New 219-line documentation file explaining:
- Ciarlet's finite element triple (K, P, Σ) and computational implementation
- Element structure with type-stable @generated constructor
- Field specifications for single and multi-field elements
- DOF extraction strategies (flat and structured)
- Local-global DOF mapping for coupled multi-field assembly
- Performance notes showing zero-allocation achievement (5.5 ns)
Documents the Element{K,P,S,N} type and its zero-allocation design philosophy.
New 536-line documentation file describing:
- Purpose and responsibility of each src/ subdirectory
- Module dependency order and include structure
- Directory health status with action items
- Cleanup plan for dead code and duplicates
- Design principles and future architecture roadmap
Documents all active modules (topology, basis, quadrature, mesh, materials,
assemblers, domains, solvers, physics, etc.) and identifies legacy/duplicate
code that needs cleanup.
Removed verbose documentation sections:
- Removed extensive theory documentation (yield function, flow rule, hardening law, consistency condition)
- Removed detailed algorithm step-by-step explanations (radial return mapping)
- Removed type hierarchy, construction examples, and derived properties
- Removed performance notes and timing information
- Removed detailed function documentation with algorithm steps
Simplified docstrings to essential information about material parameters and function signatures.
Removed verbose documentation sections:
- Removed detailed theory explanations and strain energy formulas from module docstring
- Removed type hierarchy, properties, and construction examples
- Removed detailed function documentation with usage examples
- Removed performance notes and automatic differentiation explanations
- Removed stress conversion formulas and notes
Simplified docstrings to essential formulas (strain energy, stress computation) and function signatures.
Removed verbose documentation sections:
- Removed detailed theory explanations and formulas from module docstring
- Removed type hierarchy and properties sections
- Removed usage examples from function docstrings
- Removed performance notes and implementation details
- Simplified docstrings to essential formulas (Hooke's law, elasticity tensor)
Kept core mathematical formulas and function signatures.
Removed verbose documentation sections:
- Removed detailed theory references (Simo & Hughes)
- Removed algorithm step-by-step explanations
- Removed performance notes and timing information
- Removed detailed field descriptions and invariants
- Simplified docstrings to essential information about fields and function signatures
Kept core information about multiplicative decomposition and material parameters.
Removed extensive documentation from material API docstrings:
- Removed detailed interface requirements sections
- Removed type hierarchy explanations
- Removed design philosophy sections
- Removed usage examples and 'See Also' references
- Kept only essential type and function descriptions
This simplifies the API documentation while maintaining core information
about material types and their behavior traits.
Remove demos/README.md as all demos and examples have been moved
to the website repository (juliafem.github.io) where they belong
as documentation rather than core library code.
- Move all demos/ and examples/ files to juliafem.github.io
- Update STRUCTURE.md to reflect examples/ and demos/ live in website
- Remove demos/README.md from main codebase
- Keep main repository focused on library code only
Update test_cantilever_regression.jl to use new DOF system, DOFManager,
and proper constraint elimination instead of old Physics struct API.
- Create elements using new @DOFSet and Element{K,P,S} API
- Use DOFManager for DOF allocation and management
- Apply forces using get_node_dofs API instead of NeumannBC
- Apply boundary conditions using constraint elimination (proper method)
- Solve reduced system (K_ff * u_f = f_f) instead of manipulating full matrix
- Reconstruct full solution from reduced solution
- Remove old Physics struct, DirichletBC, NeumannBC usage
- Update step numbering and comments for clarity
Add test includes for new architecture components: fields, physics,
element interpolation, materials, topology, and validation.
- Add LocalField type test
- Add fields test suite (test_local_field.jl)
- Add physics test suite (test_strain.jl)
- Add element interpolation test suite (test_interpolate_local_fields.jl)
- Add material test suites (state variables, traits, global cache, plasticity, workspace)
- Add topology test suites (segments, triangles, quadrilaterals, entities, etc.)
- Add validation test (test_plasticity_simple.jl)
- Comment out obsolete topology type extraction test
Replace tests for old Physics struct with tests for new physics
category types and trait-based dispatch.
- Remove tests for Physics struct, DirichletBC, NeumannBC, Constraint
- Add tests for Elasticity{Dim} and Thermal{Dim} category types
- Add tests for required_field_type trait function
- Test type stability and dispatch behavior
- Verify dimension-dependent field type mapping
Update test_reset_functions.jl to use get_stress/get_tangent getters
and relax allocation requirements for reset! function.
- Rename test set from MaterialStateCache to AssemblyMaterialWorkspace
- Update field access to use get_stress/get_tangent getter functions
- Relax allocation test for reset! (not in hot path, NamedTuple overhead acceptable)
- Add missing test_helpers.jl include
Update test_kernel_functions.jl to use get_tangent getter function
instead of direct field access to material_cache.𝔻, matching the
new AssemblyMaterialWorkspace API.
Update test_cache_updates.jl to support both legacy and new
GlobalMaterialCache API with proper allocation testing.
- Rename legacy test set to indicate it uses old API
- Update field access to use get_stress/get_tangent getter functions
- Relax allocation test for legacy API (may have NamedTuple overhead)
- Add new test set for GlobalMaterialCache API
- Add zero-allocation verification using BenchmarkTools
- Test both material_cache.σ/𝔻 access patterns
Replace monolithic Physics struct and boundary condition types with
simpler physics category types used for trait-based dispatch.
- Remove Physics struct (Formulation, Field, Mesh, Material coupling)
- Remove DirichletBC and NeumannBC boundary condition storage types
- Remove Constraint type
- Add Elasticity{Dim} and Thermal{Dim} physics category types
- Add required_field_type trait function for physics-to-field mapping
- Simplify to type tags for material trait dispatch
- Support compile-time field type inference from physics
- Align with new architecture: physics types are dispatch tags, not problem containers
Replace monolithic PlasticityState struct with compositional NamedTuple
state system and add new material trait system support.
- Remove PlasticityState struct (now using NamedTuple composition)
- Add supported_physics trait (Elasticity{3})
- Add required_state_variables trait (PlasticStrain, Backstress, EquivalentPlasticStrain)
- Change compute_stress to use NamedTuple for state_old and state_new
- Use get() with defaults for state extraction (handles empty initial state)
- Update state creation to use NamedTuple syntax (ε_p, α, κ)
- Remove state_type trait (replaced by required_state_variables)
Update LinearElastic to support new material trait system and
compositional state design with NamedTuple instead of Nothing.
- Add supported_physics trait (Elasticity{3})
- Add required_state_variables trait (empty tuple for stateless)
- Change state_old parameter to Union{Nothing,NamedTuple}
- Return NamedTuple() instead of nothing for state update
- Support compositional state system in compute_stress
Update AbstractMaterialState and EmptyState docstrings to reference
new AssemblyMaterialWorkspace and GlobalMaterialCache instead of
old MaterialStateCache API.
Add quantity type extraction trait for field types and optimize DOF
mapping function for zero-allocation performance.
- Add quantity_type trait function for field types
- Implement quantity_type for Displacement, Temperature, DisplacementRotation
- Add @inline to get_dof_mapping! for better inlining
- Optimize get_dof_mapping! loop to use indexed access instead of iterator
- Avoid iterator allocation in DOF mapping computation
Update JuliaFEM.jl to reflect major architectural changes: DOF system,
material workspace API, and microkernel architecture.
- Add DOF system includes (dofs.jl, dof_manager.jl, dof_connectivity.jl)
- Add material state variables and traits includes
- Add GlobalMaterialCache API exports
- Add DOF-based assembler includes and exports
- Add microkernel and formulation exports (evaluate, field_type_for_dispatch)
- Update MaterialStateCache → AssemblyMaterialWorkspace exports
- Add LocalField and strain extraction exports
- Add element interpolation and DOF extraction includes
- Update readers → io directory includes
- Remove/comment legacy includes (boundary_conditions.jl, assemble_v2.jl, etc.)
- Add DOF system exports (DOFSet, @DOFSet, DOFManager, etc.)
- Add element accessor exports (element_id, element_dofs, field_dof_range)
- Update physics exports (Elasticity, Thermal, required_field_type)
- Add material state variable exports (PlasticStrain, Backstress, etc.)
- Comment out contact problems (uses old Element API)
- Update documentation comments for new architecture
Major architectural refactoring: replace old field-based element system
with modern DOF-based design following Ciarlet's finite element triple.
- Redesign Element from Element{N,NIP,F,B} to Element{K,P,S,N}
- Replace connectivity with dof_indices (flat tuple of global DOF indices)
- Remove all old field system (sfields, dfields, fields, update_field!, etc.)
- Remove old constructors and compatibility shims
- Add compile-time DOF computation via @generated ndofs(K, S)
- Add field_dof_range for compile-time local DOF range computation
- Add local_to_global_map for type-stable DOF mapping
- Add topology_type, basis_type, dof_type query functions
- Add element_id, element_dofs, n_element_dofs, nnodes accessors
- Remove 593 lines of legacy code (963 → 370 lines)
- Simplify API: elements now store DOF indices, not node connectivity
- Support multi-field elements via DOFSet specifications
- Zero-allocation design with compile-time type information
Major refactoring: migrate material cache updates to new workspace API
and add support for GlobalMaterialCache alongside legacy API.
- Change all material_cache to material_workspace parameter names
- Remove M<:AbstractMaterialState type constraints from functions
- Change state handling from EmptyState() to NamedTuple()
- Change material_cache.σ/𝔻/states to material_workspace.fields/states
- Add new overloads for GlobalMaterialCache API (NEW API)
- Add zero-allocation optimizations using getfield and pre-allocated NamedTuples
- Add backward compatibility handling for legacy state types
- Update all function signatures and docstrings
- Add imports for GlobalMaterialCache and helper functions
- Add StatelessConstantTangent overload with pre-allocated NamedTuples
- Add StatelessStrainDependent and StatefulStrainDependent overloads for GlobalMaterialCache
- Keep legacy API overloads for Matrix{<:AbstractMaterialState}
Optimize update_geometry_cache! to eliminate iterator allocations and
update to use coords field.
- Add @inline to update_geometry_cache! for better inlining
- Change enumerate loop to indexed loop to avoid iterator allocation
- Add @inbounds annotation for bounds-check elimination
- Change ip.ξ to ip.coords to use new QuadraturePoint API
Optimize update_element_cache! to eliminate iterator and view allocations.
- Add @inline to update_element_cache! for better inlining
- Change enumerate loop to indexed loop to avoid iterator allocation
- Add @inbounds annotations for bounds-check elimination
- Remove @view allocation by passing array directly to get_dof_mapping!
Refactor continuum kernel to add atomic scalar computation and microkernel
interface while maintaining backward compatibility.
- Add compute_stiffness_value for atomic scalar K[k,l][α,β] computation
- Add compute_stiffness_block for D×D block building using atomic kernel
- Refactor compute_block_at_point to use compute_stiffness_block internally
- Add evaluate function for microkernel interface with field coupling dispatch
- Add basevec import for unit vector construction
- Add conditional imports for evaluate and get_tangent
- Update documentation to reflect new atomic operation structure
Major refactoring: replace material state cache with compositional
workspace using NamedTuple fields for better flexibility and zero-allocation.
- Rename MaterialStateCache to AssemblyMaterialWorkspace
- Change from M<:AbstractMaterialState to FieldType, StateType parameters
- Use AoS pattern: fields::Vector{FieldType} instead of separate σ, 𝔻 vectors
- Add zero-allocation field access via @generated functions
- Add extract_tangent! for type-stable zero-allocation tangent extraction
- Add get_tangent_vector, get_stress_vector helper functions
- Add @field_vector macro for compile-time field access
- Add get_stress, get_tangent, get_field accessor functions
- Update reset! to use create_zero_field and create_zero_state
- Update create_material_cache to use trait system for type inference
- Add backward compatibility alias create_assembly_workspace
- Add extensive documentation for zero-allocation usage patterns
Update kernel interface to use new material workspace API and improve
zero-allocation performance.
- Change ip.ξ to ip.coords in compute_element_stiffness!
- Add @inline to get_dof_mapping! for better inlining
- Optimize get_dof_mapping! to use indexed access instead of iteration
- Remove compute_node_contribution! function (unused)
- Update docstrings to use material_workspace instead of material_cache
- Update compute_all_blocks! example to use get_tangent_vector
- Add FIXME comment about removing B matrix computation functions
Update geometry cache to use coords field from QuadraturePoint
instead of deprecated ξ field.
- Change ip.ξ to ip.coords in create_geometry_cache function
- Update both gradient computation and detJ*w computation
Remove unused NodeCache code and simplify integration points call in
element cache creation.
- Remove NodeCache struct and create_node_cache function
- Simplify integration_points call to use auto-selected quadrature
- Remove integration_scheme parameter from create_element_cache
Refactor element-based COO assembler to use new material workspace API
and add support for GlobalMaterialCache alongside legacy API.
- Add new assemble_element! overload using GlobalMaterialCache (NEW API)
- Add new assemble! overload using GlobalMaterialCache (NEW API)
- Update legacy assemble_element! to use material_workspace and
𝔻_vec_buffer parameter
- Update legacy assemble! to use material_workspace and 𝔻_vec_buffer
- Change counter from Ref{Int} to Int for zero-allocation access
- Use extract_tangent! with pre-allocated buffer for zero-allocation
tangent extraction
- Add imports for GlobalMaterialCache and tangent extraction functions
- Update all function signatures and docstrings to reflect new API
Refactor COOCache to use AssemblyMaterialWorkspace and improve
zero-allocation performance.
- Add FieldType and StateType type parameters to COOCache struct
- Change material_cache to material_workspace field name
- Change counter from Ref{Int} to Int for zero-allocation access
- Add 𝔻_vec_buffer for pre-allocated tangent vector extraction
- Update constructor to infer FieldType and StateType from workspace
- Update reset! and extract_system to use direct counter access
- Add Tensors import and AssemblyMaterialWorkspace import
Add new DOF-by-DOF assembler and refactor material state cache type
parameters for better flexibility.
- Add DOFBasedCOOAssembler struct with comprehensive documentation
- Change AbstractMaterialStateCache from M<:AbstractMaterialState to
FieldType, StateType parameters
- Update material cache docstring to reflect new type structure
- Clarify purpose as per-element workspace vs persistent storage
Remove include statement for topology.jl which was deleted after
its content was moved to topology/api.jl.
- Remove include("topology/topology.jl") line
Major cleanup of quadrature module: simplify docstrings to minimal
code-focused style, remove all legacy symbol-based API, and fix Vec
constructor syntax throughout.
Documentation simplification:
- Remove all examples from docstrings
- Simplify file-level docstrings to one-line descriptions
- Remove verbose explanations, accuracy details, references, and
"See also" sections
- Keep only essential information about types and functions
Legacy code removal:
- Remove all legacy symbol-based API (Val{:GLTRI*}, Val{:GLTET*},
Val{:GLPYR*}, Val{:GLWED*}, Val{:GLSEG*}, Val{:GLQUAD*},
Val{:GLHEX*})
- Remove get_order() functions for legacy symbols
- Remove _legacy_tensor_product helper function
- Remove legacy symbol generation code from gl_tensor_product.jl
gauss.jl refactoring:
- Remove old Gauss{N} type and AbstractIntegration hierarchy
- Remove get_rule_name() function and all rule name mappings
(hundreds of lines for segments, triangles, quads, tets, hexes,
wedges, pyramids)
- Remove IntegrationPointNEW alias
- Replace with simple integration_points() function using new API
- Change npoints() signature to take only topology parameter
api.jl improvements:
- Uncomment and fix _infer_basis_order implementations
- Add proper type constraints (<:) for all topology types
- Add Triangle{10} case for cubic triangles
- Add _quadrature_topology_type mapping functions for all topologies
- Enable topology-only dispatch for default_quadrature
Vec constructor fixes:
- Fix all Vec{D} constructor calls to use tuple syntax
- Change Vec{2}(x, y) to Vec{2}((x, y))
- Change Vec{3}(x, y, z) to Vec{3}((x, y, z))
- Applied consistently across all quadrature rule files
Reorganize API documentation with organized sections for better
navigation and discoverability.
- Add structured sections: Elements, Physics, Fields, Formulations,
Materials, Mesh, Topology, Basis Functions, DOF System, Solvers
- Include key types and functions in each section
- Change title from "API documentation" to "API Reference"
- Keep index at the end for complete reference
Simplify documentation build script to generate API docs only.
Remove legacy code for generating examples, packages, and guides.
- Replace complex build system with simple Documenter.jl setup
- Generate markdown output for Quarto integration
- Build directly to juliafem.github.io/api/ directory
- Remove functions: copy_docs, add_page!, generate_* functions
- Use DocumenterMarkdown format instead of HTML
Remove documentation files from docs/ directory as they have
been moved to juliafem.github.io/ according to the website
structure. Manual documentation now lives in the website
repository directly.
Removed files:
- docs/book/adr-004-integration-points-api.md
- docs/book/adr-005-element-integration-point-indices.md
- docs/book/deformation_gradient_implementation.md
- docs/book/design/backend_transparent_architecture.md
- docs/book/design/gmres_algorithm_gpu.md
- docs/book/design/matrix_free_dirichlet_bc.md
- docs/book/element_assembly_implementation.md
- docs/book/linear_elastic_implementation.md
- docs/book/migration-guide-basis-api.md
- docs/book/neo_hookean_implementation.md
- docs/book/perfect_plasticity_implementation.md
Jukka Aho