Skip to main content
Scientific computing · math-native

JMax.
The math-native programming language.

Write math, see the result, and measure the energy it took. One environment for symbolic algebra, simulation, and visualization, compiled to run anywhere and metered in joules. Energy-optimized scientific computing for the AI and compute era.

symbolic + numeric + sim
joules measured
wasm·gpu runs anywhere
jmax · session ⚡ 0.33 J
  • > ∂u/∂t = α ∇²u · heat on a plate
  • differentiate x² · sin x symbolic 0.000 J
  • assemble FEM operator sparse 0.02 J
  • solve A·u = b (CG) numeric 0.31 J
  • render field → SVG viz 0.001 J
Symbolic stays cheap. You pay for the solve.

Math in. Result, plot, and joules out.

The language

Symbolic, numeric, and seen

One environment from the equation to the simulation to the plot.

Math-native syntax

Write mathematics directly. A computer algebra system simplifies and differentiates by construction, so symbolic work stays exact and nearly free.

Simulation built in

Native FEM and PDE solvers with sparse linear algebra. Take a model from equation to result without leaving the language.

Visualization is the output

Results render to vector graphics and WebGPU from one scene description. Seeing the answer is part of computing it.

The discipline

Scientific computing, on a SWaP-2C budget

Simulation and data science have always burned compute without counting it. JMax puts scientific computing on a SWaP-2C budget, key to unlocking the AI and compute revolution.

Size

One environment instead of a stack of separate tools.

Weight

No glue code or data copies between symbolic, numeric, and plotting tools.

Power

Every computation metered in joules, symbolic to solve.

Cost

Cheap symbolic steps stay cheap; you pay for the solve, not the setup.

Cooling

Less wasted computation means less heat per result.

Write math. Measure energy.

charlot-lang.dev →