English · Español
Type a function, pick a method, and NumLab shows every iteration, the plot, the error
and a step-by-step derivation with the numbers plugged in — exactly as you would write it on an exam.
Plus a symbolic CAS for everything in between.
🌐 Try it in your browser · ⬇️ Download for desktop · 🛠️ Develop
| 58 pages | Symbolic CAS | 906 tests | 3 platforms | 2 languages |
|---|---|---|---|---|
| a full numerical methods course | derivatives · integrals · limits · solve | against textbook examples | macOS · Windows · Linux | English · Spanish |
Works offline, runs locally. Every computation happens in your browser or in the desktop app: no server, no accounts, no tracking.
- Exam-style step by step. Every method renders its formulas with KaTeX and plugs the actual numbers into the first iterations: Gaussian elimination multipliers, mantissa bits, divided differences, Runge–Kutta stages…
- A workspace like TI-Nspire or GeoGebra. Pages with a calculator, graphs (sliders, points, implicit and
parametric curves, pan and zoom), a spreadsheet (formulas, named columns as lists,
linreg, paste from Excel) and notes with TeX — all sharing the same variables. Save and open documents as.numlab.json. - A real CAS inside. Exact derivatives, antiderivatives and definite integrals, limits, Taylor series, simplification, factoring, partial fractions, equation and system solving, and matrix algebra — and the same symbolic engine powers the methods (exact Newton derivatives, exact Taylor coefficients, exact reference integrals).
- Faithful to the course. Notation, stopping criteria and method variants follow the course textbook; its worked examples are one click away ("Ex. 2.8", "Ex. 4.12"…) and reproduce its tables digit by digit — including the 4-digit calculator when an example uses one.
- Plots that don't lie. Adaptive sampling never misses narrow peaks or oscillations, and it breaks the curve at asymptotes and jumps instead of drawing fake vertical lines.
- Error and convergence analysis. Estimated order of convergence, theoretical bounds vs. actual error, significant digits and condition numbers.
- Scilab export. Every page generates a ready-to-run
.scescript with your data; tables export to CSV. - Pleasant to use. Symbol keyboard, built-in help (
?), instant search (⌘K/Ctrl+K), light and dark themes, mobile-friendly layout, and an English/Spanish interface (auto-detected, switch anytime).
A notebook-style console with exact symbolic results (rendered in TeX), plus a function plotter that marks roots, extrema, double roots and intersections.
| Category | Commands |
|---|---|
| Calculus | diff(f, x, n) · integrate(f, x) · integrate(f, x, a, b) (also ±∞) · limit(f, x, a) · taylor(f, x, x0, n) · sum(f, k, a, b) |
| Algebra | simplify · expand · factor · partfrac · solve(eq, x) · solve_system(eq1, eq2, …) · roots(f, a, b) |
| Matrices | A = [1 2; 3 4] · det · inv · A' · rref · lu · lusolve(A, b) · eig · cond · norm(A, p) |
| Everything else | variables and user functions (f(x) = x^3 - 2x - 5), N(expr), plot(f, g, a, b), Scilab syntax (%pi, [1 2; 3 4]) |
Spanish command names (derivada, integrar, resolver, graficar…) work too.
 Workspace · calculator, graphs, spreadsheet and notes sharing variables |
|
 Home · course topics and exam countdown |
 Symbolic calculator (CAS) · exact results in TeX |
 Newton–Raphson · iterations, tangents and order of convergence |
 Gaussian elimination · pivoting and the augmented matrix, step by step |
 16-bit binary machine · sign, exponent and mantissa |
 Simpson's rule · exact symbolic reference and error bound |
 Cubic splines · natural and clamped |
 Runge–Kutta · stages, exact solution and global error |
 Function plotter · roots, extrema and intersections |
|
| Topic | Methods |
|---|---|
| 1 · Number representation and errors | Base conversion · 16-bit binary machine · IEEE 754 · Floating-point system F(β, t, L, U) · Machine epsilon · Absolute and relative error · 7-digit decimal machine · Error propagation · Catastrophic cancellation · Condition number · Numerical stability · Taylor series |
| 2 · Nonlinear equations | Bisection · Fixed point · Aitken (Δ²) · Steffensen · Newton–Raphson · Secant · False position · Modified Newton (multiple roots) · Comparison |
| 3 · Systems of linear equations | Gaussian elimination · Thomas algorithm · LU factorization · Jacobi · Gauss–Seidel and SOR · Condition number · Newton for systems · Fixed point for systems · Power method |
| 4 · Interpolation | Lagrange · Divided differences · Finite differences · Error estimate · Runge phenomenon · Cubic splines |
| 5 · Numerical differentiation and integration | Finite differences · Trapezoidal rule · Simpson's rule · Newton–Cotes · Romberg–Richardson · Gauss–Legendre · Double integrals · Comparison |
| 6 · Ordinary differential equations | Euler · Midpoint · Trapezoidal / modified Euler · Implicit trapezoidal · Adams–Moulton · Taylor · Runge–Kutta · First-order systems · Higher-order equations · Comparison · Pursuit missile in R³ |
| Tools | Workspace (calculator, graphs, spreadsheet, notes) · Symbolic calculator (CAS) · Function plotter |
Get the installer for your system from Releases:
| System | File |
|---|---|
| macOS (Apple Silicon and Intel) | NumLab-x.y.z-mac-arm64.dmg · NumLab-x.y.z-mac-x64.dmg |
| Windows | NumLab-x.y.z-win-x64.exe |
| Linux | NumLab-x.y.z-linux-*.AppImage · NumLab-x.y.z-linux-*.deb |
Note
The apps are not code-signed yet.
macOS: the first time, right-click the app → Open (or run xattr -cr /Applications/NumLab.app).
Windows: if SmartScreen appears, click More info → Run anyway.
Don't want to install anything? The web version is the same app.
Correctness comes first:
pnpm test| Test | What it checks |
|---|---|
test:metodos |
906 assertions reproducing the worked examples of all six chapters of the course textbook, Burden & Faires tables and closed forms. When a book contains an arithmetic slip, the app uses the correct value and the test documents it. |
test:graficas |
That the plotted curve stays close to the true function: narrow peaks, oscillations, asymptotes and jumps. |
test:tex |
That no formula loses its backslashes inside the code (\;, \frac…). |
Requirements: Node.js 22+ and pnpm.
pnpm install
pnpm dev # web version at http://localhost:5173
pnpm desktop # build and open the desktop app (Electron)
pnpm dist # build the installer for your system into release/
pnpm capturas # regenerate the icon, banners and screenshots in docs/src/
modules/<topic>/ one module per topic: pure algorithms, pages, theory and index.ts (registry)
components/ shared UI (MethodPage, DataTable, ScilabCode, Plot, help, symbol keyboard…)
lib/ expressions (mathjs), number formatting, plot sampling
i18n.ts English/Spanish: L('texto', 'text')
config.ts course data (name, exam dates, repository)
electron/ desktop app main process
scripts/ tests and image generation
.github/workflows/ CI, web demo (GitHub Pages) and desktop installers
Using it for your own course? Edit src/config.ts. Leave the exam dates empty to hide the
countdown.
Releasing: push a vX.Y.Z tag (matching version in package.json). GitHub Actions builds the installers for
all three platforms and attaches them to a draft release.
Contributions are welcome: numerical bug reports, new methods, UI improvements and translations. See CONTRIBUTING.md.
MIT © 2026 Fernando Daza and contributors.
Built with React, TypeScript, Vite, mathjs, nerdamer, KaTeX, Plotly and Electron.