Research map / Probabilistic and scientific modeling / Variational and generative sampling
LLM-Guided Evolutionary Program Synthesis for Quasi-Monte Carlo Design
Amir Sadikov
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| evolve Sobol' direction numbers for dims 4-6 instead of Joe-Kuo parameters | Sobol' sequence · Asian option rQMC, training example, N=8192 | MSE: 4.523e-05 -> 4.104e-05 | Improved |
| evolved direction numbers, small sample sizes | Sobol' sequence · Asian option rQMC, N=32..256 | MSE: 0.2484 -> 0.2523 at N=32 | Hurt |
| evolved direction numbers, N>=512 on out-of-the-money option | Sobol' sequence · Asian option out-of-the-money, N=1024 | MSE: 0.001457 -> 0.001313 | Improved |
| evolved direction numbers on in-the-money option, N=1024 | Sobol' sequence · Asian option in-the-money | MSE: 0.000261 -> 0.000260 (p=0.2581) | No clear effect |
| evolved direction numbers on Barrier options | Sobol' sequence · Barrier options, N>=512 | MSE | No clear effect |
| evolved direction numbers on Lookback, Basket, Bermudan options | Sobol' sequence · exotic options, N>=512 | MSE | Improved |
| LLM evolutionary search for 2D point sets N=100 | Fibonacci/MPMC/Clement · 2D point set N=100 | star discrepancy: 0.0188 -> 0.0150 | Improved |
| LLM evolutionary search for 2D N=16 versus known optimum | Clement et al. · 2D point set N=16 | star discrepancy: 0.0739 vs 0.0745 | Hurt |
| LLM search for 2D N=40 | Clement et al. · 2D point set N=40 | star discrepancy: 0.0332 -> 0.0331 | Improved |
| LLM search for 3D N=8 | Clement et al. · 3D point set N=8 | star discrepancy: 0.1875 vs 0.1937 | Hurt |
| add Phase 2 iterative numerical optimization after direct construction (ablation) | LLM evolutionary search · 2D point set N=16 | star discrepancy: 0.0924 -> 0.0744 | Improved |
| add Phase 1 direct construction (tuned shifted Fibonacci lattice) over initial lattice (ablation) | LLM evolutionary search · 2D point set N=16 | star discrepancy: 0.0962 -> 0.0924 | Improved |
Closest papers by meaning
- MOTIF: Multi-strategy Optimization via Turn-based Interactive Framework (AAAI 2026 · similarity 0.85)
- Property-Guided LLM Program Synthesis for Planning (NeurIPS 2026 · similarity 0.85)
- PathWise: Planning through World Model for Automated Heuristic Design via Self-Evolving LLMs (ICML 2026 · similarity 0.85)
- SolverLLM: Leveraging Test-Time Scaling for Optimization Problem via LLM-Guided Search (NeurIPS 2025 · similarity 0.85)
- HSEvo: Elevating Automatic Heuristic Design with Diversity-Driven Harmony Search and Genetic Algorithm Using LLMs (AAAI 2025 · similarity 0.84)
- BayeSQP: Bayesian Optimization through Sequential Quadratic Programming (NeurIPS 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
