physlang 0.1.0

Write math, run code. A DSL for mathematical physics.

PhysLang

Write math, run code.

PhysLang is a domain-specific language that lets you write mathematical notation directly and execute it as Python. No more translation errors between paper and code.

> let beta = 1.0
> let mu = U/2
> compute sum(n, 1, 100, n)
  = 5050
> compute integrate(sin(x), x, 0, pi)
  = 2.0

Installation

pip install physlang

That's it! The first time you run physlang, it automatically:

• Creates a desktop shortcut with the Ĥ (Hamiltonian) icon
• Adds PhysLang IDE to your applications menu

For Students

Just run these two commands:

pip install physlang
physlang

You'll see the Ĥ icon appear on your desktop. Double-click it anytime to open the IDE!

Manual Desktop Setup

If the automatic setup didn't work, you can run:

physlang --install     # Add desktop shortcut
physlang --uninstall   # Remove it

Optional: JAX Support

For GPU acceleration and automatic differentiation:

pip install physlang[jax]

Quick Start

Command Line

# Start interactive REPL
physlang

# Execute single expression
physlang -c "sum(n, 1, 100, n)"

# Execute file
physlang program.phi

# Show generated Python
physlang -p -c "integrate(x^2, x, 0, 1)"

# Use JAX backend
physlang -j -c "grad(f, 3.0)"

Web IDE

PhysLang includes a full web-based IDE with symbol palette, real-time translation, and code export:

# Open the IDE in your browser
physlang --ide

# Or start with a local server (for some browsers)
physlang --ide --serve

From Python:

from physlang import launch_ide
launch_ide()

The IDE features:

• Math Editor: Write with Unicode symbols (α, β, ∑, ∫, ², etc.)
• Symbol Palette: Click to insert Greek letters, operators, subscripts/superscripts
• Auto-conversion: Type \alpha → α, ^2 → ², \sum → ∑
• Real-time Translation: See Python/JAX code generated instantly
• Export: Download as .txt, .py, or JAX .py
• Works Offline: Fully self-contained, no server needed

Python API

from physlang import run, compile_to_python, compile_to_jax

# Execute directly
result = run("sum(n, 1, 100, n)")  # 5050

# Get Python code
code = compile_to_python("let f(x) = x^2\ncompute f(5)")
print(code)

# Get JAX code (GPU + autodiff)
jax_code = compile_to_jax("let f(x) = x^2\ncompute grad(f, 3.0)")

Typing Math

PhysLang accepts both Unicode math symbols and LaTeX notation. Type whichever is easier:

LaTeX	Unicode	Description
\alpha	a	Greek letters
\beta	b
\pi	p
\sum	``	Summation
\int	``	Integral
\infty	``	Infinity
^2	``	Superscripts
_1	``	Subscripts
\dagger	``	Hermitian conjugate
\leq	``	Less or equal
\RR	``	Real numbers

In the REPL, press Tab after typing LaTeX to convert it to Unicode.

Language Reference

Variables

let x = 5
let beta = 1.0
let mu = U/2

Functions

let f(x) = x^2 + 2*x + 1
let g(x, y) = x^2 + y^2
compute f(5)  # 36

Arithmetic

3 + 4 * 2       # 11
2^10            # 1024 (or 2 for Unicode)
sqrt(2)         # 1.414...
exp(1)          # 2.718...
log(e)          # 1.0
sin(pi/2)       # 1.0

Summation

# ASCII syntax
sum(n, 1, 100, n)           # 5050
sum(k, 1, 10, k^2)          # 385

# Unicode syntax (equivalent)
sum_{n=1}^{100} n           # 5050

Integration

# Definite integral
integrate(sin(x), x, 0, pi)     # 2.0
integrate(x^2, x, 0, 1)         # 0.333...

# Unicode syntax
_0^ sin(x) dx              # 2.0

Linear Algebra

let A = matrix([[1, 2], [3, 4]])
compute det(A)              # -2.0
compute trace(A)            # 5.0
compute norm(array(3, 4))   # 5.0
compute inv(A)              # inverse
compute eig(A)              # eigenvalues

Quantum Notation

# Hermitian conjugate
A^dagger                    # or A

# Commutator
comm(A, B)                  # [A, B] = AB - BA

# Anticommutator
anticomm(A, B)              # {A, B} = AB + BA

# Trace
Tr(rho * H)

# Tensor product
kron(A, B)                  # A x B

Operators (Quantum Field Theory)

operator c[i, sigma] : fermion
operator a[k] : boson
operator S[i] : spin

JAX Backend

Use -j flag or compile_to_jax() for:

• GPU acceleration
• Automatic differentiation
• JIT compilation

from physlang import compile_to_jax

code = compile_to_jax("""
let f(x) = x^3 + 2*x^2 + x
compute grad(f, 2.0)
""")
# Computes df/dx at x=2.0 using JAX autodiff

Built-in Functions

Math Functions

sin, cos, tan, exp, log, log10, sqrt, abs sinh, cosh, tanh, arcsin, arccos, arctan floor, ceil, round, min, max factorial, gamma

Linear Algebra

det, trace, norm, inv, eig, eigvals matrix_power, kron, outer, inner, cross

Array Operations

array, matrix, zeros, ones, eye, diag

Calculus

sum(i, start, end, expr) - Summation prod(i, start, end, expr) - Product integrate(expr, var, lower, upper) - Integration grad(f, x) - Gradient (JAX only) hessian(f) - Hessian matrix (JAX only)

Examples

Gauss Sum

compute sum(n, 1, 100, n)  # = 100*101/2 = 5050

Basel Problem

let s = sum(n, 1, 10000, 1/n^2)
compute s  # approximates pi^2/6

Gaussian Integral

compute integrate(exp(-x^2), x, -10, 10)  # approximates sqrt(pi)

Matrix Operations

let A = matrix([[1, 2], [3, 4]])
let B = matrix([[5, 6], [7, 8]])
compute A @ B               # matrix multiplication
compute det(A @ B)          # determinant of product

Gradient Descent (JAX)

let f(x) = (x - 3)^2
let x = 0.0
let lr = 0.1
# x = x - lr * grad(f, x)  # update step

File Extension

PhysLang files use .phi extension:

physlang program.phi

Why PhysLang?

Every physics paper has equations. Implementing them in code introduces bugs:

• mu = 0 instead of mu = U/2
• Wrong index ordering in tensor products
• Missing signs in commutators
• Incorrect Metropolis-Hastings ratios

PhysLang eliminates translation errors by letting you write the math directly.

Links

• PyPI: https://pypi.org/project/physlang/
• GitHub: https://github.com/dmjdxb/PhysLang

Contributing

Contributions welcome. See CONTRIBUTING.md.

License

Apache License 2.0