README.md

In mludkov/mlOSP: Machine Learning and Regression Monte Carlo Algorithms for Optimal Stopping

mlOSP

Regression Monte Carlo for Optimal Stopping

Description: A suite of regression Monte Carlo algorithms utilizing Machine Learning for Optimal Stopping Problems (mlOSP).

Includes both static and sequential experimental designs. We implement the original Longstaff-Schwartz and Tsitsiklis-van Roy algorithms, as well as machine learning approaches that explicitly specify the underlying experimental designs. The mlOSP template then allows to mix and match the choice of the regression method, the experimental design, and the stochastic simulator. Key solver functions are osp.prob.design (original LSM), osp.fixed.design (a variety of space-filling or user-specified designs, generally assumed to be batched), osp.seq.design (sequential designs using a collection of pre-specified Expected Improvement Criteria), osp.seq.batch.design ( sequential design with adaptive batching) and osp.tvr (TvR method). Also implements the Bouchard-Warin hierarchical adaptive partitioning with linear regression (osp.design.piecewisebw). The library currently works with 10+ regression emulators, see documentation.

The Bermudan_demo vignette provides a short illustration with a 2D Bermudan basket Put. The two demo R files provide the source code for the http://arxiv.org/abs/2012.00729 article and the respective benchmarked solvers.

Work partially supported by NSF-1521743 and NSF-1821240.

mludkov/mlOSP documentation built on April 29, 2023, 7:56 p.m.