ParamRF, or pmrf, is an open-source radio frequency (RF) modeling framework. It provides a declarative syntax for creating RF circuit and surrogate models using JAX.

The library provides tools for model simulation, optimization, fitting, statistical analysis, and Bayesian inference.

Version:
Author:: Gary Allen
GitHub:: https://github.com/gvcallen/paramrf
Docs:: https://gvcallen.github.io/paramrf
Paper:: https://doi.org/10.48550/arXiv.2510.15881

Features

Declarative syntax: Models can be composed and defined using a self-documenting, object-oriented syntax.
Hierarchical modeling: By nesting models within models, it is easy to create deep, hierarchical structures.
Differentiable: Since the library is built on jax (as opposed to numpy), derivatives are available via auto-differentiation, enabling faster performance and new design opportunities.
Hardware flexible: Functions are compiled just-in-time (JIT), reducing overhead and allowing computation on high-performance hardware (CPU, GPU, TPU).
Extensible: Power users can easily add additional models and algorithms by extending the library’s built-in classes and interfaces.

Installation

ParamRF can be installed directly using pip (requires Python 3.11+):

$ pip install paramrf

Example

The code below demonstrate how to define and optimize an RLC model to satisfy a given goal function. See the documentation for more examples.

import pmrf as prf
from pmrf.models import Resistor, Inductor, Capacitor

model = Resistor(50) ** Inductor(1.0e-9) ** Capacitor(1.0e-12)
goal = prf.evaluators.Goal('s11_db', '<', -20)
passband = prf.Frequency(3, 4, 101, 'GHz')

result = prf.optimize.minimize(goal, model, passband, solver=prf.optimize.NelderMead())

plot_freq = prf.Frequency(1, 6, 101, 'GHz')
model.plot_s_db(plot_freq, m=0, n=0, label='initial')
result.model.plot_s_db(plot_freq, m=0, n=0, label='optimized')

Next steps

For an overview of the library’s features, see the examples page.
For step-by-step guides that you can follow, check out the tutorials.
To delve a bit deeper into the library’s core building blocks and philosophy, head off to core concepts.

Optional dependencies

Several additional dependencies are required/recommended for more advanced use-cases.

For Bayesian inference, you may need this fork of distreqx:

$ pip install git+https://github.com/gvcallen/distreqx

For BlackJAX’s Bayesian solvers:

$ pip install git+https://github.com/handley-lab/blackjax.git@v0.1.0-beta

For the PolyChord solver:

$ pip install git+https://github.com/PolyChord/PolyChordLite.git anesthetic mpi4py

Development Warning

ParamRF is an active research project with an evolving API. Although a large portion of the API has stabilized, until V1.0.0 is reached, minor version bumps (0.X.0) will indicate breaking changes, while patch versions (0.x.Y) will imply new features and/or bug fixes.

Citation

If you have used ParamRF for academic work, please cite the arXiv preprint (https://doi.org/10.48550/arXiv.2510.15881) as:

G.V.C. Allen, D.I.L. de Villiers, (2025). ParamRF: A JAX-native Framework for Declarative Circuit Modelling. arXiv, https://doi.org/10.48550/arXiv.2510.15881.

or using the BibTeX:

@article{paramrf,
   doi = {10.48550/arXiv.2510.15881},
   url = {https://doi.org/10.48550/arXiv.2510.15881},
   year = {2025},
   month = {Oct},
   title = {ParamRF: A JAX-native Framework for Declarative Circuit Modelling},
   author = {Gary V. C. Allen and Dirk I. L. de Villiers},
   eprint = {2510.15881},
   archivePrefix = {arXiv},
   primaryClass = {cs.OH},
}