Learning quantum systems via out of time order correlators

Value of the Information & Strength of the Argument

The talk provides substantial value by introducing a new class of observables (OTOCs) into the quantum learning toolbox and demonstrating clear advantages in realistic scenarios. The argumentation is solid: it combines physical intuition from statistical mechanics (e.g., Loschmidt echo, scrambling) with numerical simulations and rigorous hardness results. The speaker carefully delineates when time reversal helps and when it does not, addressing potential limitations. The discussion of DQC1 and restricted access is particularly compelling, as it connects to known complexity-theoretic results. Overall, the claims are well-supported and the presentation is persuasive.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high: the work is based on a peer-reviewed arXiv paper (2208.02254) and includes formal proofs for some results. Sources are cited clearly (the main arxiv link is provided in the description). The talk is well-structured, with a clear progression from motivation to concrete examples to formal claims. The title accurately reflects the content, focusing on learning quantum systems via OTOCs. The inclusion of numerical experiments and robustness checks strengthens the reliability. No discrepancies were detected between the title and the actual content.

The title accurately reflects the content, which focuses on using out-of-time-order correlators to learn quantum systems, with emphasis on time reversal experiments and their advantages in learning tasks.

8/10

The talk is based on a peer-reviewed arXiv paper (2208.02254) and presented by a researcher from Google Quantum AI and UC Berkeley. The content is rigorous, well-structured, and includes both physical intuition and provable results. Minor uncertainties in pronunciation and informal presentation style do not detract from the scientific reliability.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 02:00Introduction to the talk and acknowledgment of collaborators; overview of the goal: learning properties of quantum processes.
• 10:00Discussion of challenges in learning quantum systems: thermalization, restricted access, and weak parameter estimation.
• 20:00Introduction to time reversal experiments and out-of-time-order correlators; explanation of the butterfly effect and information spreading.
• 35:00Physical examples: numerical results showing advantages of OTOCs for learning Hamiltonians in systems with restricted access and weak interactions.
• 50:00Provable results: binary classification task solvable in constant time with OTOCs; discussion of DQC1 hardness and exponential separation.
• 1:00:00Concluding remarks and outlook; mention of related work on Hamiltonian learning using time-reversal-inspired algorithms.

• Learning quantum systems via out-of-time-order correlators — Main paper discussed in the talk; provides theoretical and numerical evidence for advantages of OTOCs in quantum learning.