Friday, June 26, 2009

Hitler's army created Operations Research proof bombers?


(Photograph by Linda Reynolds/Flying Wing Films)

Those that study the history of Operations Research know that a lot of its genesis came from World War II. British experts used mathematical principles to identify the optimal places to locate the new radar technology to identify incoming enemy aircraft. A lot of the credit can go to the work at Bawdsey Research Station under the supervision of A. P. Rowe. A. P. Rowe's team analyzed and improved upon how the radar would track targets and setup optimal conditions for the radar activities. An excellent article on the work of the World War II researchers can be found at Wikipedia.

Yet Hitler's Nazi army was developing a secret "stealth" plane that fortunately never took flight. Scientists at Northrop Gruman have recreated the Horten 2-29 at their facilities. Surprisingly they discovered that using legacy World War II radar equipment that it did indeed would be radar resistant. Perhaps if the Nazi army had a fleet of these planes Operations Research would not be where it is today? Thank goodness to the efforts of A. P. Rowe and other leading mathematicians that they were able to institute foundations in Operations Research. Another reason to thank the founding fathers of modern day Operations Research.

Monday, June 22, 2009

Machine Learning Open Source Software - mloss.org

Dan at the OpenScience Project tells about a new community driven project called Machine Learning Open Source Software. Mloss.org is devoted to open source software projects related to machine learning, algorithms, and statistics. The site is essentially an online software research journal. Mloss.org does a good job of explaining in detail the software projects. They also do a good job of maintaining citations to specific software projects so that the community of developers will maintain credit and recognition.

Machine Learning (wikipedia) is the science of applying algorithms and formulas to allow machines to learn based on given data or other input. This is a growing field as technology advances the processing power of computers.

Thursday, June 18, 2009

Open source commonality with science


Here is a great essay about What Open Source shares with Science from ZDNet UK. The essay writer basically explains that science and open source software share essentially the same underpinnings of the scientific method. That idea that hypothesis, theories, experiments, thoughts and ideas should be documented and shared with the rest of the (scientific) community. From the essay...

The speed of progress is greatly enhanced by virtue of the fact the practitioners of Science publish not only results, but methodology, and techniques. In programmatic terms, this is equivalent to both the binary and the source code. This not only helps 'bootstrap' others into the field, to learn from the examples set, but makes it possible for others to verify or refute the results (or techniques) under investigation.


I've postulated before that proprietary software might be a hindrance to Operations Research. This essay does a great job explaining my point. In order to advance Operations Research concepts in software the best thing for the field is to allow peer review. This will only allow the field to grow and prosper. Otherwise are we not just competing against our own peers?

Tuesday, June 9, 2009

PyMathProg releases version 0.14

The developer of PyMathProg released version 0.14 today. Here is the news from the developer...

This release enables solver options, and you can actually utilize different methods to solve your problems. If you wish, you can now setup pymprog into your python onsite packages. You can either download the zip or the tgz archive. Documentations are also online at:http://pymprog.sf.net
PyMathProg is a Python reincarnation of GLPK and the GNU MathProg modelling language. The Python programming language is simpler to use than most languages and a quick learning curve.

Monday, June 8, 2009

Operations Research for fun

Those of us in the Operations Research profession often get consumed in our work. I do not take for granted what I do for a job. I love doing Operations Research on a daily basis. Yet sometimes our minds get caught up in the work world. There are plenty of areas that we can take a side route and enjoy our talents in other ways.

Here is a list of things an Operation Research professional can do for fun. If you can think of any more please do not hesitate to add in the comments section.

1. Project Euler

This is one of my favorite problem solving websites. It is not explicitly Operations Research as more applied mathematics. Yet it is very interesting to try to determine ways to solve some mathematical, numerical, and logical word problems. It will more than likely require the use of some computer programming skills.

2. Netflix Prize

A couple of years ago Netflix, the movie DVD-to-the-door rental company, decided to open up to the public ways to improve its movie recommendation engine. The rules were very straight forward. Given a user defined list of movies find similar movies that the Netflix users might pick. Netflix would provide a pretty comprehensive data set of user lists. Numerous people and organizations have provided suggestions. There is even a leaderboard based on RMSE. Oh yea, and the prize is $1,000,000. That ought to get you interested.

3. Volunteering with High School Operations Research

High School OR is a volunteer organization that helps to bring Operations Research to the high school level. High School OR develops and maintains workshops for high school teachers so that they may use Operations Research in the classroom. This is a great way to get the Operations Research profession to the younger generations. High School Operations Research is sponsored by INFORMS.

4. Math tutoring

The Operations Research professional can use their skills to help elementary, high school and even college students in mathematics. This can be an excellent way to contribute to the community. Perhaps even to make a little more money on the side. There will always be a need to provide math assistance since students are subjected to more and more testing. There can be a great intrinsic reward to helping those achieve their fullest potential. This can also be a great way to show how math can lead to a future profession, specifically Operations Research.

5. Contribute at the COIN-OR project

Open source software is growing in the Operations Research community. Paraphrasing from the COIN-OR website...
COIN-OR's goal is to help gather a community for software in Operations Research what open literature in the past did for mathematics.
There are many ways to help with COIN-OR open source projects. I describe these ways in an IEOR Tools post Top 5 Things to Get Involved with Open Source in Operations Research. There are many ways to contribute to Open Source projects. You need not be a programmer to get involved. You may be a debugger, tester, or even help with documentation. The ways are limitless to get involved. Open Source in Operations Research can be a fun way to learn new ways to solve those classic optimization problems.

Wednesday, June 3, 2009

Optimization with WolframAlpha metasearch site

Wolfram Alpha bursted onto the scene with a lot of fan-faire but little in know-how. There was a lot of interesting thoughts in the Operations Research blog world as far as potential applications. I decided to give it a try and wouldn't you know that with the "optimization" keyword search term gave me a really interesting result

http://www45.wolframalpha.com/input/?i=optimization

Apparently Wolfram Alpha will return several algorithm resources to find global, constrained, and local points on a graph. Sure this isn't mathematical programming but I bet we can search further.

Second I used the keyword "linear programing" and got the following return.

http://www45.wolframalpha.com/input/?i=linear+programming


Wolfram Alpha returned a mathematica function for applying linear programming algorithm. I put in an example from the Definition site and it return a set of optimized values. This is very interesting indeed. Of course this is very simple but it would be really interesting to benchmark the Wolfram|Alpha website for some more complex formula and data sets.

I am very intrigued by the possibilities this could lead to for the Operations Research community. This is a free service provided by Wolfram|Alpha. Could there be more community-based or "cloud" services for the Operations Research community. Currently the NEOS Server project provides a free optimization solver. Wolfram|Alpha has a variety data sources as well as other mathematical formulations. Perhaps this is the future of optimization solvers.