After a follow-up paper vote on Monday, the most popular presentation selected for Irv Lustig’s last Opti session of the day was “Writing Better Mixed Integer Programming Models.”

I suspect Irv organized a second vote because he wanted to avoid having to give the “Introduction to ILOG CP Optimizer” presentation which actually won the original electronic vote. Whether or not there was any vote-rigging involved, I don’t know, but I do know that the vote was quite spread, showing that there was interest in all the proposed topics. I thought of suggesting to Irv that he combine all seven proposed topics into a single talk, but who knows how that would have turned out!

Whether it’s because math was never really my strong point at school, or whether it’s because my brain was fried by 3.30pm on Day 3, I’m afraid Irv’s suggestions for improving MIP models was way over my head. I suppose I should have guessed that when he started with “It’s going to include a lot of math, and if you get lost in the math, so be it.” Oh boy.

If you’re interested in this topic, I think your best bet is to wait until the video is published online some time in the next few weeks! In the mean time, though, if you have any questions for Irv, feel free to post a comment here. Anyone got any Advil?

In this session, Bob Bixby, ILOG Chief Science Officer, discussed how optimization has become a key component in addressing common business and manufacturing problems today. He explained what has and hasn’t changed in approaches to optimization over the many years that he has been in the business.

Bob used IBM’s East Fishkill 300mm semiconductor wafer manufacturing plant as an example to explain the benefits that optimization and automated scheduling can bring to highly complex manufacturing processes.

People have been trying to find ways to improve process optimization for years, of course. The difference today is that processes are now so complex that traditional methods can no longer cope. It’s no longer feasible to try and optimize an entire project. Instead, optimization solutions must break down processes into separate sub-processes and optimize those individually. The old generation of solutions, despite enormous progress in processing capacity, required too much time to generate optimum solutions. even with a turnaround time of less than 24 hours, solutions would often be obsolete before they could even be implemented.

By combining constraint programming optimization, process modeling and decision management systems, ILOG’s optimization suite can be used to build real-time optimization solutions. Bob’s slides contained plenty of statistics from the IBM Fishkill application, showing how this approach has been used to bring dramatic optimization improvements.

He highlighted, for example, the 25.3% reduction in cycle time (production time + waiting time) that the ILOG solution was able to achieve in certain plant sub-processes, which contributed to overall efficiency improvements of around 6%. Considering that the plant handles some 15,000 batches a day, that’s a considerable achievement and represents considerable cost savings for the plant.

These cutting-edge optimization solutions are no longer a nice-to-have. They are essential for survival in today’s white hot competitive environments.

Dialog 08 is now fully underway - we’ve already seen Jerome Boyer talking about Agile Business Rules Development and Pierre Berlandier covering Rule Governance (thanks to James Taylor for the coverage).

Just to recap, these are the tracks that will be running over the next couple of days:

• Business Rule Management Systems (BRMS)
• Optimization
• Visualization
• Supply Chain Management (SCM)
• Financial Services Industry (FSI)

Dialog attendees: feel free to use this post for comments on the tracks in general and anything around the event.