Learning SYK Hamiltonians
Abstract
We study the problem of learning the dense Sachdev--Ye--Kitaev (SYK) Hamiltonian from copies of its Gibbs state. Existing algorithms for Hamiltonian learning typically rely on geometric locality or bounded interaction degree and therefore do not apply to SYK, where each quartic interaction overlaps with Θ(n3) others. We show that this obstruction can be overcome by exploiting the random mean-field structure of the model. At any constant temperature, we prove that with high probability over the SYK couplings, the entire Hamiltonian can be learned to inverse-polynomial accuracy using polynomially many samples. Furthermore, when the inverse temperature is restricted to be a sufficiently small constant, we construct a quasipolynomial-time learning algorithm which is qualitatively different from the sample-efficient algorithm.
Related
- Operator Score Matching for Learning Quantum Hamiltonians
- Classical Hardness of Learning Functions of Hamiltonians
- Scalably learning one-dimensional quantum many-body Hamiltonians from dynamical data
- Learning Many-Body Hamiltonians Using a Local Probe
- Deep Learning GW Quasiparticle Hamiltonians for Many-Body Excited-State Electronic Structure at Scale
BibTeX
@misc{anshu2026learning,
title = {{Learning SYK Hamiltonians}},
author = {Anurag Anshu and Srinivasan Arunachalam and Sitan Chen and Yeongwoo Hwang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.02178},
eprint = {2610.02178},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.02178},
}
RIS
TY - UNPB
TI - Learning SYK Hamiltonians
AU - Anurag Anshu
AU - Srinivasan Arunachalam
AU - Sitan Chen
AU - Yeongwoo Hwang
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.02178
UR - https://arxiv.org/abs/2610.02178
AB - We study the problem of learning the dense Sachdev--Ye--Kitaev (SYK) Hamiltonian from copies of its Gibbs state. Existing algorithms for Hamiltonian learning typically rely on geometric locality or bounded interaction degree and therefore do not apply to SYK, where each quartic interaction overlaps with Θ(n^3) others. We show that this obstruction can be overcome by exploiting the random mean-field structure of the model. At any constant temperature, we prove that with high probability over the SYK couplings, the entire Hamiltonian can be learned to inverse-polynomial accuracy using polynomially many samples. Furthermore, when the inverse temperature is restricted to be a sufficiently small constant, we construct a quasipolynomial-time learning algorithm which is qualitatively different from the sample-efficient algorithm.
ER -