Quantuva is a research effort in quantum software. We build the compilers, simulators, and error-mitigation methods that sit between a quantum algorithm and the noisy machines that run it — and we publish everything we learn.
Quantum hardware has improved enough that the interesting problem is no longer "does it exist" but "what do you do with the noise." The algorithms we want to run were designed for ideal machines; the machines we have are far from ideal. The gap between the two is a software problem, and that's the one we work on.
Concretely, we study circuit transpilation — rewriting an algorithm for a specific topology and gate set — and error mitigation — reading a trustworthy number out of a noisy measurement. The SDK is the working artifact of that research: what we learn gets folded into the tooling, and the tooling is how we test what we think we know.
We are deliberately not a hardware company, and we are not chasing a qubit-count headline. Our work is about the layer above the chip, and it is measured by whether a researcher can trust the number their circuit reports.
Code, lab notes, and papers are released without an embargo. There is no separate academic edition; what you see here is the whole of it.
Every paper ships with the code and data that produced its figures. If you can't reproduce a claim, the claim is not finished.
We write down where a method stops working, not just where it works. The lab notes are full of open questions, on purpose.
Works on qubit mapping and SWAP routing, and on the question of when minimizing circuit depth stops being the right objective.
Builds the simulators — statevector, density-matrix, and tensor-network — and the noise models that make them honest.
Works on zero-noise extrapolation, Pauli twirling, and readout correction, including the regimes where each stops helping.
Works on surface-code decoding and stabilizer extraction, and on how a code patch behaves when the underlying qubits are noisy.
Works on gate decomposition and synthesis for trapped-ion and superconducting backends, and on compiling to native two-qubit interactions.
Works on benchmarking and validation — how to tell when a compiled circuit is actually trustworthy, and how to measure it.
Founded as a small research group with one goal: make near-term quantum computers useful to people who aren't building the hardware.
First public SDK with a statevector simulator, a transpiler, and a noise model — released alongside the first open papers.
Density-matrix and tensor-network simulators, error-mitigation passes, and the beginning of the circuit library.
The work continues in the open — transpilation, mitigation, and the surface code, with everything written down as we go.