The math-native programming language

Write the math.
See the result.

Matrices are values. Plots are statements. Every computation measured in joules. Compiled to native speed.

Scroll

I · Write

Math is the syntax.

Multiply matrices with *. Transpose with '. Solve linear systems with \. JMax reads like the equations you already think in.

fit_model.jmax

II · See

Plot is a statement.

Type plot and see the chart. One word. The result renders inline.

plot line data

plot scatter data

plot bar results

plot heatmap correlation

Export to SVG, PNG, PDF, CSV, or JSON. Publish-ready from the REPL.

III · Four Ways In

Your way. One platform.

Write code, drag blocks, talk to AI, or type in a terminal. JMax compiles to native speed. Pick the way that fits how you think.

Visual Scripting

AI Coding

you: fit a linear model and plot the residuals

ai:

fn analyze(X: mat, y: vec):
  let beta = (X' * X) \ (X' * y)
  let residuals = y - X * beta
  plot scatter residuals

714 pJ consumed

Editor

1 fn main():

2 let A = [[2,1],[1,3]]

3 let b = [5,10]

4 let x = A \ b

5 plot bar x

▶ Run

Output

[1.0, 3.0]

412 pJ

Command Line

$ jmax

jmax> eig([[4,1],[1,3]])

[4.618, 2.382]

jmax> det([[1,2],[3,4]])

-2.0

jmax> softmax([1, 2, 3])

[0.09, 0.24, 0.67]

Energy: 89 pJ total

IV · Built In

Every domain. One language.

Linear algebra, statistics, signal processing, machine learning, symbolic math, finance, bioinformatics, and data analysis. All built in. No packages. No dependencies. Just write.

Linear Algebra

matmul, solve, eig, svd, qr, cholesky, inv, det, trace, rank

Statistics

mean, std, median, t_test, correlation, regression, percentile

Signal Processing

fft, ifft, convolve, fir_filter, spectrogram, hann, hamming

Machine Learning

softmax, relu, gelu, sigmoid, attention, cross_entropy

Symbolic Math

differentiate, integrate, find_root, sin, cos, exp, log

Finance

black_scholes, npv, irr, sharpe, sma, ema, max_drawdown

Bioinformatics

smith_waterman, needleman_wunsch, gc_content, translate

Data Analysis

dataframe, groupby, merge, pivot, describe, sort, filter

V · The Receipt

Know what it costs.

Every JMax program produces an energy receipt. You see exactly how many joules your computation consumed.

Energy Receipt

programfit_model.jmax

operations0

duration0 ms

avg power0 W

total energy0 pJ

SCI: 0 ISO 21031 CSRD Scope 3

VI · Try It

Start computing.

Open a terminal. Type an equation. See the answer.

Terminal

# Install $ cargo install jmax-cli # Start the REPL $ jmax jmax> 2 + 3 5 jmax> sqrt(2.0) 1.4142135623730951 jmax> [[1,2],[3,4]] * [[5,6],[7,8]] [[19, 22], [43, 50]]

Get JMax Read the Docs

Pure Rust REPL Inline Plots Energy Metered NxN Linear Algebra FFT Machine Learning DataFrames