DataModel: validate constraint array lengths against the row count

Aperta
  • Qualcosa non va?
#2,254 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
3/5
Tempo stimato
Mezza giornata
Idoneità per principianti
50/100
Tipo di issue
Bug
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
cpp, python
Ambito
backend

Direzione di ricerca

Start in the setters in data_model_wrapper.pyx that store rhs, row types and the constraint lower/upper bounds, and compare their lengths with len(A_offsets) - 1. First check whether the C++ data_model_view already validates these sizes, since that decides where the check belongs. Done means a clear ValueError for mismatched lengths or half-set bounds, plus a test that covers the bounds-only path used by cvxpy, pyomo and linopy.

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

Descrizione

Problem

DataModel accepts two row representations: rhs + row types (set_constraint_bounds, set_row_types) and lower/upper bounds (set_constraint_lower_bounds, set_constraint_upper_bounds). The setters in data_model_wrapper.pyx store whatever array they receive. Nothing checks that its length equals the row count implied by A_offsets (len(A_offsets) - 1).

The solve path then passes each array to the C++ data_model_view with its own shape[0]. Any consumer that assumes one representation, or assumes the sizes match, can read out of bounds. #2249 is one instance, in _compute_slack_csr, which indexed rhs[i] and sense[i] when only the bounds were set. cvxpy, pyomo and linopy all use set_constraint_lower_bounds, so this path is exercised by third-party interfaces.

Proposal

Validate at the Python boundary (the setters, or the solve entry once the matrix is known):

  • rhs, row_types, constraint_lower_bounds and constraint_upper_bounds must each be empty or have length equal to the row count.
  • Lower and upper bounds must be set together or not at all.
  • Raise a clear ValueError otherwise.

Optionally, as Ishika Roy suggested, normalize so that rhs + sense is always converted to lower/upper bounds when the model is built, so consumers only deal with one representation. That is a larger change and could be a second step.

Open question

Does the C++ data_model_view already validate these sizes against the row count? That needs checking, since the Python-side check would then be a duplicate of it.

Related: #2249, #2253, #1615.

Lingua principale
Cuda
Stelle
1.1k
Fork
249
Merge medio
1g 15h
PR unite (30g)
124

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 NVIDIA/cuopt

Tutte le issue di NVIDIA/cuopt

Issue simili

Altre issue su Backend & API Design

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.