This Small Business Innovation Research (SBIR) Phase I project will address the challenges of supporting large-scale optimization models that are most naturally represented by nonnumerical and other nontraditional forms of objectives and constraints. Building on the project team's experience in designing, implementing, and marketing more traditional optimization modeling software, the project will address the feasibility of a unified approach that enables one natural and convenient modeling language to support a broad variety of nontraditional model forms and solver types. In addition to considering the design of language extensions, the research will address the creation of interfaces for diverse solvers, particularly as it relates to issues of automated problem detection and conversion. The expected outcome is a prototype optimization modeling language that will alleviate frustrations with current systems and encourage experimentation with a broader variety of solution methods. Optimization has increasingly become an established paradigm for modeling of phenomena throughout science, engineering, and business. The new product envisioned will address a significant customer need by advancing the ability of optimization modeling languages to describe and generate problems in the ways that are most natural to modelers
