Example gallery¶

This section maps core solver capabilities to runnable workflows in the repository, so users can quickly see the public examples that are supported today.

How examples are classified¶

Each example-local README is authoritative for its exact command, runtime, outputs, limitations, and evidence boundary.

Representative results¶

The panels below are lightweight documentation thumbnails, not raw benchmark archives. They point to the same workflows listed in the sections that follow.

Start With These¶

Use the capability matrix before assuming a workflow status. Use the promoted example contract when adding or auditing a gallery entry.

Quasi-static fracture

Implicit AT1/AT2 crack-path workflows with comparison reports and standard run metadata.

Dynamic fracture

Explicit dynamics for impact, branching, and perforated-plate crack-growth benchmarks.

Microstructured media

Perforated-plate and heterogeneous fracture cases used for dataset and morphology studies.

Engineering outputs

Force-displacement, energy, convergence, lockfile, and visualization artifacts from reproducible runs.

Solid mechanics

Linear elastic, hyperelastic, and J2 mechanics examples backed by the same sparse-solve and material-update paths used by the solver.

Capability-first pathways¶

Quasi-static fracture¶

The quasi-static family provides documented pathways for selected literature comparisons.

• SENT / SENS / TPB / L-panel: benchmark configs under configs/benchmarks/quasistatic/ and comparison scripts in matching example directories.

• Run example: python -m phast run examples/quasistatic/notched_holed_plate/config.yaml --validate-only

• Expected outputs: config.yaml, run_lockfile.json, results.csv, plots. Run compare.py in examples/quasistatic/notched_holed_plate/ for benchmark-level pass/fail summary.

Explicit dynamics fracture¶

• Branching, impact, perforation, Kalthoff-Winkler: configs/benchmarks/dynamic/*.yaml plus matching run folders under examples/dynamic/*.

• B7 dynamic branching: examples/dynamic/B7_dynamic_crack_branching_comsol/config.yaml is the full-plate COMSOL cross-check.

• Benchmark workflow:

  1. Use the command in the example-local README and write to an explicit --output_dir.

  2. Inspect the generated manifest, histories, and damage fields.

  3. If the README provides a case-specific comparison command, run that exact command. Do not assume that every dynamic example provides a generic compare.py.

  4. Where retained comparison evidence is supplied, the standard filenames are compare_report.txt and compare.png. Retained evidence is not a substitute for a comparison generated from a new run.

Solid mechanics FEA¶

• python -m phast run examples/solid_mechanics_beta/linear_plate/config.yaml Linear elastic plate FEA with displacement, stress, deformed-shape, response, and visual-manifest outputs.

• python -m phast run examples/solid_mechanics_beta/neohookean_plate/config.yaml Nonlinear neo-Hookean cantilever FEA with Newton convergence and final-state field outputs.

• python -m phast run examples/solid_mechanics_beta/j2_bar/config.yaml Mesh-level J2 plasticity bar FEA with von Mises and equivalent-plastic-strain fields.

Heterogeneous material fields¶

• python examples/heterogeneous_fields/run.py --config examples/heterogeneous_fields/parameters.yaml --output-dir runs/heterogeneous_fields constructs element-ordered E(x) and Gc(x) fields, evaluates a mechanics-derived tensile history under an imposed affine strain, and solves bounded AT2 damage.

• This is a script-contract teaching example because arbitrary material-field maps are not currently represented by the general YAML runner.

• It is not a coupled equilibrium benchmark, cohesive interface model, or validated microstructure-fracture study. See the example README before adapting image or segmentation data.

Beta plasticity, cohesive, and PF-CZM validation¶

• J2 plasticity and ductile PF-plasticity: examples/plasticity_interface_beta/run_j2_validation.py and examples/plasticity_interface_beta/run_ductile_pf_plasticity_validation.py.

• Cohesive/interface benchmarks: mode-I jump, mixed-mode, contact-compression, delamination patch, structural DCB, diffuse-interface, and coupled PF-cohesive validation workflows live under examples/plasticity_interface_beta/.

• Manifested reproduction set: configs/benchmarks/plasticity_interface/reproducibility_contracts.yaml.

• Generated visual outputs: the scripts write *.png, manifest metadata, and summaries into their chosen output directories. These are beta validation artifacts until the nonlinear support criteria in the capability matrix are satisfied.

Benchmark Catalogue¶

YAML-driven benchmark configurations live under configs/benchmarks/. Runnable public examples carry their config.yaml files directly in examples/. Prefer the example-local command when a case is curated, because the folder also contains the README, setup image, result visuals, and postprocessing scripts expected by the public example contract.

Use python -m phast run <config.yaml> --validate-only to parse and schema-check a benchmark without running the solve. This is configuration validation, not scientific or mesh-convergence validation. The full YAML schema is documented in configs/REFERENCE.yaml.

Quasi-static Benchmarks¶

Miehe tension is already documented as an example-local YAML workflow under examples/quasistatic/miehe_tension/. Older Miehe shear and three-point bending variants should stay outside the gallery until they satisfy the same flat YAML, manifest, and visualization contract.

Dynamic Benchmarks¶

Acceptance Checks¶

Some curated benchmark examples provide a local comparison utility, such as compare.py; others provide only retained comparison evidence. Follow the example-local README rather than inferring a command from another case. When a comparison utility is present, run it after the simulation completes so that it loads the new output and reference data and writes case-specific artifacts. The acceptance metric may concern peak load, load-displacement envelope error, branching onset time, final crack-path morphology, or required-output checks.

For load-displacement comparisons, configure the reaction writer explicitly:

output:
  reaction_node_set: <named_node_set>
  reaction_component: 1  # 0=x, 1=y

The benchmark command should not be the only record of a claim. Keep the validated configuration, lockfile, metadata, histories, retained visuals, and comparison output together with the example or run artifact.

Capability boundaries¶

For current non-hardened features, review Capability matrix. This doc is also the source of truth for what can be promised publicly.

Sparse direct vs CG inner solve¶

Driver: sparse_solve. Speedup is wall-time of the sparse direct path versus the matrix-free CG inner-solve path on the same mesh, tolerance, backend stack, and output settings. See Performance and Reproducibility for the reporting checklist before publishing numbers.