Reproduce “Warped geometric information on the optimisation of Euclidean functions” 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
Marcelo Hartmann, Bernardo Williams, Hanlin Yu, Mark Girolami, Alessandro Barp, Arto Klami
Researchers developed an optimization algorithm for real-valued Euclidean functions by rephrasing the problem on a specifically constructed warped Riemannian manifold. This approach integrates a 3rd-order geodesic approximation and efficient retraction maps, reducing iteration counts on challenging benchmarks.
Key insights
- Embedding a Euclidean function's graph into a warped Riemannian manifold enables the leverage of intrinsic geometric information while maintaining computational tractability.
- Higher-order (3rd-order) approximations of geodesic paths can provide more accurate search directions by capturing more of the manifold's curvature, significantly improving iteration efficiency without incurring full Hessian costs.
- The specific choice of a warped metric simplifies the calculation of key Riemannian components, allowing the proposed RCG algorithm to achieve linear memory and quadratic arithmetic complexity in problem dimensionality.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.