Start · 01

Install¶

TSDynamics requires Python ≥ 3.12.

pip install tsdynamics

or, with uv:

uv add tsdynamics

The C compiler requirement¶

Continuous and delay systems are compiled to native code by JiTCODE / JiTCDDE, which generate C from your symbolic equations and build it with your platform's compiler. A working C toolchain must be on the path:

LinuxmacOSWindows

sudo apt-get install build-essential python3-dev

(or the equivalent gcc + Python headers package for your distribution)

xcode-select --install

Install the MSVC Build Tools (the "Desktop development with C++" workload is sufficient).

Discrete maps do not need the C toolchain — they are JIT-compiled by Numba, which ships its own backend.

First call is slow, every later call is fast

The first integrate() on a system class compiles a shared library and caches it under ~/.cache/tsdynamics/. Subsequent calls — even in new Python sessions — load the cached binary. See the compilation pipeline for details.

Optional extras¶

Extra	Installs	When you want it
`tsdynamics[plot]`	`matplotlib`	Plotting trajectories and diagrams in the examples
`tsdynamics[diffsol]`	`pydiffsol`	Experimental Rust-backed ODE solver backend

pip install "tsdynamics[plot]"

Verify the install¶

import tsdynamics as ts

print(ts.__version__)
print(ts.registry.families())   # {'ode': 118, 'dde': 5, 'map': 26}

traj = ts.Henon().iterate(steps=100)   # no C compiler needed for maps
print(traj)                            # Trajectory(n_steps=100, dim=2, ...)

If ts.Lorenz().integrate(final_time=1.0) also succeeds, the C toolchain is correctly set up.

Next¶

02 · First trajectory — a complete worked example across all three system families.