Patrick Mihelich: Linear programming made easy with Boost Proto

Описание: Patricks' talk from C++Now! 2012

Problems in fields as diverse as operations research, finance, and chip design can be modeled as linear programs. As a highly general method for decision problems, with efficient algorithms, LP is a valuable addition to your toolbox.

Powerful software libraries such as GLPK exist for solving LPs, but these low-level APIs can be hard to use, requiring the user to transform the problem into a less human-readable form. Modeling languages such as AMPL make it easy to express a problem declaratively, but lack the power and familiarity of a general-purpose programming language.

This presentation introduces CVX++, a Boost.Proto-based DSEL for expressing and solving linear programs. Proto allows the best of both worlds; a declarative programming style embedded in C++. CVX++ uses Proto transforms to convert the user's objective function and constraints into a more machine-friendly representation, with GLPK as the backend solver.
---

*--*

---