Skip to main content

​
From PyPI

Or with uv:
Prebuilt wheels bundle the compiled Rust extension, so no toolchain is needed on the supported platforms: Wheels are published for CPython 3.10 through 3.14.

​
Verify the install

The _lib import is the meaningful part: it confirms the native extension loaded, since generation runs in Rust with no Python fallback.

​
Runtime dependencies

pip install synforecast pulls in narwhals, numpy, pandas, polars, and pydantic. There are no optional extras — both pandas and polars are required, because output is materialized through narwhals rather than one fixed backend.

​
Optional dataframe engines

engine="pandas" (the default) and engine="polars" work out of the box. Three further engines are accepted but not installed by default:
engine="cudf" requires a RAPIDS cuDF build and a CUDA GPU; install it from the RAPIDS channel for your CUDA version.

​
Building from source

Installing from the source distribution compiles the Rust extension, which needs a stable Rust toolchain (install via rustup):

​
From a git checkout

For development, clone the repository and use the devenv target, which runs uv sync --dev and installs the pre-commit hooks:
See CONTRIBUTING for the test and lint workflow.
Alpha versioning SynForecast is in alpha. APIs and seed-identical outputs may change between releases, so pin a version if you depend on exact generated values.

​
Next

  • Quick start — generate a panel and control the generators.
  • Balanced pool — what generate_series draws from by default.