Repository navigation
Description
Goal
The repository has been rewritten end to end: code, tests, documentation, workflows. Most of it comes from #310, but not all of it. This issue tracks a review of the result by theme. It does not check that the plan of #310 was followed. It checks that the result holds together:
- Coherence: code, docstrings, READMEs, documentation and
AGENTS.mdsay the same thing. - Conventions: the code follows ct-conventions v0.14.0 (pinned in
AGENTS.md). Cite rules by identifier. - Presentation: developers find their way through the READMEs, and users through the documentation.
- Improvements: what could be simpler, clearer, or more automated.
How to work on this
- Review one theme at a time. Post findings as a comment here, headed by the theme number (
T3 — …). - Trivial fixes: one small PR per theme,
Refs #<this issue>. Anything larger gets its own issue, linked from the theme. - Tick a theme when it has been reviewed and its findings are fixed or turned into issues.
- The "Known leads" below were spotted while preparing this issue. Check them first, but the review is not limited to them.
Suggested order: T1 → T2 → T3 → T4 → T5 → T8, with T6 and T7 in parallel. The user documentation describes everything else, so it is best re-read after the code has settled.
Themes
T1 — Package architecture and public API (src/)
Scope: src/OptimalControlProblems.jl, the submodules Modelings, Metadata, References, Catalogue, Problems, src/README.md.
- Each submodule has one responsibility, and they depend on each other in the right order (
MOD1,MOD2,MOD7). - Imports are qualified and the public / internal split is explicit (
MOD3,MOD4,MOD8). - Errors are typed and chosen by the choice rule (
EXC1). Hot paths are type-stable (TYP8,PERF*), and 1-D values are scalars (DIM1). - Docstrings of public symbols are complete and match the code (
DOC1–DOC8). - The API is right-sized: no redundant accessors, and names are consistent with OptimalControl v2.
Reviewed and fixed in #457 (findings); the API levels page is #458.
Known leads:
- OptimalControlProblems is a user-facing package. Like OptimalControl, it exports its user API at top level ([Dev] Split src into submodules (ct-conventions MOD1, MOD2), keeping the public exports #413). Only the internal CT packages export nothing.
MOD8now says so (ct-conventions v0.12.0, rules(MOD8): user-facing packages export their user API ct-conventions#56), and the pin andAGENTS.mdare updated (chore: pin ct-conventions v0.12.0, MOD8 for a user-facing package #450). Still to check: is every exported name really user API, and does thetest_modules.jlcheck enforce it? - API levels page ([Doc] An API levels page: exported, qualified, internal #458): a page that tells users which names are exported, qualified (
OptimalControlProblems.status) or internal, generated from the module, as OptimalControl'sapi/levels.md. The T1 review found the documentation already follows this split.
T2 — Problem data: metadata, descriptions, references (ext/MetaData, with the descriptions since #465, ext/MetaData/references)
-
METADATA_SCHEMAcovers every field in use, and no field is unused. - Every
<problem>.jluses the same structure (:grid_size,:parameters,:status,:init,:notes,:reference_solution…). - Descriptions: same outline, correct mathematical formulation (matching the models), and sourced
### References. - Reference TOML files: provenance is clear, and
tools/referencesregenerates them. -
:draftstatuses and:notesare still current.
Reviewed and fixed in #460 and #461 (findings). The descriptions share one outline, checked by a test. One thing is left: whether each mathematical formulation matches its models (to read in T6/T7). The descriptions are now next to the metadata, ext/MetaData/<problem>.md (#465).
T3 — Modelings (ext/*Models, ext/*.jl)
Scope: OptimalControl, OptimalControlScalarized, JuMP (and JuMPModels/schemes/), ModelingToolkit (proof of concept, #308).
- Every problem file in a modeling follows the same template (as described in
ext/README.md): checked by the linter (feat(tools): a linter of the problems (T3 of #449) #467), and a new problem starts from the skeleton oftools/new/problem.jl(feat(tools): the skeleton of a new problem (T3 of #449) #468). - The same problem is written the same way across modelings (names, ordering, parameters, initial guess).
- JuMP schemes:
build.jl/data.jl/tableaux.jlare clear and documented, and matchguide/jump-schemes.md. - Signs of costates and duals (
rules/signs.md; open: [Dev] Align the costate and dual signs of the JuMP models with OptimalControl #326, [Dev] Expose the duals of the JuMP path, boundary and box constraints, aligned with signs.md #430). -
ext/README.mdmatches the actual layout.
Progress: the linter of the problems (#467: what it checks, what other tools check, what is left to check by hand, as a PR checklist; bounds from the parameters, dead reads removed), the skeleton of a new problem or modeling (#468), bryson_denham JuMP named as in OptimalControl and :valid (#469), qualified imports in the OptimalControl and JuMP extensions (#471, MOD4). The cart_pendulum comment on J = mcart is left for the check of the formulations (T6/T7). The layout block of AGENTS.md goes to T8.
Known leads:
- The layout block in
ext/README.mddoes not listModelingToolkitModels.jl/ModelingToolkitModels/. - The layout block in
AGENTS.mdlistsOptimalControlScalarizedModels,MetaDataandDescriptionsas extensions and omitsModelingToolkitModels.Project.tomldeclares three extensions:JuMPModels,ModelingToolkitModels,OptimalControlModels.
T4 — Test suite (test/)
Scope: test/runtests.jl, test/common/, test/suite/<env>/, test/cron/, test/robustness/, test/extras/, test/README.md.
- The layout per environment is consistent: every environment has the same kinds of files (
test_smoke,test_methods,test_problem…). The shared tests are written once for the four modelings (refactor(test): one set of tests for the four modelings (T4 of #449) #476); a table intest/README.mdsays what each modeling tests, and why not. - The three kinds of tests (does not break / same NLP / convergence) are clearly separated, and each test belongs to one kind.
-
test/common/holds no duplicated or dead helpers. The GPU smoke test is shared (Share the GPU smoke test between OptimalControl and OptimalControlScalarized #445). The modeling tests are shared (refactor(test): one set of tests for the four modelings (T4 of #449) #476), the GPU smoke test too andcommon/status.jlis split (refactor(test): the GPU smoke test written once, status.jl split, coverage gaps (T4 of #449) #477). - Testing rules: fakes at top level, names, structure (
TST*). Coverage is checked: 92.3 % on Codecov, the misses being the ModelingToolkit extension, now covered (ci(coverage): the ModelingToolkit extension covered (T4 of #449) #478, feat(coverage): a test_args input, as in ci.yml CTActions#178), and theshowmethods and the discovery of the problems, now tested (refactor(test): the GPU smoke test written once, status.jl split, coverage gaps (T4 of #449) #477). -
test/extras/andtest/robustness/still have a clear purpose, and are documented:test/extras/was a dead script, removed (refactor(metadata): no grid without a modeling (T4 of #449) #475); the robustness is shared (refactor(test): one set of tests for the four modelings (T4 of #449) #476). -
test/README.md(300 lines) is accurate and easy to read (refactor(metadata): no grid without a modeling (T4 of #449) #475, refactor(test): one set of tests for the four modelings (T4 of #449) #476).
T5 — CI, crons, references pipeline, release (.github/, tools/)
Scope: .github/workflows/*.yml, cron-reusable.yml, labels, tools/ci/, tools/cron/, tools/references/, .github/workflows/README.md, CHANGELOG.md, BREAKING.md.
- Every workflow is still needed. Triggers and labels are consistent with the label list and
test/README.md(label dispatch, CTActions#149). - Cron issues ([Cron] cron-failure #435, [Cron] cron-convergence — 1 finding #436) are readable and actionable, as
.github/workflows/README.mddescribes. - The Python tools are tested and documented. Is a tool duplicated in CTActions?
-
CHANGELOG.md/BREAKING.mdcover all user-visible changes since 0.5 (CHG*), and the migration guide matches them. - The release path is ready (
REL1–REL7, ct-agents v0.4.0).
Known leads:
Project.tomlis still at0.5.0, while [Tracking] Per-modeling lists of problems, drift-proof tests and cron #310 planned0.6.0-betaonmain. Is this intended until the release?- The full CI matrix is still to be restored (Restore the full CI matrix (Julia 1.10, macOS, Windows) after the reorganisation #344).
- Rotating crons with a time budget (decided 2026-11-20) — done for ModelingToolkit in feat(cron): a rotation with a time budget, one process per problem (#435) #466, with a runner that leaves no report shown as a finding. Still to apply to the other crons. Each run tests the problems tested longest ago, read from the date of their last result on the
resultsbranch, as many as fit in a time budget (e.g. 1h).- A new problem comes first, and a crashed run makes no problem lose its turn. Where needed, the cadence goes up (ModelingToolkit: daily).
- Start with the ModelingToolkit cron ([Cron] cron-failure #435): its Ubuntu runner dies after ~1h20, most likely out of memory, and its Windows job exceeds the 6h limit. Run one Julia process per problem. Then apply it to the other crons.
- The main work is in the reports. The cron issues ([Cron] cron-failure #435, [Cron] cron-convergence — 1 finding #436, [Cron] cron-drafts — 2 ready drafts #454), the drafts check and the references must be built from the last result of each problem, not from the current run only. Otherwise a problem that did not run looks "resolved".
- The modelings are independent (decided 2026-11-20) but for the definition of the problems (parameters, initial guess, description): each has its own configurations, grid and references. Done:
- perf(test): ModelingToolkit solves with a quasi-Newton Hessian, warmed up on the smoke grid (#435) #472: the ModelingToolkit solves with a quasi-Newton Hessian, as Corleone does, warmed up on the smoke grid;
- feat(metadata): the grid of a problem by modeling, coarser for ModelingToolkit (#435) #473:
grid_size_data(problem, backend), an optional:grid_sizesper problem, at most 50 steps for ModelingToolkit (solved by shooting).bioreactorin the ModelingToolkit cron: converged in 24 s and 2.2 GB, instead of not converged after 682 s and 7.5 GB; - feat(references): the references by modeling, each from its own solves (#435) #474: the references by modeling,
[<modeling>.<scheme>], each from its own solves; the reference API takes the backend (breaking, inCHANGELOG.md/BREAKING.md); how close the modelings are is computed on the problem pages. - Next: the ModelingToolkit references on its grid, from the next cron (
tools/references/update.jl).
T6 — User documentation, hand-written pages (read by the maintainer)
Scope: docs/src/index.md, getting-started/*, guide/*, problems-overview.md, migration.md.
- Reading path: Introduction → Installation → First problem → Guided tour. A newcomer can follow it.
- Each guide page answers one question, without overlapping the others.
- The code examples run and are up to date with OptimalControl v2 (dev site).
- Tone, length and terminology are consistent (problem / modeling / backend / scheme).
-
migration.mdmatchesBREAKING.md. - ModelingToolkit / Corleone: no page shows how to solve a ModelingToolkit problem with Corleone. The
ModelingToolkitsnippet indocs/modelings.jlstops atsys = model(pb). Add a "Solve with ModelingToolkit" guide page, similar to "Solve with OptimalControl" and "Solve with JuMP". It should cover the grids,initial_guess(pb)and the Ipopt solve, and link to the Corleone documentation for details. The problem-page tab is in [Cron/Doc] ModelingToolkit convergence cron, trajectories, and documentation (#308) #439.
Per-page reading checklist:
-
index.md -
getting-started/installation.md -
getting-started/first-problem.md -
getting-started/guided-tour.md -
guide/get.md -
guide/solve-optimalcontrol.md -
guide/solve-jump.md -
guide/jump-schemes.md -
guide/initial-guess.md -
guide/compare.md -
guide/plot.md -
problems-overview.md -
migration.md
T7 — Generated documentation and site machinery
Scope: docs/make.jl, docs/{problems,modelings,results,browser,status,api_reference}.jl, docs/src/components/*.vue, docs/src/.vitepress/.
- Problem pages: content, tabs per modeling, plots, results from the crons. Read a few pages in full (one per kind of problem).
- Problems browser, gallery and Status page: usable, filters useful, client-side rendering (
SITE*). - API Reference: complete, nothing internal leaking, extensions documented.
- Build scripts: readable, no logic duplicated with
src/(one source of truth), and the build fails loudly when data is missing. - Light/dark theme, mobile width, version picker.
Known leads:
- Local preview that stays browsable during a build, as OptimalControl's
docs/preview.sh:make.jlclearsdocs/build/first, so a server on it returns 404 while it builds. - API levels page ([Doc] An API levels page: exported, qualified, internal #458), generated by
docs/api_reference.jl.
T8 — Entry points for developers and repository metadata
Scope: README.md, README.template.md, CONTRIBUTING.md, AGENTS.md, CLAUDE.md, docs/src/developers.md, CITATION.*, .zenodo.json, Project.toml (authors, compat).
- A developer arriving on the repository knows where to go:
README.md→developers.md→ the per-directory READMEs. Nothing is duplicated, and nothing is missing. -
AGENTS.mdmatches the repository (layout, key conventions, recurring tasks). - Citation metadata is consistent: authors, title, DOI.
Known leads:
README.mdsays "Each problem is available both in the OptimalControl DSL and in JuMP". This is no longer true: modelings have their own lists, and ModelingToolkit has been added.README.mdsays the package requires Julia 1.10 or later. Check this against CI (Restore the full CI matrix (Julia 1.10, macOS, Windows) after the reorganisation #344).- The LICENSE badge in
README.mdpoints to themasterbranch, but the default branch ismain. CONTRIBUTING.mdcopies the "Contributing" and "Developer Resources" sections ofREADME.mdword for word. Keep one, and link to it from the other.Project.tomlauthorslists one person, while the citation lists four.
Versions: ct-conventions v0.14.0, ct-agents v0.5.0 (updated from v0.11.0 / v0.4.0 on 2026-11-19).
Activity
ocots commentedon Nov 19, 2026
T1 — Package architecture and public API (
src/)Reviewed against ct-conventions v0.14.0, ct-agents v0.5.0, on
mainateba3d484.What holds
MOD1,MOD2,MOD7). One directory per submodule, manifests that only declare, dependencies in the documented order (Modelings,Metadata→References,Catalogue→Problems).src/README.mdmatches the code.MOD3,MOD4). All qualified (using X: X,using ..Sub: Sub), noimport.MOD8), the known lead. The split is consistent: what the package exports is what the documentation writes unqualified (problems, the backends, the problem functions,Problem, the accessors,problem_metadata,grid_size_data,parameters_data); what it leaves to its submodules is what the documentation writes qualified (status,features,reference_objective,initial_guess_data).test/suite/core/test_modules.jlenforces the structure (every package export is a binding of a submodule, every submodule export is reachable from the package). Whether a name is user API is a judgment no test can make; nothing to change here.DIM1. No hot path insrc/: it reads metadata (OrderedDict{Symbol,Any}by design) and builds values once. Nothing to check with JET here; the hot paths are in the extensions (T3).Findings
EXC1: untyped errors at load time.src/Metadata/loading.jl(4×error(...): missing, unknown or mistyped metadata field) andsrc/References/objectives.jl(2×: malformed reference table). Each says that one value of a data file is wrong →CTBase.Exceptions.IncorrectArgumentwithgot/expected.EXC1:features(:unknown)throws anUndefVarError(getfield(Catalogue, problem)insrc/Problems/problem.jl) instead of theIncorrectArgumentofproblem_metadata(:unknown). Check the name first, throughMetadata.problem_metadata.Catalogue.status. It repeats the unknown-problem check ofMetadata.problem_metadata, with an unstructured message (nogot/expected/suggestion). It can beMetadata.problem_metadata(problem)[:status].Metadata.mergere-implementsBase.merge. Four methods of a new functionmerge(shadowingBase.mergeinMetadata), each with a full docstring though they are internal, andparameters_data(problem, parameters)catches thePreconditionErrorof one of them to rethrow another. Simpler: one internal helper on the four cases (nothing/NamedTuple), throwing the finalPreconditionErrordirectly, andBase.mergefor twoNamedTuples.src/OptimalControlProblems.jl("OptimalControl, its scalarized form, JuMP") andsrc/Modelings/Modelings.jl(lists three backends) predate the ModelingToolkit modeling.Metadata/discovery.jlcomputes the path ofext/again instead of sharingModelings.EXT_PATH. Acceptable:Metadatadepends on no submodule, and moving the constant would add a dependency for one path. Leave as is.Proposal
Findings 1–5 in one small PR,
Refs #449. No change to the exported API.🤖 Generated with Claude Code
ocots commentedon Nov 19, 2026
T2 — Problem data: metadata, descriptions, references
Reviewed against ct-conventions v0.14.0, ct-agents v0.5.0, on
mainateba3d484, with the cron results of theresultsbranch (JuMP 2026-11-17, OptimalControl 2026-11-18, OptimalControlScalarized 2026-11-19).What holds
METADATA_SCHEMAand every field of the schema is used (notes:bryson_denham,mountain_car;reference_solution:beam,bryson_denham). The schema is enforced at load time (typed errors since refactor(src): typed errors and simpler parameter merge (T1 of #449) #457) and bytest_consistency.jl, which also checks that a status is declared exactly for the modelings having a model file.grid_size,parameters,status,notes,init,reference_solution. The status written on one line or several is the formatter's choice.tools/references/update.jland must not be edited by hand, and each result keeps its modeling, configuration, platform and run.tools/references/README.mdsays when to compute them again. Last regenerated on 2026-11-07 ([Metadata] Reference solutions with how they were obtained: typed references and API #409).### Referenceswith entries (test_sources).Findings
bryson_denham: a stale note and draft. JuMP is:draft, and the note says "JuMP uses ASCII state component names while the OptimalControl model uses Unicode names". This is no longer true: both models usex1,x2. JuMP passes the convergence cron on Ubuntu and Windows and the robustness cron, with the objective of OptimalControl (4.000143). → Remove the note and set JuMP to:valid. The drafts cron ([Cron] cron-drafts — 2 ready drafts #454) did not list it as ready: to check in T5.mountain_car: an inaccurate note. It puts the gap on the scalarized model with MadNLP (102.381 against 102.398). The crons show the same 102.381 formodeler=exa solver=madnlpin both OptimalControl and OptimalControlScalarized, and 102.3976 with Ipopt in every modeling. The gap comes from the configuration (exa + MadNLP), not from the scalarized modeling. → Rewrite the note.beam,brachistochrone,bryson_denham,cart_pendulum,ducted_fan).balanced_fieldhas no "Qualitative behaviour".robbinshas no "Mathematical formulation" heading: its formulation is in the introduction.space_shuttlehas "References and relevance".→ Use one outline: Mathematical formulation, Parameters (if any), Qualitative behaviour, Characteristics, References. Check it with a test next to
test_sources, so that it cannot drift again.features(docs/problems.jl,facts): dimensions, free final time, the kinds of constraints (e.g.mountain_car,chain). They are correct today, but they are written by hand and can drift from the models. → Keep only what cannot be computed (the structure of the solution, why the problem is a benchmark), and drop the restated facts. Maintainer's call.bryson_denham(Windows, 2026-11-11,shooting=single solver=ipopt) has the objective 6.875, against 4.000 for the other modelings.update.jlruns again, does a ModelingToolkit row (a shooting method, not a scheme) go into thetrapezetable?Proposal
Findings 1–3, with the outline test, in one small PR,
Refs #449. Finding 4 waits for the maintainer.🤖 Generated with Claude Code
docs: a page whose description has no qualitative behaviour points to…
fix(data): keep bryson_denham draft in JuMP, its note is right
ocots commentedon Nov 19, 2026
T2 — correction of finding 1. The note of
bryson_denhamis right: the OptimalControl model declaresx ∈ R², whose components arex₁,x₂, while the JuMP model names themx1,x2; the comparison of the modelings (test_features) fails when JuMP is:valid. The review had only looked at the parameter names. The status stays:draft(reverted in #460). Making it valid means renaming the JuMP componentsx₁,x₂(T3).🤖 Generated with Claude Code
Merge pull request #457 from control-toolbox/fix/t1-review
ocots commentedon Nov 19, 2026
Status on 2026-11-19, and what comes next
Conventions: ct-conventions v0.14.0, ct-agents v0.5.0.
Done
bryson_denham: its JuMP draft and note are right).docs/preview.sh(docs: a local preview that stays browsable while the docs are rebuilt #459, merged).mountain_car(feat(models): mountain_car with a free final time, in the ModelingToolkit modeling (#308) #452). Both merged.Open pull requests
features, and a placeholder on the page when a description has no qualitative behaviour.:parameter_info, the unit and the meaning of every parameter, required in the metadata and shown by the generated Parameters section. The descriptions lose their hand-written parameter sections, so the values are written once. Retarget it tomainonce fix(data): stale notes, one outline for the descriptions (T2 of #449) #460 is merged.Next
The ModelingToolkit cron ([Cron] cron-failure #435). It fails on Ubuntu after about 1h20, with no failed step and no log. GitHub's annotation: "The hosted runner lost communication with the server … starves it for CPU/Memory". The runner dies, not the tests: most likely out of memory. The whole ModelingToolkit/Corleone suite runs in one Julia process. The plan:
This is also T5.
T2, after fix(data): stale notes, one outline for the descriptions (T2 of #449) #460 and feat(data): the unit and the meaning of every parameter, in the metadata (T2 of #449) #461: move
ext/Descriptions/<problem>.mdnext toext/MetaData/<problem>.jl, one directory for the hand-written data of a problem. No renaming, no folder per problem.T3, already found:
brachistochronebounds its control by ±1.57 written in the code (u_min,u_maxare unused), and its final time by 0.1 and 20;balanced_field:v_min,v_max,α_minare unused, and its bounds are written in the code;space_shuttle:α_t0is unused;cart_pendulum: a comment questionsJ = mcart;bryson_denham: the JuMP componentsx1,x2should bex₁,x₂for it to become valid.T3/T5:
bryson_denham(6.875 against 4.000);trapezetable of the references;bryson_denham.ModelingToolkit: the other problems with the reduced-time convention, from
main, one commit per problem; and the documentation page on the reduced time. Wait for a working cron (1.) before setting them:valid.The remaining themes: T3 → T4 → T5 → T8, with T6 and T7 in parallel.
🤖 Generated with Claude Code
fix(data): mountain_car note, now that ModelingToolkit is valid
Merge pull request #460 from control-toolbox/fix/t2-review
Merge pull request #461 from control-toolbox/feat/parameter-info
Merge pull request #465 from control-toolbox/refactor/descriptions-ne…
ocots commentedon Nov 20, 2026
T3 — Modelings
Reviewed against ct-conventions v0.14.0, ct-agents v0.5.0, on
mainat79465cdf. The checks of the template, of the parameters and of the bounds were run by script on the 72 model files, not read by eye.What holds
parameters::Union{Nothing,NamedTuple}=nothing,init=nothing,scheme/grid_sizefor JuMP;_warn_if_draft);parameters_data;_initial_guess, [Metadata] Shared initial guess in the metadata, read by every modeling #377).:validin two modelings, the names of the components and the continuous functions are compared by the tests (test_features, the continuous comparison), and the NLPs by the comparison cron (Rationalize JuMP and comparison test suites: separate generic transcription unit tests from per-problem modeling checks #441).schemes/{data,tableaux,build}.jlhave docstrings, andguide/jump-schemes.mdmatches them: the four schemes (:trapeze,:euler,:euler_implicit,:midpoint), their tableaux, the number of controls, and the names the section "In the code" cites.ext/README.mdmatches the layout. Its first known lead (ModelingToolkit missing) is fixed, and the descriptions are now next to the metadata (refactor(data): the description of a problem next to its metadata (T2 of #449) #465).Findings
Bounds written in the code, parameters unused.
brachistochrone(OptimalControl, scalarized, JuMP) bounds the control by-1.57 ≤ u ≤ 1.57and the final time by0.1 ≤ tf ≤ 20, written in the code. The parametersu_min,u_max(±π) are read but unused.balanced_fieldwritestf ≥ 0.1andv ≥ 1in its OptimalControl models, andv ≥ 1,h ≥ 0,tf ≥ 0.1in JuMP. Itsα_minis read but unused (the models use0 ≤ α), andv_min,v_maxare never read.→ Read every bound from the parameters, with the defaults set to the values the models use today, so that no objective changes. Then drop the "not used yet" from their
:parameter_info.Dead reads.
mountain_carreadstf_start, andspace_shuttlereadsα_s,β_s, in all three modelings; only the initial guess of the metadata uses them, a leftover of [Metadata] Shared initial guess in the metadata, read by every modeling #377.space_shuttle'sα_t0is used nowhere. The ModelingToolkit models ofchainandjacksonreadt0without using it. → Remove the reads, and the parameterα_t0.MOD4in two extensions.ext/OptimalControlModels.jlandext/JuMPModels.jluse bareusing OptimalControlProblems,using OptimalControl,using JuMP,using DocStringExtensions;ext/ModelingToolkitModels.jlis compliant. Fixing it qualifies every name in about 60 problem files (OptimalControl.@def,OptimalControlProblems.parameters_data…). It is mechanical, but it changes how the files read. Maintainer's call.bryson_denhamin JuMP: rename the componentsx1,x2tox₁,x₂, those of the OptimalControl model (x ∈ R²). Then it can become:valid(see the correction in T2).cart_pendulum: the dynamics carry the commentJ = mcart # should be -(m+mcart) but was mcart?. Check it against the source of the problem.AGENTS.md(the second known lead): its layout block listsOptimalControlScalarizedModelsandMetaDataas extensions and omitsModelingToolkitModels.Project.tomldeclares three:JuMPModels,ModelingToolkitModels,OptimalControlModels. → Fix the block (T8).Signs of the costates and duals: still open in [Dev] Align the costate and dual signs of the JuMP models with OptimalControl #326 and [Dev] Expose the duals of the JuMP path, boundary and box constraints, aligned with signs.md #430, nothing new here.
Proposal
🤖 Generated with Claude Code
Merge pull request #467 from control-toolbox/feat/problem-lint
ocots commentedon Nov 20, 2026
T3, progress (ct-conventions v0.14.0, ct-agents v0.5.0)
tools/lint/problems.jl, what it checks, what other tools check, what is left to check by hand as a PR checklist), bounds from the parameters (brachistochrone,balanced_field), dead reads removed, "Add a problem" inAGENTS.md.tools/new/problem.jl new|add), passing the linter but for itsTODOs.bryson_denhamJuMP componentsx₁,x₂, JuMP:valid.OptimalControlModels.jl/JuMPModels.jl; thecart_pendulumJ = mcartcomment.AGENTS.mdlayout block: moved to T8.Merge pull request #468 from control-toolbox/feat/new-problem
Merge pull request #469 from control-toolbox/fix/bryson-denham-jump
ocots commentedon Nov 20, 2026
Progress, 2026-11-20 (ct-conventions v0.14.0, ct-agents v0.5.0)
bryson_denhamJuMP), refactor(ext): qualified imports in the OptimalControl and JuMP extensions (MOD4) #471 (MOD4) merged. The template item is ticked: the linter checks it. Thecart_pendulumJ = mcartcomment goes to the check of the formulations (T6/T7), theAGENTS.mdlayout to T8.ocots commentedon Nov 20, 2026
T4 — review of the test suite (2026-11-20, ct-conventions v0.14.0, ct-agents v0.5.0)
Findings:
grid_size_data(problem), not the grid of the modelinggrid_size_data(problem, backend)(feat(metadata): the grid of a problem by modeling, coarser for ModelingToolkit (#435) #473): the JuMP convergence cron, the JuMP and OptimalControl robustness. Same value today, wrong as soon as a problem has:grid_sizes. Decided: one public grid function,grid_size_data(problem, backend); the one-argument form goes (breaking, with the reference API of feat(references): the references by modeling, each from its own solves (#435) #474 in[Unreleased]).test_smoke,test_convergence,test_robustness, and OptimalControl its owntest_robustness, beside the sharedcommon/modeling_tests.jlof the three modelings that return aProblem. Proposal: a JuMP adapter and a sharedtest_robustness: one place for the four modelings.test_draft), OptimalControlScalarized has notest_featuresnortest_reference_solution, robustness exists for JuMP and OptimalControl only. Proposal: a table modeling × kind of test intest/README.md, each gap with its reason, the forgotten ones added.test/extras/beam.jlis a dead script (Pkg.activate("test"), bareusing, Plots, never run nor referenced): removed.test/README.mdis stale: ModelingToolkit has references of its own since feat(references): the references by modeling, each from its own solves (#435) #474, a coarser grid (feat(metadata): the grid of a problem by modeling, coarser for ModelingToolkit (#435) #473) and a quasi-Newton Hessian (perf(test): ModelingToolkit solves with a quasi-Newton Hessian, warmed up on the smoke grid (#435) #472); the modelings are independent.common/status.jl(615 lines) does several things (rows, results, trajectories, reports, environment files); left as is for now.Already right: no fake type inside a function (TST3); every test file has its entry function
test_<name>()(TST1).Plan: PR 1 = the grid function (b) + 1 + 4 + 5; PR 2 = 2 + 3.
Merge pull request #475 from control-toolbox/refactor/grid-by-modelin…
Merge pull request #476 from control-toolbox/refactor/shared-modeling…
Merge pull request #477 from control-toolbox/refactor/t4-leftovers
Merge pull request #478 from control-toolbox/feat/coverage-modelingto…