arXiv:2610.06828
The practical cost of magic state cultivation
Rohan Mehta, Varun Menon, Hengyun Zhou, Mikhail D. Lukin, J. Pablo Bonilla Ataides
Abstract
Magic state cultivation is a promising method for preparing high-fidelity non-Clifford resource states - a central component of fault-tolerant quantum computing - with low spacetime overhead. In benchmarking cultivation, existing resource estimates assume access to noiseless transversal measurement of the output magic state to allow for postselection during the final escape stage of the cultivation process. However, such measurements would destroy the magic state and are therefore unavailable during computation. We develop an open-boundary postselection method, Caliper, that uses only nondestructive mid-circuit syndrome information and substantially outperforms existing methods operating under these constraints. In some regimes, our postselection method uniquely recovers the resource efficiency of prior estimates. However, in other cases, we observe a tradeoff. At spacetime costs reported in previous works, the achievable logical error rates can be orders of magnitude higher, and recovering the reported logical error rate requires substantially larger codes and longer escape stages. Our results highlight how practical considerations involving mid-circuit postselection can impact the design and performance of magic state cultivation protocols.
AI Summary
Magic-state cultivation estimates often use a noiseless terminal measurement to decide escape-stage postselection, although this would destroy the output state. This paper asks what error rates and resource costs remain achievable when acceptance depends only on nondestructive syndrome information available mid-circuit. Caliper estimates eventual decoding-failure probability by searching over possible hidden terminal syndromes and combining minimum-weight corrections from both logical classes. Precomputed graph distances bound correction-weight changes and reduce repeated matching calls. Stim simulations benchmark it against closed-boundary postselection and existing open-boundary heuristics for color- and fold-transversal surface-code cultivation, using $\ket{\bar Y}$ proxies and circuit-level depolarizing noise at $p=10^{-3}$. Caliper matches or improves the tested open-boundary methods at fixed acceptance. For fault-distance-five color-code cultivation, reaching $10^{-7}$ costs about $10\%$ more spacetime than closed-boundary postselection, versus roughly $7\times$ for the next successful benchmark; reaching $10^{-9}$ costs $3.7\times$ more. For fault-distance-three surface-code cultivation, full visible-syndrome error detection yields $1.4\times10^{-5}$, above the $2\times10^{-6}$ target. Among tested modifications, recovering that target requires increasing escape distance/rounds from $(7,3)$ to $(11,6)$, with roughly $5\times$ total expected spacetime. A low-noise theorem establishes that zero visible syndrome maximizes the conditional-error suppression exponent, but not necessarily finite-noise fidelity. Limitations include heuristic hidden-syndrome search, miscalibrated probability estimates, and uncertain transfer from $\ket{\bar Y}$ to exact $\ket{\bar T}$ cultivation. Median Caliper latency is $25.3\,\mu\mathrm{s}$ for the smaller surface-code protocol; waiting costs could dominate on superconducting hardware without acceleration. Adaptive escape optimization remains open.
Error found: The abstract does not identify that its quantitative cultivation-cost conclusions are based entirely on Clifford |Y⟩ proxies, while their extension to exact non-Clifford |T⟩ cultivation with open-boundary postselection remains untested.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.