[Cron] cron-convergence — 1 finding

Aperta
  • Qualcosa non va?
#436 3 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
4/5
Tempo stimato
3-5 giorni
Idoneità per principianti
32/100
Tipo di issue
Bug
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
julia
Ambito
machine-learning, testing

Direzione di ricerca

Read the Workflow README section linked in the issue, then reproduce the moonlander JuMP robustness finding with Pkg.test(; test_args=["JuMP", "robustness", "problem=moonlander"]). Compare problems/moonlander/result.json and cron/JuMP-robustness.md with the linked run report to determine whether the loss of convergence is expected or a regression. The issue does not specify a code change; it remains open as a monitoring dashboard.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

cron-convergence

Solver Convergence and Robustness Report

About this tracking:
Numerical convergence is reported, not tested (#360): a solver failing to converge or finding an objective that shifts does not fail the CI. These scheduled crons periodically re-solve the entire problem catalogue to catch silent regressions, sensitivity to solver parameters, or better local optima.

What to do when an alert (⚠️) is flagged:

  1. Check which problem and solver configuration lost convergence or shifted objective.
  2. Reproduce locally with the Julia command provided in the section.
  3. Decide if the change is expected (e.g. updated solver version, improved initial guess) or a regression.
  4. If expected, recompute reference objectives (tools/references/README.md). This issue stays open as a permanent dashboard.
    Details: Workflow README.

Overall Scope Status

Scope Focus Cadence Status
JuMP — Nominal Convergence Default solver configuration (IPOPT/MadNLP on default discretization grids) Weekly (Tuesday 04:00 UTC) ✅ Clean
JuMP — Solver Robustness Perturbs IPOPT solver parameters (mu_strategy, mu_init) to test numerical stability (#378) Monthly (1st, 04:00 UTC) ⚠️ 1 finding
OptimalControl — Nominal Convergence High-level OptimalControl.jl modeling on standard grids and default solvers Weekly (Wednesday 04:00 UTC) Awaiting run
OptimalControl — Solver Robustness Perturbs solver parameters for OptimalControl.jl formulations (#378) Monthly (1st, 04:00 UTC) ✅ Clean
OptimalControlScalarized — Nominal Convergence Scalarized modeling where vector states/controls are unrolled into scalar components Weekly (Thursday 04:00 UTC) Awaiting run

JuMP — Nominal Convergence

Focus: Default solver configuration (IPOPT/MadNLP on default discretization grids)
Cadence & runners: CronJuMP.yml · weekly (Tuesday 04:00 UTC) · Julia 1.13, Ubuntu + Windows, CPU

✅ Clean · Run 42494171122 · 2026-11-08 15:27 UTC · Full scope report · Previous run

No findings.

JuMP — Solver Robustness

Focus: Perturbs IPOPT solver parameters (mu_strategy, mu_init) to test numerical stability (#378)
Cadence & runners: CronRobustness.yml · monthly (1st, 04:00 UTC) · Julia 1.13, Ubuntu + Windows, CPU

⚠️ 1 finding · Run 42494459775 · 2026-11-08 15:32 UTC · Full scope report · Previous run

Platform Problem Modeling Test configuration Change
1.13-ubuntu-latest moonlander JuMP solver=ipopt mu_strategy=monotone mu_init=0.001 converged → not_converged
🛠️ How to investigate
  • Local reproduction: 
    using Pkg; Pkg.test(; test_args=["JuMP", "robustness", "problem=moonlander"])
  • Data files: problems/moonlander/result.json · cron/JuMP-robustness.md
  • Outcome guide:
    • converged → <status>: problem converged in the previous run, but not in this run
    • objective a → b (±x): converged in both runs, but the objective value moved by more than relative threshold 1e-4
    • new solution: converged to a previously unknown local solution (#378)
📜 Recent history
  • 2026-11-08 · run — new: moonlander (converged → not_converged)
## OptimalControl — Nominal Convergence

Focus: High-level OptimalControl.jl modeling on standard grids and default solvers
Cadence & runners: Weekly (Wednesday 04:00 UTC) · Julia 1.13, Ubuntu + Windows (CPU + GPU)

Awaiting next scheduled run.

OptimalControl — Solver Robustness

Focus: Perturbs solver parameters for OptimalControl.jl formulations (#378)
Cadence & runners: CronRobustness.yml · monthly (1st, 04:00 UTC) · Julia 1.13, Ubuntu + Windows, CPU

✅ Clean · Run 42494459775 · 2026-11-08 15:56 UTC · Full scope report · Previous run

No findings.

📜 Recent history
  • 2026-11-08 · run — resolved: insurance (all clear)
  • 2026-11-08 · run — new: insurance (converged → not_converged)
## OptimalControlScalarized — Nominal Convergence

Focus: Scalarized modeling where vector states/controls are unrolled into scalar components
Cadence & runners: Weekly (Thursday 04:00 UTC) · Julia 1.13, Ubuntu + Windows (CPU + GPU)

Awaiting next scheduled run.

Lingua principale
Julia
Stelle
13
Fork
1
Merge medio
3h 33m
PR unite (30g)
55

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di control-toolbox/OptimalControlProblems.jl

Tutte le issue di control-toolbox/OptimalControlProblems.jl

Issue simili

Altre issue su Julia

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.