Direct/adaptive-mixture phase-gradient learning for neural-network quantum states with complex phase structure
arXiv:2606.13912v1 Announce Type: cross Abstract: Neural-network quantum states (NQS) are a leading variational tool for quantum many-body physics, yet their optimization is fragile whenever the ground state carries a non-trivial sign or complex phase structure, a situation generic to gauge fields, broken time-reversal symmetry, and fermionic statistics. We trace this fragility to the stochastic estimator of the phase gradient rather than to network expressiveness. The phase sector of the Monte Carlo energy gradient is a noisy score-function estimator; differentiating the local energy instead
The continuous push for more robust and efficient quantum computing models necessitates addressing fundamental challenges in quantum state optimization, which is currently a limiting factor.
This research addresses a core fragility in quantum computing's foundational tools, potentially unlocking more powerful and stable quantum machine learning applications for complex systems.
Improved methods for optimizing neural-network quantum states will enhance the accuracy and stability of quantum simulations, especially for systems with intricate phase structures.
- · Quantum computing researchers
- · Quantum hardware developers
- · Materials science
- · Pharmaceutical research
- · Current less efficient optimization methods
- · Classical simulation methods for complex quantum systems
More accurate variational quantum simulations become feasible for previously intractable problems.
Accelerated discovery of new materials and molecular structures with desired quantum properties.
Enhanced development of fault-tolerant quantum algorithms and a broader applicability of quantum machine learning.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG