Reproduce “Hamiltonian Metric Learning and Energy-Based Training: A Dissipative Geometric Framework for Optimization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sparsho Chakraborty, Mohammad Alamgir, Nishanth M, Akshit Nanda, Ram Prasad Padhy, Sayan Mukherjee
HAMLET models optimization as damped motion through a loss landscape, using a positive-definite metric to map momentum into parameter velocity and accounting for geometric force when that metric varies. In a controlled quadratic image-restoration test, setting the metric to the image Hessian equalizes curvature modes. HAMLET approaches the known optimum faster by objective gap and gradient norm, not runtime.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.