Scientific Computing
Preliminary / method-support scientific computing: numerical workflow review, approximation / convergence / error summary, solver-selection support, interpolation / discretization framing, numeric pipeline checklist, and computational cost / stability notes. PRELIMINARY ONLY — not a substitute for a computational-science specialist, production-grade solver stacks (PETSc / SUNDIALS / FEniCS / OpenFOAM / Trilinos / etc.), HPC execution, or certified numerical accuracy for regulated applications.
Numerical Workflow Review
StandardExpected-stage coverage check by problem type (ODE / PDE / linear-algebra / optimisation / root-finding / integration / Monte-Carlo / data-fit / generic). Gap reporting, not an execution audit.
⏱ 3-6 minutes
Convergence / Error Summary
StandardLog–log fit on {h, error} runs to estimate the empirical order of convergence, with pairwise order estimates and optional gap-vs-expected-order. For real verification prefer method-of-manufactured-solutions.
⏱ 3-6 minutes
Solver Selection Support
StandardHeuristic solver shortlist by problem class (ODE-IVP / ODE-BVP / PDE-elliptic / parabolic / hyperbolic / linear systems / nonlinear / eigenvalue / optimisation / integration / Monte-Carlo), size, stiffness, and structure. Heuristic matching — not a benchmark.
⏱ 3-6 minutes
Interpolation / Discretization Framing
StandardMethod-card shortlist for interpolation / differentiation / integration / PDE-spatial / PDE-temporal by data shape and smoothness. Framing only — validate with a known-answer test before production.
⏱ 3-6 minutes
Numeric Pipeline Checklist
StandardCanonical V&V (verification / validation) + reproducibility checklist by pipeline type (simulation / inverse-problem / parameter-study / data-assimilation / uncertainty-quantification). Framework only — does NOT execute the checks.
⏱ 3-6 minutes
Computational Cost / Stability Notes
StandardPer-operation complexity tiering (O(1) / log / linear / n log n / n² / n³ / exponential) with optional condition-sensitivity and stability notes. Qualitative profile — no real profiling or constant-factor analysis.
⏱ 3-6 minutes