Julia Code Generation

Generate Julia simulation scripts with DifferentialEquations.jl integration.

Features

DifferentialEquations.jl: Tsit5, Vern9, Rodas5, CVODE_BDF, RK4 solvers
Plots.jl visualization: Time histories, phase space portraits
CSV/DataFrame export: Via CSV.jl and DataFrames.jl
Energy tracking: Conservation monitoring from Lagrangian
Adaptive tolerances: Configurable abstol and reltol

Basic Usage

from mechanics_dsl.codegen.julia import JuliaGenerator
import sympy as sp

theta, g, l = sp.symbols('theta g l')

gen = JuliaGenerator(
    system_name="pendulum",
    coordinates=["theta"],
    parameters={"g": 9.81, "l": 1.0},
    initial_conditions={"theta": 0.5, "theta_dot": 0.0},
    equations={"theta_ddot": -g/l * sp.sin(theta)},
    solver="Tsit5",
    abstol=1e-8,
    reltol=1e-8
)
gen.generate("pendulum.jl")

Parameters

solver: ODE solver name (default: Tsit5)
abstol: Absolute tolerance (default: 1e-8)
reltol: Relative tolerance (default: 1e-8)

Supported solvers: Tsit5, Vern9, Rodas5, CVODE_BDF, RK4

Generated Code Structure

The generated Julia script includes:

Package imports: DifferentialEquations, Plots, CSV, DataFrames
Physical constants: Parameters as const globals
``equations_of_motion!``: In-place ODE function for DifferentialEquations.jl
``simulate()``: Configurable simulation with solver and tolerances
``plot_results()``: Time history subplots saved to PNG
``plot_phase_space()``: Phase portrait visualization
``export_csv()``: DataFrame-based CSV export
Energy function: compute_energy() from Lagrangian when available

Running the Generated Code

Install Julia dependencies first:

using Pkg
Pkg.add(["DifferentialEquations", "Plots", "CSV", "DataFrames"])

Then run:

julia pendulum.jl

Solver Selection Guide

Solver	Best For	Order
`Tsit5`	General non-stiff problems (default)	5th
`Vern9`	High-accuracy requirements	9th
`Rodas5`	Stiff systems	5th
`CVODE_BDF`	Very stiff / large systems	Variable
`RK4`	Fixed-step, educational	4th