Analog quantum machine learning for near-term hardware
Talkat Near-Term Quantum Computers: Fault Tolerance + Benchmarking + Quantum Advantage + Quantum Algorithms · 2024-04-22..04-26·
summary
Demonstrates a concrete pipeline that maps chemists’ model Hamiltonians onto Rydberg atom arrays by encoding spin-S as 2S qubits, moving atoms for programmable all-to-all connectivity, and implementing multi-qubit GRAPE gates with ~10^-3 error up to 8 qubits to avoid exponential two-qubit overhead. Introduces an operator-resolved density of states extracted from snapshot ensembles (10k–50k samples cited) and hybrid classical postprocessing; reproduces spectra, resolves a disputed molecular spin configuration, and shows gate-count speedups ranging from ~2x to 10–100x depending on model.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:10Quantum chemistry needs correlations
- 5:22Model Hamiltonians for hardware
- 7:10High-spin interactions
- 11:56Challenges in analog ML
- 12:37Rydberg atom connectivity
- 14:47Encoding high spins
- 19:14Multi-qubit gates with pulses
- 21:28Floquet spin interactions
- 25:51Efficient multi-qubit operations
- 26:11Snapshot measurements
- 33:42Resolving spectra
- 38:09Chemical accuracy analysis
- 39:13Beyond quantum chemistry
- 40:39Future directions
- 41:41Summary
related talks
28mAnalog and Digital Quantum Computing with Neutral Atoms
33mAtomic Quantum Processors and the Error-Correction Frontier
01h 10mArchitectural mechanisms of a universal fault-tolerant quantum computer
44mQuantum machine learning: Prospects and challenges
02h 42mQuantum Computing Exposed: Deep Dive
47mQuantum Computing and You
01h 07mQuantum architecture for utility scale with HPC, AI and Quantum
32mComputational quantum chemistry: Classical gaps, quantum opportunities, open questions Part 2