OQFT is the studio’s first quantum project. It builds the optimistic quantum Fourier transform, the step in Shor’s algorithm that reads out a hidden period, in a form that’s right on almost every input and much cheaper than the textbook circuit. The explainer covers the idea: cut the register into blocks, keep only the rotations between neighbours, and let every block run at once.
Three releases went out in one day. 0.1 brought the circuit model, a simulator and the textbook and approximate QFT. 0.2 added the optimistic QFT, sized from an error target. 0.3 added oqft check, which runs inputs through the simulator and shows where the optimistic circuit fails. Circuits come out as OpenQASM 3, and a Qiskit example loads them and checks the matrix against Qiskit’s own QFT.
On 2,048 qubits at an error target of 0.01, OQFT’s circuit is 198 layers deep against 4,096, with about 104,000 gates against 2.1 million, and no qubits beyond the register. It isn’t a win everywhere: the crossover case study found the textbook circuit shallower below 68 qubits. The testing post shows the failure where theory puts it, at the top of a block’s range.
Like RMQA and the rest of the studio’s Go tools, it’s one binary with releases for Linux, macOS and Windows, built and smoke-tested by the release workflow. Open source under Apache 2.0, with the latest release on GitHub. A resource estimator for factoring is next.