 ms 1.62 ± 0.29 ms 0.989 ± 0.22
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = false) 1.86 ± 0.66 ms 1.9 ± 0.66 ms 0.979 ± 0.49
Simple Pendulum/IIP/MultipleShooting(10, Tsit5; grid_coarsening = true) 3.14 ± 0.99 ms 3.16 ± 0.97 ms 0.994 ± 0.44
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = false) 0.0672 ± 0.0068 s 0.0676 ± 0.007 s 0.994 ± 0.14
Simple Pendulum/IIP/MultipleShooting(100, Tsit5; grid_coarsening = true) 0.0802 ± 0.0076 s 0.0802 ± 0.012 s 1 ± 0.18
Simple Pendulum/IIP/Shooting(Tsit5()) 0.251 ± 0.078 ms 0.256 ± 0.078 ms 0.982 ± 0.43
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK2() 0.73 ± 0.0057 s 0.741 ± 0.016 s 0.984 ± 0.022
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK3() 16.3 ± 5 ms 16.7 ± 5.1 ms 0.98 ± 0.42
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK4() 3.58 ± 0.16 ms 3.6 ± 0.15 ms 0.995 ± 0.06
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK5() 10.7 ± 1.1 ms 10.8 ± 0.71 ms 0.989 ± 0.12
Simple Pendulum/OOP/BoundaryValueDiffEqMIRK.MIRK6() 1.86 ± 0.14 ms 1.9 ± 0.14 ms 0.975 ± 0.11
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = false) 3.56 ± 2.9 ms 3.62 ± 2.9 ms 0.982 ± 1.1
Simple Pendulum/OOP/MultipleShooting(10, Tsit5; grid_coarsening = true) 6 ± 5.1 ms 6.13 ± 5.2 ms 0.978 ± 1.2
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = false) 0.115 ± 0.0032 s 0.117 ± 0.0044 s 0.987 ± 0.046
Simple Pendulum/OOP/MultipleShooting(100, Tsit5; grid_coarsening = true) 0.143 ± 0.0053 s 0.142 ± 0.0072 s 1.01 ± 0.064
Simple Pendulum/OOP/Shooting(Tsit5()) 0.647 ± 0.036 ms 0.678 ± 0.045 ms 0.954 ± 0.083
time_to_load 5.34 ± 0.0033 s 5.29 ± 0.0076 s 1.01 ± 0.0016
### Benchmark Plots A plot of the benchmark results has been uploaded as an artifact to the workflow run for this PR. Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).

ErikQQY and others added 27 commits July 15, 2025 01:56
@ErikQQY
Don't fail other solvers
1cc250d
@ErikQQY
Fix unrelated failings
5931430
@ErikQQY
Fix more concrete solve algorithm
fac21aa
@ErikQQY
Add default cost
f9cd33f
@ErikQQY
Nested FIRK should work with optimization solvers
c694c26
@ErikQQY
More docstrings
93ea85a
@ErikQQY
Proper assert
54943e8
@ErikQQY
Expanded FIRK should work
5c21462
@ErikQQY
Fix Aqua error
ce2b3e0
@ErikQQY
Fix typo in compat
3b905c9
@ErikQQY
Shooting and MultipleShooting work fine
a041b90
@ErikQQY
MIRKN work fine
b7f5762
@ErikQQY
All solvers should be done
130da34
@ErikQQY
Fix kwargs issue with shooting
d7e4ef4
@ErikQQY
Fix inplace assert in multiple shooting
9901afb
@ErikQQY
Fix inplace assert in single shooting
f710cd7
@ErikQQY
Merge branch 'master' into qqy/bvp_opt
5cc940a
@ErikQQY
Use correct setup in FIRK testing
81da921
@ErikQQY
Unify testing environment
4d677c7
@ErikQQY
Fix ensemble test for FIRK
66a6b89
@ErikQQY
Fix more FIRK tests
44d556f
@ErikQQY
Fix FIRK on simple pendulum
2237f70
@ErikQQY
Set default abstol for nonlinear solving in shooting
318f015
@ErikQQY
Fix kwargs issue with Shooting
ebc6dca
@ErikQQY
Fix multiple shooting only use inplace problem constructor
3137c0b
@ErikQQY
Fix typo
3d0dc52
@ErikQQY
Use proper Jacobian algorithm
f16dc9a
@ErikQQY ErikQQY marked this pull request as ready for review August 13, 2025 19:10
@ErikQQY ErikQQY closed this Aug 14, 2025
@ErikQQY ErikQQY reopened this Aug 14, 2025
@ErikQQY

This comment was marked as outdated.

Sign in to view
@ErikQQY ErikQQY mentioned this pull request Aug 26, 2025
Closed
@ErikQQY ErikQQY closed this Aug 28, 2025
@ErikQQY ErikQQY reopened this Aug 28, 2025
@ErikQQY
Copy link
Member Author

ErikQQY commented Aug 28, 2025 •
edited
Loading

I think this PR is ready to go now.

The optimal control interface still needs some more work considering the complex whole-time interval inequality(similar to @variable(model, v[1:T] >= 0, start = v_0) in JuMP.jl), but let's do that in a following PR since that would involve changing the way we build the loss function here.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@github-actions github-actions[bot] github-actions[bot] left review comments

@ChrisRackauckas ChrisRackauckas Awaiting requested review from ChrisRackauckas

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@ErikQQY