Research

We sit at the intersection of computation and physics: using numerical methods to simulate the physical world, and using physics to understand the nature of computation. Our full publication list lives on Google Scholar ↗.

Tensor networks

Contraction-order optimization and generic tensor networks — a unified framework we use to compute solution-space properties of combinatorial optimization problems and to run probabilistic inference on graphical models.

Quantum computation & error correction

Quantum algorithm design and simulation (Yao.jl), tensor-network decoders for quantum error-correcting codes, universal quantum computing with a single arbitrary gate, and algorithms for neutral-atom (Rydberg) hardware.

Computational complexity

The nature of computation from a physics perspective: problem reductions, automated discovery of branching rules with optimal complexity, and embedding hard problems into physical devices.

AI-native scientific computing

Automatic differentiation through numerical methods, and human-AI collaboration workflows for building research software — test-driven vibe coding, agent infrastructure, sustainable automation.

Open-source software

Research here ships as open-source Julia packages. Group organizations: CodingThrust, QuantumBFS, Happy-Binaries.

Yao.jl co-author

Extensible, efficient quantum algorithm design framework — one of the fastest quantum circuit simulators.

OMEinsum.jl maintainer

Einstein-summation engine with hyper-optimized contraction-order search, the backbone of our tensor-network stack.

GenericTensorNetworks.jl author

Solution-space properties of combinatorial optimization problems via generic tensor networks.

TensorQEC.jl group project

Tensor-network decoding and simulation of quantum error-correcting codes.

ProblemReductions.jl group project

Reductions between computationally hard problems (spin glass, SAT, ...).

TensorInference.jl author

Probabilistic inference on graphical models with tensor networks and differentiable programming.

Selected talks

20 talks since 2023 — expand