Citing neuraLQX¶

neuraLQX would not exist without NetKet. If you got results with neuraLQX, then NetKet is part of the scientific and engineering chain that produced those results: sampling, VMC drivers, SR/QGT machinery, operator interfaces, logging, distributed computing integration patterns, and the surrounding ecosystem.

You have to always cite NetKet whenever you cite neuraLQX!

This page gives ready-to-use citation text and BibTeX snippets for common usage scenarios.

Important

If you publish work that used neuraLQX, you have to cite neuraLQX and you absolutely have to cite NetKet at minimum. If you used MPI-distributed runs with older neuraLQX versions, also cite mpi4jax. You should additionally cite JAX (and Flax) as it is the main workhorse behind everything.

What should I cite?¶

Use one of the bundles below as a default. Pick the one that matches how you ran neuraLQX.

Bundle A: neuraLQX + NetKet (a must): Cite neuraLQX and both NetKet papers. This is the standard citation for essentially all neuraLQX-based work.
Bundle B: neuraLQX + NetKet + MPI: If you ran across multiple CPUs/nodes using MPI-backed workflows, add mpi4jax.
Bundle C: neuraLQX + NetKet + JAX/Flax: Ideally, you cite NetKet, JAX as well as Flax.
Bundle D: Full stack: If you used MPI, ideally, you should cite NetKet, JAX, Flax and mpi4jax.

Suggested citation sentences¶

Below are suggested sentences you can paste into your manuscript. Use one as-is or adjust to your style. All variants include NetKet by construction.

LaTeX examples¶

Minimal (recommended default)

Simulations for this work were performed using neuraLQX~\cite{neuralqx1:2026}. neuraLQX is a high-performance
simulations toolkit for loop quantum gravity built on top of NetKet~\cite{netket3:2022,netket2:2019}.

If you used MPI / multi-node runs

Simulations for this work were performed using neuraLQX~\cite{neuralqx1:2026}. neuraLQX is a high-performance
simulations toolkit for loop quantum gravity built on top of NetKet~\cite{netket3:2022,netket2:2019}.
Distributed execution used MPI-enabled JAX communication via mpi4jax~\cite{mpi4jax:2021}.

The ideal non-MPI case:

Simulations for this work were performed using neuraLQX~\cite{neuralqx1:2026}. neuraLQX is a high-performance
simulations toolkit for loop quantum gravity built on top of NetKet~\cite{netket3:2022,netket2:2019}.
The implementation relies on JAX~\cite{jax2018github} and Flax~\cite{flax2020github}.

Full-stack acknowledgement:

Simulations for this work were performed using neuraLQX~\cite{neuralqx1:2026}. neuraLQX is a high-performance
simulations toolkit for loop quantum gravity built on top of NetKet~\cite{netket3:2022,netket2:2019}.
The implementation relies on JAX~\cite{jax2018github} and Flax~\cite{flax2020github}. MPI-distributed execution
used mpi4jax~\cite{mpi4jax:2021}.

“Built with / derived from” (when neuraLQX inspired internal code)

The software used in this work was developed with neuraLQX~\cite{neuralqx1:2026}
as an implementation reference and was built on concepts and components from
NetKet~\cite{netket3:2022,netket2:2019}.

BibTeX¶

Bundle A: neuraLQX + NetKet¶

Bundle B: neuraLQX + NetKet + MPI¶

Bundle C: neuraLQX + NetKet + JAX/Flax¶

Bundle D: Full stack (everything on this page)¶

Hint

You can also see the BibTeX above in neuraLQX by doing the following

import neuralqx as nqx
nqx.cite.all  # show BibTeX for neuraLQX, NetKet, Jax, Flax, mpi4jax
nqx.cite.neuralqx.netket  # show BibTeX for neuraLQX and NetKet only

# you can also compose them
nqx.cite.neuralqx.netket.mpi
# or
nqx.cite.neuralqx.jax.netket

FAQ¶

Do I really need both NetKet citations?: Yes.
What if I used neuraLQX for prototyping but rewrote the final code?: Cite neuraLQX as an implementation reference (see the “Built with / derived from” sentence above) and still cite NetKet!
What if my journal has a strict citation limit?: Use Bundle A at minimum, you have to cite neuraLQX, and you also must cite NetKet. If MPI was essential for feasibility, prefer Bundle B over adding JAX/Flax.