Inferix: An (experimental) unified interface for probabilistic inference in JAX + Equinox
| Inferix | |
|---|---|
| Author | Gary Allen |
| Homepage | github.com/gvcallen/inferix |
NB: This library is in very early stages, and is likely not currently functional.
Installation
Inferix can be installed using pip directly:
pip install inferix
Motivation
In the JAX ecosystem, you typically have to choose between two extremes for Bayesian inference:
- Lower-level drivers (like BlackJAX), where you have to manually manage while-loops, PRNG keys, buffers, and algorithmic states.
- High-level Probabilistic Programming Languages (PPLs) (like NumPyro or jaxns), which are user-friendly but requiring rewriting your model using their specific domain-specific languages and distribution primitives.
The goal of Inferix is to be a middle option that mirrors the API of Optimistix. It is designed for engineers and scientists who already have a forward likelihood and prior model written in pure JAX (perhaps using Equinox), and just want to sample from it without managing boilerplate or adopting a heavy PPL framework.
Inferix wraps low-level algorithms in a unified interface (inferix.mcmc_sample or inferix.nested_sample) and handles any XLA-compiled control flow, reparameterizations and data packaging. Current kernels include JAX-native NUTS and Nested Slice Sampling (via BlackJAX).
import jax
import jax.numpy as jnp
import inferix
# 1. Define your target functions (Pure JAX)
def my_likelihood(theta, args):
# e.g., A physics-based model
return ...
def my_prior_transform(u, args):
# A mapping from the uniform unit hypercube coordinates u to physical parameters theta
return ...
# 2. Instantiate your sampler of choice e.g. inferix.NUTS or inferix.NSS
sampler = inferix.NSS(num_delete=10, num_inner_steps=20, logZ_convergence=1e-3)
# 3. Execute the run
key = jax.random.key(42)
result = inferix.nested_sample(
log_likelihood_fn=my_likelihood,
prior_transform_fn=my_prior_transform,
sampler=sampler,
key=key,
)
# 4. Access the results
print(f"Final log-Evidence (logZ): {result.logZ} ± {result.logZ_err}")
print(f"Num samples: {result.samples.shape[0]}")
Metadata
Release files for inferix 0.1.10
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| inferix-0.1.10.tar.gz | 3.3 MB | With provenance | Details for inferix-0.1.10.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| inferix-0.1.10-py3-none-any.whl | Python 3 | none | any | Details for inferix-0.1.10-py3-none-any.whl |
Total release size: 3.3 MB
Release files / inferix-0.1.10.tar.gz
| Download URL | inferix-0.1.10.tar.gz |
|---|---|
| Size | 3.3 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
f0023a7ad97e9c976de452cf7fbe6bc1aaa7121d5e853a0e8ce33cad2032575e
|
|
BLAKE2b-256 checksum How to use checksums |
25d1951b7a9e92b88371627f9b8d0a9b0be88bf0e760cddd8c2e635e75c28aa6
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Apr 23, 2026.
Transparency logRelease files / inferix-0.1.10-py3-none-any.whl
| Download URL | inferix-0.1.10-py3-none-any.whl |
|---|---|
| Size | 15.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
d2b4b3277b155dac26794e1bcf5571b4f09c40d8dffd39c9a266fdb84fad7482
|
|
BLAKE2b-256 checksum How to use checksums |
c5b82174f8dc4807272a493852cb81c69282059c804a77948ebb6c5dce0d0ac9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.12
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Apr 23, 2026.
Transparency logSecurity for inferix 0.1.10
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for inferix-0.1.10.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Apr 23, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.