Backends

livn supports three simulation backends, each providing a full implementation of the Env protocol. Select a backend via the LIVN_BACKEND environment variable before importing livn:

export LIVN_BACKEND=brian2     # default
export LIVN_BACKEND=diffrax    # requires livn[diffrax] dependencies
export LIVN_BACKEND=neuron     # requires livn[neuron] dependencies and MPI

All backends share the same user-facing API - you write your simulation code once and switch backends by changing the environment variable.

brian2 (default)

brian2 is the default backend, suitable for rapid prototyping and small-to-medium systems. It models neurons as point processes using brian2's equation-based description language. No additional dependencies are required beyond the base install.

Strengths:

• Fast setup, no external system libraries needed
• Good for systems up to ~1,000 neurons on a single machine
• Includes Izhikevich and Leaky Integrate-and-Fire (LIF) models

# brian2 is the default, just import and go
from livn import make

env = make("EI2")
it, t, iv, v, *_ = env.run(100)

Jax

A JAX-based backend that enables differentiable simulations through Diffrax and Equinox. This allows you to compute exact gradients through the simulation and use gradient-based optimization to learn stimulus parameters, decode neural activity, or train surrogate models end-to-end.

Strengths:

• End-to-end differentiable: backpropagate through the full simulation
• GPU-accelerated via JAX
• JIT-compiled for fast repeated evaluation

Install with:

pip install livn[diffrax]

import os
os.environ["LIVN_BACKEND"] = "diffrax"

from livn import make
import jax

env = make("EI2")
# Gradients through the simulation are now available

See the Differentiable Simulation example for a full walkthrough.

NEURON

The NEURON backend provides high-fidelity, multi-compartment biophysical simulations with MPI-based parallelism. It integrates with the MiV-Simulator for large-scale network simulations on HPC infrastructure.

Strengths:

• Detailed biophysical neuron models (multi-compartment, ion channels, calcium dynamics)
• MPI-parallel: scales to millions of neurons on supercomputers
• Best choice for generating realistic synthetic data

Requires system-level MPI and HDF5 libraries. See Installation for setup instructions.

pip install livn[neuron]

import os
os.environ["LIVN_BACKEND"] = "neuron"

from livn import make

env = make("EI2")

Comparison

Feature	brian2	Diffrax	NEURON
Differentiable	No	Yes	No
GPU support	No	Yes	No
Multi-compartment models	No	Yes	Yes
MPI parallelism	No	No	Yes
Setup complexity	Low	Medium	High
Ideal scale	≤1,000 neurons	≤10,000 neurons	≤millions