Mechanistic Interpretability and Causal Feature Steering of Neural Quantum States via Sparse Autoencoders
arXiv:2607.01336v1 Announce Type: cross Abstract: Neural Quantum States (NQS) are a remarkably expressive class of variational ans\"atze for quantum many-body wavefunctions, yet little is understood about their internal mechanisms: trained on variational objectives alone, how do NQS accurately capture physical observables that they have never been explicitly optimized for? In this work, we present a systematic approach to analyze the internal activations of NQS using sparse autoencoders. We extract features from the residual stream and demonstrate that these features strongly correlate with ph
This research provides a methodical approach to understanding the 'black box' nature of Neural Quantum States, which is crucial as these models mature and become more complex.
Understanding the internal mechanisms of Neural Quantum States is vital for their reliability, explainability, and broader adoption in scientific discovery and quantum computing.
This work introduces a validated methodology (sparse autoencoders) for interpreting NQS, potentially accelerating their development and application by providing insights into their decision-making processes.
- · Quantum Computing Researchers
- · Materials Science
- · Drug Discovery
- · AI Explainability Tools
Improved interpretability of Neural Quantum States leads to faster development cycles for quantum algorithms and simulations.
More reliable and trusted NQS enable breakthroughs in areas like novel material design and complex molecular interactions.
A deeper understanding of NQS could inspire new architectural designs for classical AI models aiming for similar expressiveness and efficiency.
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