The field of convex optimization continues to grow in popularity as algorithms continue to be improved, new software is developed, and new applications are identified. The recent surge in interest in machine learning and data mining models, many of which are convex, has certainly contributed to this growth.
CVX is a software package developed by Michael Grant and Stephen Boyd that greatly simplifies the process of specifying and solving convex optimization problems. It transforms MATLAB into a modeling framework, allowing constraints and objectives to be expressed using natural MATLAB syntax. CVX imposes a modest set of conventions, called the disciplined convex programming ruleset, that its users must adopt when constructing models. These conventions enable CVX to automatically verify convexity, and to perform the transformations required to express models in standard, solvable form.
When the fundamentals of CVX were first conceived, its authors expressed concern over whether the work was sufficiently "academic" in nature. Thankfully, those concerns were alleviated, as CVX has been widely adopted for use in teaching, and cited extensively in published research. Last year, Dr. Grant and Prof. Boyd were awarded the Beale Orchard Hays Prize by the Mathematical Optimization Society for their work on CVX.
In this talk, Dr. Grant will provide a brief introduction to convex optimization and an overview of the CVX modeling language and its more unique capabilities. He will offer some personal history of its development, and discuss how his experience has shaped his views on the notion of "academic" software.
Current 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996