Multi-start module #555

EveCharbie · 2022-10-19T21:28:17Z

All Submissions:

Have you followed the guidelines in our Contributing document [docs/contribution.md]?
Have you checked to ensure there aren't other open [Pull Requests] for the same update/change?
Have you opened/linked the issue related to your pull request? -> Issue Have a module for multistart OCP which use MPI #393
Have you used the tag [WIP] for on-going changes, and removed it when the pull request was ready?
When ready to merge, have you sent a comment pinging @pariterre in it?

New Feature Submissions:

Does your submission pass the tests (if not please explain why this is intended)?
Did you write a proper documentation (docstrings and ReadMe)
Have you linted your code locally prior to submission (using the command: black . -l120 --exclude "external/*")?

Changes to Core Features:

Have you added an explanation of what your changes do and why you'd like us to include them?
Have you written new examples for your core changes, as applicable?
Have you written new tests for your core changes, as applicable?

This change is

Ipuch

Reviewable status: 0 of 5 files reviewed, 1 unresolved discussion (waiting on @EveCharbie)

bioptim/optimization/multi_start.py line 67 at r1 (raw file):

                        else:
                            list_out += list_now

list_out = [instance for instance in product(*self.args_dict.values())]

pariterre

Reviewed 5 of 5 files at r1, all commit messages.
Reviewable status: all files reviewed, 12 unresolved discussions (waiting on @EveCharbie)

bioptim/examples/getting_started/example_multistart.py line 113 at r1 (raw file):

def solve_ocp(args: list = None):

Remove this function

Code quote:

def solve_ocp(args: list = None):

bioptim/examples/getting_started/example_multistart.py line 130 at r1 (raw file):

    biorbd_model_path = args[0]
    final_time = args[1]

All this should be in prepare_multistart (as suggested later)

Code quote:

    biorbd_model_path = args[0]
    final_time = args[1]
    n_shooting = args[2]
    seed = args[3]

    ocp = prepare_ocp(biorbd_model_path, final_time, n_shooting)
    ocp.add_plot_penalty(CostType.ALL)
    sol = ocp.solve(Solver.IPOPT(show_online_optim=False))  # You cannot use show_online_optim with multi-start

bioptim/examples/getting_started/example_multistart.py line 136 at r1 (raw file):

    ocp = prepare_ocp(biorbd_model_path, final_time, n_shooting)
    ocp.add_plot_penalty(CostType.ALL)
    sol = ocp.solve(Solver.IPOPT(show_online_optim=False))  # You cannot use show_online_optim with multi-start

This should be called by the run()

Code quote:

sol = ocp.solve(Solver.IPOPT(show_online_optim=False))  # You cannot use show_online_optim with multi-start

bioptim/examples/getting_started/example_multistart.py line 137 at r1 (raw file):

    ocp.add_plot_penalty(CostType.ALL)
    sol = ocp.solve(Solver.IPOPT(show_online_optim=False))  # You cannot use show_online_optim with multi-start
    ocp.save(sol, f"solutions/pendulum_multi_start_random{seed}.bo", stand_alone=True)

The user can do this with the output from .run()

Code quote:

ocp.save(sol, f"solutions/pendulum_multi_start_random{seed}.bo", stand_alone=True)

bioptim/examples/getting_started/example_multistart.py line 144 at r1 (raw file):

        solve_ocp,
        n_random=10,
        n_pools=4,

Suggestion:

    ocp = solve_ocp(biorbd_model_path, final_time, n_shooting, seed=42)
    ocp.add_plot_penalty(CostType.ALL)
    return MultiStart(
        ocp,
        n_random=10,
        n_pools=4,
    )

bioptim/examples/getting_started/example_multistart.py line 155 at r1 (raw file):

        biorbd_model_path=["models/pendulum.bioMod"], final_time=[1], n_shooting=[30, 40, 50]
    )
    multi_start.run()

This should returns all the solution

multi_start.solve(solver=Solver.IPOPT(show_online_optim=False))

Suggestion:

multi_start.solve(Solver.IPOPT(show_online_optim=False))

bioptim/optimization/multi_start.py line 23 at r1 (raw file):

        n_random: int = 1,
        n_pools: int = 1,
        args_dict: dict = None,

Suggestion:

    def __init__(
        self,
        ocp,
        n_random: int = 1,
        n_pools: int = 1,
        **kwargs,
    ):

bioptim/optimization/multi_start.py line 38 at r1 (raw file):

        """

        self.solve_ocp = solve_ocp

self.ocp = ocp
self.solver = solver

bioptim/optimization/multi_start.py line 41 at r1 (raw file):

        self.n_random = n_random
        self.n_pools = n_pools
        self.args_dict = args_dict

Suggestion:

self.args_dict = **kwargs

bioptim/optimization/multi_start.py line 79 at r1 (raw file):

        """
        with Pool(self.n_pools) as p:
            p.map(self.solve_ocp, self.combined_args_to_list)

Suggestion:

p.map(self.ocp.solve, self.combined_args_to_list)

tests/test_global_getting_started.py line 1146 at r1 (raw file):

        biorbd_model_path=[bioptim_folder + "/models/pendulum.bioMod"], final_time=[1], n_shooting=[5, 10]
    )
    multi_start.run()

.solve()

EveCharbie

It is not exactly what you asked for, but I think it respects the idea the most possible

Reviewable status: 3 of 5 files reviewed, 12 unresolved discussions (waiting on @Ipuch and @pariterre)

bioptim/examples/getting_started/example_multistart.py line 113 at r1 (raw file):

Previously, pariterre (Pariterre) wrote…

Remove this function

Done.

bioptim/examples/getting_started/example_multistart.py line 130 at r1 (raw file):

Previously, pariterre (Pariterre) wrote…

All this should be in prepare_multistart (as suggested later)

Done.

bioptim/examples/getting_started/example_multistart.py line 136 at r1 (raw file):

Previously, pariterre (Pariterre) wrote…

This should be called by the run()

Done.

bioptim/examples/getting_started/example_multistart.py line 137 at r1 (raw file):

Previously, pariterre (Pariterre) wrote…

The user can do this with the output from .run()

Not possible to return sol because it cannot be pickled.

bioptim/examples/getting_started/example_multistart.py line 144 at r1 (raw file):

        solve_ocp,
        n_random=10,
        n_pools=4,

I cannot put ocp since the arguments would be fixed, but I put prepare_ocp

bioptim/examples/getting_started/example_multistart.py line 155 at r1 (raw file):

Previously, pariterre (Pariterre) wrote…

This should returns all the solution

multi_start.solve(solver=Solver.IPOPT(show_online_optim=False))

Done.

bioptim/optimization/multi_start.py line 23 at r1 (raw file):

        n_random: int = 1,
        n_pools: int = 1,
        args_dict: dict = None,

Done.

bioptim/optimization/multi_start.py line 38 at r1 (raw file):

Previously, pariterre (Pariterre) wrote…

self.ocp = ocp
self.solver = solver

Done.

bioptim/optimization/multi_start.py line 41 at r1 (raw file):

        self.n_random = n_random
        self.n_pools = n_pools
        self.args_dict = args_dict

Done.

bioptim/optimization/multi_start.py line 67 at r1 (raw file):

Previously, Ipuch (Pierre Puchaud) wrote…

from itertools import product
list_out = [instance for instance in product(*self.args_dict.values())]

Done

bioptim/optimization/multi_start.py line 79 at r1 (raw file):

        """
        with Pool(self.n_pools) as p:
            p.map(self.solve_ocp, self.combined_args_to_list)

Done.

tests/test_global_getting_started.py line 1146 at r1 (raw file):

Previously, pariterre (Pariterre) wrote…

.solve()

Done.

pariterre

Reviewed 4 of 4 files at r2, 1 of 1 files at r3.
Reviewable status: 4 of 6 files reviewed, 4 unresolved discussions (waiting on @EveCharbie and @Ipuch)

bioptim/examples/getting_started/example_multistart.py line 85 at r3 (raw file):

    u_bounds[1, :] = 0  # Prevent the model from actively rotate

    # To be changed when the issue on NoisedInitialGuess is fixed

TODO

bioptim/examples/getting_started/example_multistart.py line 143 at r3 (raw file):

        prepare_ocp,
        solver=Solver.IPOPT(show_online_optim=False),  # You cannot use show_online_optim with multi-start
        callback_function=save_results,

post_optimization_callback=save_results,

Code quote:

callback_function

bioptim/optimization/multi_start.py line 46 at r3 (raw file):

        self.prepare_ocp = prepare_ocp
        self.solver = solver
        self.callback_function = callback_function

post_optimization_callback

Code quote:

.callback_function = callback_function

small steps

pariterre · 2022-10-28T00:12:52Z

@EveCharbie I rebased this PR to integrate the new tests. Please use reset --hard (don't merge) to move to that PR if you need to push something later. Call me if needed

pariterre · 2022-10-28T18:58:25Z

@EveCharbie
Tests fails so renaming to normal name!

codecov · 2022-11-02T16:39:03Z

Codecov Report

Base: 83.76% // Head: 80.47% // Decreases project coverage by -3.28% ⚠️

Coverage data is based on head (8136cc5) compared to base (5027e98).
Patch coverage: 70.55% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #555      +/-   ##
==========================================
- Coverage   83.76%   80.47%   -3.29%     
==========================================
  Files          91       95       +4     
  Lines       10155    10382     +227     
==========================================
- Hits         8506     8355     -151     
- Misses       1649     2027     +378

Impacted Files	Coverage Δ
bioptim/examples/acados/static_arm.py	`42.85% <0.00%> (ø)`
...tim/examples/getting_started/example_multiphase.py	`89.28% <0.00%> (-1.63%)`	⬇️
bioptim/interfaces/sqp_interface.py	`0.00% <0.00%> (ø)`
bioptim/optimization/optimization_variable.py	`85.29% <ø> (ø)`
...tim/examples/getting_started/example_multistart.py	`28.12% <28.12%> (ø)`
bioptim/optimization/optimal_control_program.py	`87.03% <50.00%> (-0.31%)`	⬇️
bioptim/interfaces/solver_options.py	`71.11% <61.45%> (-1.74%)`	⬇️
bioptim/optimization/multi_start.py	`88.46% <88.46%> (ø)`
bioptim/optimization/solution.py	`86.26% <88.88%> (-0.10%)`	⬇️
bioptim/interfaces/interface_utils.py	`90.38% <90.38%> (ø)`
... and 10 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

pariterre

Reviewed 2 of 2 files at r4, 2 of 2 files at r6, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @Ipuch)

pariterre

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @Ipuch)

Ipuch requested changes Oct 20, 2022

View reviewed changes

pariterre requested changes Oct 20, 2022

View reviewed changes

EveCharbie commented Oct 21, 2022

View reviewed changes

EveCharbie force-pushed the Multi-start branch from f26c550 to 780f33e Compare October 21, 2022 21:29

pariterre requested changes Oct 27, 2022

View reviewed changes

pariterre changed the title ~~[RTR] Multi-start module~~ [RTM_special] Multi-start module Oct 27, 2022

EveCharbie added 16 commits October 27, 2022 20:10

added an example of how to use multi-start

f852555

added the multi-start module

b7ef059

blacked

11c4980

updated readme

eba70ac

added warning in comments since I cannot chaek it

873a6de

added a test

7feb5b2

Tryied to improve readme

c1885a0

requested changes

4ee7ba6

small steps

kwargs

5ad607d

still working

59339f5

working

9422c6a

still working

546aae0

I think we are good :)

c0147b2

addapted test + blacked

65ccd5a

added TODO

47768a3

post_optim_callback name

ff72b50

pariterre force-pushed the Multi-start branch from 215b60f to ff72b50 Compare October 28, 2022 00:12

pariterre changed the title ~~[RTM_special] Multi-start module~~ Multi-start module Oct 28, 2022

EveCharbie added 2 commits November 2, 2022 11:50

fixed Rotation test but I did not play with that :/

43834c2

same for quaternions

8136cc5

pariterre approved these changes Nov 2, 2022

View reviewed changes

pariterre merged commit a49807b into pyomeca:master Nov 2, 2022

EveCharbie deleted the Multi-start branch May 30, 2023 07:14

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Multi-start module #555

Multi-start module #555

EveCharbie commented Oct 19, 2022 •

edited

Ipuch left a comment •

edited

pariterre left a comment

EveCharbie left a comment

pariterre left a comment

pariterre commented Oct 28, 2022

pariterre commented Oct 28, 2022

codecov bot commented Nov 2, 2022

pariterre left a comment

pariterre left a comment

Multi-start module #555

Multi-start module #555

Conversation

EveCharbie commented Oct 19, 2022 • edited

All Submissions:

New Feature Submissions:

Changes to Core Features:

Ipuch left a comment • edited

Choose a reason for hiding this comment

pariterre left a comment

Choose a reason for hiding this comment

EveCharbie left a comment

Choose a reason for hiding this comment

pariterre left a comment

Choose a reason for hiding this comment

pariterre commented Oct 28, 2022

pariterre commented Oct 28, 2022

codecov bot commented Nov 2, 2022

Codecov Report

pariterre left a comment

Choose a reason for hiding this comment

pariterre left a comment

Choose a reason for hiding this comment

EveCharbie commented Oct 19, 2022 •

edited

Ipuch left a comment •

edited