Optimization Direct at INFORMS 2022 ANNUAL MEETING, Indianapolis IN, October 16-19, 2022

We are holding technology tutorials and workshops at INFORMS

Technology Tutorial at INFORMS

ODH Python Primer

Presented by: Robert Ashford
Monday, October 17, 11:40am-12:15pm

 
This short tutorial shows participants how to build a basic model using the ODH|CPLEX in Python. This session includes setting the Python environment, reading data from a csv or spreadsheet, creating variables, objective functions, and constraints, solving the model, and returning the results. Additionally, this session points the participants to further reading so that they may expand their capabilities. Furthermore, we will present the brand new ODH|Gurobi generic API and demonstrate it in Python.

Information


Technology Workshops at INFORMS

Solving the Most Difficult Mixed Integer Programming Models with ODH

Presented by: Robert Ashford
Saturday, October 15, 1-3:30pm

ODH is a package for accelerating MIP optimization and extending its range. It runs concurrently and synchronously with a MIP optimizer exchanging information with it.

Although ODH was primarily designed to enable MIP optimization to be applied to models too large or difficult to be amenable to the standard commercial MIP optimizers such as CPLEX, Gurobi or other commercial packages, it also accelerates the optimization of most models that are regularly solved with standard MIP.

ODH has been developed by Optimization Direct Inc. over the past 8 years to handle real world customer models for which standard MIP was not effective. On Optimization Direct’s standard hundred model sub-set of mid-sized models taken from its library of nearly 1000 real customer models, ODH|CPLEX and ODH|Gurobi reduce the average optimality gap after two hours to nearly half that found by using the optimizer alone. The set was chosen to be representative rather than amenable to ODH.

On the MIPLIB version 7 open set of 29 models to which no integer feasible solution had been found, ODH|CPLEX found solutions to 5 of them in 2 hours. Of the remaining 257 models to which no optimal solution had been proven, ODH found and proved optimal solutions to 13 of them and found better solutions to any that had been previously found to 45% of them. Again, with runs of just 2 hours.

Optimization Direct has customers in the scheduling, telecoms, retail, logistics and supply chain areas who use ODH. At least three of these have applications that would fail without ODH technology. In January 2022, one said ‘ODH|CPLEX is comparable to the best optimization technology available from anyone on my small models and much superior to any other optimizer on the large ones’.

ODH uses a number of heuristics, especially a novel decomposition based method, and information from a MIP optimizer to feed that MIP optimizer with good solutions. This is made possible by its new thread synchronization technology within which the ODH and MIP threads can work efficiently.

Information