Blog
Research notes and insights on my published and ongoing work.
OptiProfiler Platform: Web Sandbox for Online Benchmarking
Browser-based platform where users upload solvers, OptiProfiler runs benchmark() inside isolated Docker sandboxes, and results return through a FastAPI + Celery stack.
OptiProfiler Agent: An AI Assistant for DFO Benchmarking Workflows
An open-source agent system that guides users through the full OptiProfiler journey: questions, scripts, debugging, and interpreting benchmark results.
Exact lower bounds on the non-convergence probability of probabilistic direct search
A work-in-progress note on quantitative non-convergence phenomena in probabilistic direct search. Coming soon.
Notes on stochastic orders and submartingale-like assumptions for optimization
A short work-in-progress note on stochastic comparison ideas for randomized optimization analyses. Coming soon.
OptiProfiler: A Platform for Benchmarking Optimization Solvers
OptiProfiler is a benchmarking platform for optimization solvers, currently under active development. More details coming soon.
A unified series condition for the convergence of derivative-free trust region and direct search
A work-in-progress note on a common convergence viewpoint for trust-region and direct-search methods. Coming soon.
Gradient convergence of direct search based on sufficient decrease
A work-in-progress note on convergence behavior in direct search methods. Coming soon.
Non-convergence analysis of probabilistic direct search
We establish the non-convergence theory for probabilistic direct search, answering the question: when will your algorithm fail to converge?
Error analysis of finite difference scheme for American option pricing under regime-switching with jumps
We provide a rigorous error analysis for the finite difference scheme applied to American option pricing under regime-switching models with jumps, establishing convergence rates and stability conditions.