Utility, Computational Cost, and Effective Quantum Volume of NISQ Experiments
summary
Argues that optimized tensor network contraction is the leading classical tool, links its cost to a treewidth/effective area and shows random circuit sampling still resists classical emulation (estimated ~50 years on Frontier for certain circuits). Introduces effective fidelity and measures a smaller-than-lightcone spread via OTOCs—butterfly velocity—so many observables only need ~100 gates not ~3,000; reproducing an IBM Floquet local‑Z curve requires ~28 qubits and takes about one second per data point on a single GPU.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:31Where are experiments at?
- 4:16Noise vs. volume
- 7:27Tensor network contraction as a framework for simulation
- 10:59Random Circuit Sampling: computational cost
- 17:56Measuring observables
- 21:49Out-of-time-order correlators
- 29:21Floquet evolution of transverse field Ising model
- 37:25Non-local observable: inverted light cone
- 39:27Other simulation papers
- 42:12Asymptotics vs. finite sizes in NISQ regime
- 53:20Asymptotics vs. finite sizes in NISO regime
- 54:08Acknowledgements
related talks
01h 09mRecent Progress in Quantum Advantage
32mThe intersection of Quantum Computing and Distributed Systems...
28mThe Problem with Qubits
48mStatistics In Quantum Computational Supremacy
01h 08mPutting Quantum Supremacy on a Rigorous Footing
32mComputational quantum chemistry: Classical gaps, quantum opportunities, open questions Part 2
37mResource-Efficient Quantum Computing by Breaking Abstractions
02h 42mQuantum Computing Exposed: Deep Dive