Tool comparisons

Choose by workflow, not by a made-up score.

These tools are not interchangeable. Compare their role, the workflow you need and the evidence attached to your case. This page makes no cross-solver accuracy or speed ranking.

Choose by workflow

Start with what you need to do.

Use the tool role and the evidence available for your exact model. These routes are not ranked against each other.

Python FEM building blocks

Compose a custom workflow

SfePy and scikit-fem offer Python based tools for custom PDE or finite element workflows. The surrounding model still needs verification.

Compare roles and official sources →
Standalone solvers

Use an established FEA workflow

CalculiX and Code_Aster cover broader structural workflows. Confirm the exact formulation, version and evidence required for your case.

Compare roles and official sources →
Detailed roles, decision criteria and official sources for five tools
ToolRoleConsider it whenBoundary to checkPrimary source
QF SolverPython structural-mechanics solver with a public API and CLI.You want to inspect solver methods and run research or engineering-method workflows whose exact scope is recorded.Research and methods development. Not certified and not a general-purpose replacement for industrial FEA. Published maturity statuses carry forward from 0.2.8.GitHub release and wheel →
SfePyPython package for custom applications that solve PDEs with FEM in 1D, 2D and 3D.You want to describe a custom PDE problem and build a Python-native finite-element workflow.Check the formulation, solver and verification evidence for your own problem; a Python interface does not by itself qualify a model.Official SfePy documentation →
scikit-femPure-Python finite-element assembly library.You want FEM assembly components that fit into a custom Python/SciPy workflow.It is a library building block; assess the surrounding model, linear algebra and validation workflow you need.Official documentation →
CalculiXStandalone solver and pre/post-processor (CCX and CGX).You want an established structural FEA workflow with linear and nonlinear, static, dynamic and thermal solution routes.Confirm exact formulation, input conventions and solver version for the case; do not transfer QF Solver's qualification labels to CalculiX or vice versa.Official project site →
Code_AsterOpen-source numerical simulation software developed by EDF R&D for structural mechanics.You need a broader, established structural-mechanics solver workflow and its extensive reference documentation.Choose and verify the appropriate version, model setup and support path for the engineering context.Official product page →
A defensible numerical comparison needs comparable meshes, materials, loads, boundary conditions, solver settings, observables and tolerances. The QF Solver evidence page links only to recorded comparable cases.
QF Solver status

Current release: 0.2.11.

The public release and the qualification registry are separate facts. 0.2.11 is published; the 0.2.8 consolidated registry still governs the published maturity labels.

QF Solver fit

Use it when scope and evidence matter.

QF Solver is a research and engineering-method development project. Its strongest fit is a Python workflow where the formulation and its limits need to be inspected.

Good fit

  • You need to inspect the equations, implementation and diagnostics behind a declared structural-mechanics route.
  • You value reproducible inputs, tests and evidence that separates implementation from qualification.
  • You can keep the model inside the route’s published boundaries and perform case-specific review.

Check another workflow

  • Your organization requires a certified solver or a specific acceptance process.
  • You need general nonlinear, contact, friction or post-buckling capability beyond the recorded cases.
  • You expect hardware-independent GPU, mixed-mesh or distributed scaling guarantees.

The latest release is 0.2.11; published maturity labels still follow registry 0.2.8. Inspect the route-level matrix →

Planned comparison articles · eight reserved slots
Planned comparison articles

Four matched case slots per solver.

The menus mirror the four named solid-element cases in the current Code_Aster V&V record. Empty slots are planning placeholders, not comparison results.

Planned menu

Code_Aster

4 cases

Four named linear-static V&V cases are already part of the QF evidence record. These slots reserve the matching comparison menu; no new results are published here.

TET44-node linear tetrahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.Read the existing frozen correlation case →
HEX88-node linear hexahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.Read the existing frozen correlation case →
TET1010-node quadratic tetrahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.Read the existing frozen correlation case →
HEX2020-node quadratic hexahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.Read the existing frozen correlation case →
Planned menu

CalculiX

4 cases

Four matching case slots are prepared. Results will remain empty until model conventions, solver settings, observables, and tolerances can be compared consistently.

TET44-node linear tetrahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.
HEX88-node linear hexahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.
TET1010-node quadratic tetrahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.
HEX2020-node quadratic hexahedronPlanned · empty
Comparison article not publishedThis article slot is reserved for a future presentation of the comparison.
Every case needs the same mesh and model assumptions, equivalent loading and boundary conditions, declared solver settings, comparable observables, and stated tolerances before a numerical comparison is defensible.