Installation

SpectralBrain runs on Python 3.11–3.12.

From PyPI

pip install spectralbrain

The base install gives you the full spectral-geometry core (Laplace–Beltrami decomposition, all descriptors, distances), input-agnostic I/O, and the 2D/3D rendering core.

Optional extras

Install only what your study needs:

pip install "spectralbrain[bayesian]"   # PyMC, nutpie, NumPyro, BlackJAX, ArviZ
pip install "spectralbrain[viz]"        # vedo, scienceplots, hippunfold_plot, …
pip install "spectralbrain[gpu]"        # torch, CuPy, JAX (CUDA)
pip install "spectralbrain[neuro]"      # nilearn, dipy, pybids, templateflow, …
pip install "spectralbrain[full]"       # everything above

GPU is optional

The CPU backend (NumPy/SciPy) computes the Laplace–Beltrami eigenpairs and every descriptor on its own. The gpu extra accelerates large batches but is never required — nothing in the core depends on a CUDA device.

FreeSurfer is optional

SpectralBrain reads FreeSurfer and HippUnfold output but does not run them. If your surfaces already exist (lh.pial, hippocampal .surf.gii, …), you do not need FreeSurfer installed to analyze them.

Development install

git clone https://github.com/rdneuro/spectralbrain
cd spectralbrain
pip install -e ".[full]"