User:Natasha.zrazhevska/sandbox

Portfolio Safeguard (PSG) is an optimization package for solving a wide range of optimization, statistics, and risk management problems. The system is designed for complex, large-scale applications.

PSG is a product of American Optimal Decisions, Inc.

PSG can solve standard general classes of optimization problems: linear, quadratic, nonlinear, MIP. However, the main focus is on solving typical problems in various engineering areas, such as minimizing Value-at-Risk (quantile), maximizing Entropy or Maximum Likelihood functions with various constraints. Solution of such problems is automatically reduced to a sequence of linear, quadratic or MIP sub-problems.

PSG package uses a set of precoded nonlinear functions for every engineering area. Therefore, codes usually include only several lines.

PSG initially was focused on financial engineering problems and was called “Portfolio Safeguard”. PSG has a diverse set of risk management functions: Variance, Value at Risk (quantile), Expected_shortfall (also called Conditional Value-at-Risk, expected Value-at-Risk, expected Tail Risk, or superquntile), Drawdown, Entropy. Later versions of PSG were extended to solve other classes of problems: Advanced Statistics, Logistics, Optimization of Large Physical Systems etc.

PSG includes four nonlinear nonsmooth optimization solvers for problems with a large (up to 108) number of variables. It can solve very efficiently nonthmooth convex optimization problems, such as minimization of maximum of linear functions (e.g., Conditional Value-at-Risk) and Two Stage Stochastic optimizations.

The interface for PSG is implemented in four programming environments: R_(programming_language), MATLAB, C++, and Run-File (Text).