You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
While 0.4.0 is being finished on the release/0.4.0-rc branch (#114), this roadmap outlines the vision for FEAScript 0.5.0. The primary focus of this release is solving larger problems fast: sparse iterative solvers (moved from 0.4.0), multigrid and worker threads. Version 0.5.0 also opens FEAScript to meshes from other tools, builds meshes from several structured blocks, adds physical units with a dimensional check before solving, and brings the API closer to a stable 1.0, with one config object that describes a whole model as plain JSON.
Iterative solver robustness: The checks that 0.4.0 gave the Jacobi solver, a relative residual tolerance of 1e-8, a NOT_CONVERGED error instead of the last iterate, the run report and regression tests against the direct solvers, applied to every new iterative solver (Iterative solver robustness #119).
Undetermined systems: Detect a model whose boundary conditions leave the solution undetermined and stop it with a SINGULAR_MATRIX error. In 0.4.0, a heat conduction model with zero flux on every boundary (only "symmetry" or no conditions) and a heat source returns values of about 1e14 without an error, since the LU solver does not flag the singular matrix. Check for a Dirichlet or Robin condition before solving, or detect the zero pivot, and add a regression test (Stop undetermined models with a SINGULAR_MATRIX error #137).
Worker pool: Parallel assembly and solves on several threads. Data is shared through SharedArrayBuffer on cross-origin isolated pages and passed as messages otherwise. Assembly and the solvers talk to the threads through one small interface, so one thread and many threads run the same code (Add a worker pool for parallel assembly and solves #139).
One asynchronous solve: await model.solve({ threads: "auto" }) replaces solve() and solveAsync(), in the browser, in Node.js and in FEAScriptWorker. The old forms keep working, with a deprecation warning (One asynchronous solve: await model.solve({ threads }) #140).
Batch runs: Solve one model for a list of parameter values, sharing the work between threads, for parameter sweeps. Batch runs need no SharedArrayBuffer, so they use every thread on any page, including CodePen (Add batch runs for parameter sweeps #141).
Meshing
Multi-block meshes: Build a 2D mesh from several structured blocks, each a quadrilateral given by its four corners, with linear or quadratic elements. Blocks meet edge to edge with the same number of elements along a shared edge, and their shared nodes are merged (Build 2D meshes from several structured blocks #142).
Named block boundaries: Each block names its outer edges, e.g. "inlet" or "wall". An edge without a name gets a default one (Name the outer edges of mesh blocks #143).
Same structure as imported meshes: A generated mesh uses the common mesh structure of the readers, so every model, solver and plot works on it as on an imported mesh (Generated meshes use the common mesh structure #144).
Quadrilaterals on polygons: A single quadrilateral is meshed with the transfinite map the structured mesher already uses, and several quadrilateral blocks sharing edges are the multi-block mesher above. Splitting an arbitrary simple polygon into quadrilaterals automatically is where this stops: a trapezoidal decomposition (vertical sweep) works for simple polygons but gives slivers near slanted edges, so it may come later as an option, and triangles cover the general case. Mixed meshes stay unsupported (Mesh quadrilateral polygons with quadrilaterals #146).
Corner triangles in flow models (to investigate): The triangle generator can leave a triangle in a corner of the domain with two sides on the boundary. With Taylor-Hood elements, five of its six velocity nodes are then fixed, and the flow near that corner comes out wrong. In the lid-driven cavity (rectangle: [4, 2], minAngle: 30), sizes 0.35, 0.3 and 0.15 give such triangles and the strongest return flow moves into a lid corner (u = -0.35 instead of -0.32), while sizes 0.5, 0.4 and 0.2 give none. Investigate splitting these corner triangles in the generator, or warning about them, and add a regression test (Corner triangles give wrong flow with Taylor-Hood elements #147).
Earlier mesh settings: numElementsX, maxX, numElementsY, maxY and the boundaries "0" to "3", deprecated in 0.4.0, keep working until 0.7.0 at the earliest, so existing scripts and pens still run (Keep the deprecated mesh settings working until 0.7.0 #148).
Working with Other Tools
Common mesh structure: Every reader returns the mesh structure of 0.4.0, with nodes, elements and named boundaries, extended with node and element sets (Common mesh structure for all readers #149).
Ready for 3D: The common mesh structure does not assume two dimensions. Nodes have x, y and z coordinates, elements carry their own dimension and type, and boundaries can be made of faces as well as edges, so 3D elements (tetrahedra and hexahedra) can be added before 1.0 without changing it. 0.5.0 still solves 1D and 2D models only (Make the common mesh structure ready for 3D #150).
Import report: Unsupported keywords and element types are listed in a structured report, never dropped silently (Add a structured import report #155).
Units
Unit input: model.setUnits({ length: "mm", temperature: "C", pressure: "Pa" }). Only the quantities that are named change, and anything left out uses its SI default, so a model without units gives the same numbers as today (Add unit input: model.setUnits() #156).
Conversion at the edges of the core: Inputs are converted to SI, the model is solved in SI, and results come back in the units of the user. The physics and the solvers do not change. Coefficient functions keep receiving x and y in the units of the user (Convert units at the edges of the core #157).
Dimensional check: Every unit is reduced to its powers of mass, length, time and temperature, and compared with what the model expects. A mismatch stops the model before solving with a FEAScriptError, e.g. thermalConductivity: "W/m2" has the wrong dimension, expected W/(m*K) (Check the dimensions of every unit before solving #158).
Plot grid: The density of the contour plot follows the size of the mesh instead of a fixed number of points per length unit, so a mesh gives the same picture in m and in mm (Base the contour plot grid on the mesh size #160).
Faster interpolated plots: plotInterpolatedSolution and the VTK plots locate every point of their grid in the mesh again for each plot, which takes most of the time of a page that shows several fields. The grid points will be located once per mesh, as an element and local coordinates, and reused for every field and plot, and the boundary segments will be indexed for the inside test, so the fields of one result plot for about the cost of one (Locate plot grid points once per mesh #162).
API
Config-object model: One object defines a model. The builder methods stay as a shorter way to write it. The object is plain JSON, so a model can be saved, shared as a link and passed to other tools (Define a model with one config object #163).
Formula coefficients: Coefficients given as text, e.g. "if(x < 2, 10, 1)", evaluated without eval, so models in JSON can still have varying coefficients. JavaScript functions keep working (Accept coefficients as formulas without eval #164).
Config objects and formulas: A model written as a config object gives the same result as with the builder methods, and a coefficient given as a formula the same result as the same JavaScript function (Tests for config objects and formula coefficients #172).
Around the Core
FEAScript-mcp: An MCP server, so AI assistants such as Claude Desktop can set up and solve models with FEAScript on your own machine. It has its own repository and versions, and will use the config object of 0.5.0 (FEAScript/FEAScript-mcp#1).
Please feel free to
Propose new features or optimizations
Express interest in contributing to any of the above tasks
Help define priorities for the 0.5.0 release
Share which file formats and units matter most to you
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
While 0.4.0 is being finished on the
release/0.4.0-rcbranch (#114), this roadmap outlines the vision for FEAScript 0.5.0. The primary focus of this release is solving larger problems fast: sparse iterative solvers (moved from 0.4.0), multigrid and worker threads. Version 0.5.0 also opens FEAScript to meshes from other tools, builds meshes from several structured blocks, adds physical units with a dimensional check before solving, and brings the API closer to a stable 1.0, with one config object that describes a whole model as plain JSON.Numerical Solution & Performance
1e-8, aNOT_CONVERGEDerror instead of the last iterate, the run report and regression tests against the direct solvers, applied to every new iterative solver (Iterative solver robustness #119).SINGULAR_MATRIXerror. In 0.4.0, a heat conduction model with zero flux on every boundary (only"symmetry"or no conditions) and a heat source returns values of about 1e14 without an error, since the LU solver does not flag the singular matrix. Check for a Dirichlet or Robin condition before solving, or detect the zero pivot, and add a regression test (Stop undetermined models with a SINGULAR_MATRIX error #137).SharedArrayBufferon cross-origin isolated pages and passed as messages otherwise. Assembly and the solvers talk to the threads through one small interface, so one thread and many threads run the same code (Add a worker pool for parallel assembly and solves #139).await model.solve({ threads: "auto" })replacessolve()andsolveAsync(), in the browser, in Node.js and inFEAScriptWorker. The old forms keep working, with a deprecation warning (One asynchronous solve: await model.solve({ threads }) #140).SharedArrayBuffer, so they use every thread on any page, including CodePen (Add batch runs for parameter sweeps #141).Meshing
"inlet"or"wall". An edge without a name gets a default one (Name the outer edges of mesh blocks #143).blockskey ofsetGeometry(), beside the singleshapeof 0.4.0, with the boundary names that 0.4.0 introduced (Accept mesh blocks under a blocks key of setGeometry() #145).rectangle: [4, 2],minAngle: 30), sizes 0.35, 0.3 and 0.15 give such triangles and the strongest return flow moves into a lid corner (u = -0.35 instead of -0.32), while sizes 0.5, 0.4 and 0.2 give none. Investigate splitting these corner triangles in the generator, or warning about them, and add a regression test (Corner triangles give wrong flow with Taylor-Hood elements #147).numElementsX,maxX,numElementsY,maxYand the boundaries"0"to"3", deprecated in 0.4.0, keep working until 0.7.0 at the earliest, so existing scripts and pens still run (Keep the deprecated mesh settings working until 0.7.0 #148).Working with Other Tools
await importMesh(file, { format: "inp" }).importGmshMeshstays as an alias (Add an importer and exporter interface: importMesh() #151)..inpmesh import (Abaqus and CalculiX): Written from the published keyword documentation, with test files that we build ourselves (Add .inp mesh import (Abaqus and CalculiX) #152)..vtufiles (Add VTU import #153).Units
model.setUnits({ length: "mm", temperature: "C", pressure: "Pa" }). Only the quantities that are named change, and anything left out uses its SI default, so a model without units gives the same numbers as today (Add unit input: model.setUnits() #156).xandyin the units of the user (Convert units at the edges of the core #157).FEAScriptError, e.g.thermalConductivity: "W/m2" has the wrong dimension, expected W/(m*K)(Check the dimensions of every unit before solving #158).resultreports the units in use, and plots and exports show them (Report units in results, plots and exports #159).Visualization
plotInterpolatedSolutionand the VTK plots locate every point of their grid in the mesh again for each plot, which takes most of the time of a page that shows several fields. The grid points will be located once per mesh, as an element and local coordinates, and reused for every field and plot, and the boundary segments will be indexed for the inside test, so the fields of one result plot for about the cost of one (Locate plot grid points once per mesh #162).API
"if(x < 2, 10, 1)", evaluated withouteval, so models in JSON can still have varying coefficients. JavaScript functions keep working (Accept coefficients as formulas without eval #164).Resultshape: The same fields for every model and every solver (One Result shape for every model and solver #165)."inlet", as 0.4.0 does for geometries and Gmsh meshes (Key boundaries of imported and multi-block meshes by name #166).Verification & Benchmarks
.msh,.inpand.vtugives the same solution (Cross-format tests for the importers #170).Around the Core
Please feel free to
All reactions