Optimization Modules: our key differentiator

One of the key differentiators behind our solutions is the toolkit of mathematical techniques we employ. Coupled with decades of domain experience, our development team has customized these specialized mathematical techniques to match our optimization goals. The resultant optimization modules reside at the core of, and uniquely power our optimization engines. A representation of our customized, optimization modules include:

• Structured Genetic Algorithm Framework - uses customizable solution representations to eliminate problems with classic genetic algorithm approaches where solutions must be represented as strings of 0's and 1's.  The custom structured solution representation fits the solver to the problem using modern object-oriented modeling, instead of encoding and decoding data as is usually the case in applications of theoretical OR techniques.  In addition, this framework allows the use of heuristics designed to drive solutions toward optimality, rather than merely depending on random perturbations to find solutions.
• Fast, Combinatorial Search Tools - set of skeleton algorithms, which generate all possible combinations in a search space (such as all subsets of a given set, all sequences of a set, etc.) which can be customized to "prune" suboptimal branches of the search tree, and search in a preferred order.  These search tools are used to build fast solvers for knapsack/packing problems, sequencing problems (such as Traveling Salesman routing problem with Time Windows), and others.
• Generalized Assignment Solver - fast solver for the Generalized Assignment Problem (GAP) which uses the Ejection Chain method (similar to the look ahead function in a computer chess game) to improve solutions.  This solver is also used as a skeleton to build special purpose modules for specific problems, such as vehicle routing and warehouse slotting.
• Problem-specific Heuristics - used as further building blocks and initial solution seeding, such as traveling salesman, vehicle routing, bin-packing, etc.

Customer Centric Solutions: fitting the tools to the customer’s needs

Optricity further differentiates its approach by using its custom optimization modules (at a bit lower level) to create purpose-built engines and applications that directly operate on (and exploit the structure of) real world problems. Using good quality, object-oriented software engineering practices, our team precisely models the customer’s juncture optimization problem and then fits the tool to the embedded optimization opportunities.
The Optricity team believes customer centric solutions provide a better and more fruitful approach than to use a “divorced” black box. In the black box scenario data and constraints must be fed in using native representation, after which the result must be deciphered and validated. Optricity’s object-oriented modeling approach enables easier identification of multiple embedded optimization problems. By fitting several different modules into one solution the Optricity developers exploit the problem structure and enhance solution performance and usefulness of results.