Actually, there was a study done that shows that for most traffic conditions, even heavy traffic, randomizing the lights allows greater throughput. Syncronized lights only help if the traffic is very light, or if it is very, very heavy. However, the very heavy case is also prone to unpredictable behavior, as you might imagine, so it's not that great. So, everyone using these things to change lights will randomize them, and help overall throughput.
Pretty simple optimization problem
on
Am I Hot or Not
·
· Score: 1
We did this stuff at a place I worked called AspenTech (http://www.aspentech.com). I think the product is now called "Aspen Plus Online." When I worked on it it was called RT-Opt (Real Time Optimizer). Of course, there we were optimizing entire chemical plants -- say 3000 independant variables in a 250000 equation system. Heating a house would be a pretty trivial optimization problem. Of course Aspen's software cost, say, a quarter million US for an installation (including consulting to customize the plant model). Out of the range of most homeowners, unfortunately. There's tons of chemical engineering literature on these problems, for those interested.
Slashdot Mirror
Actually, there was a study done that shows that for most traffic conditions, even heavy traffic, randomizing the lights allows greater throughput. Syncronized lights only help if the traffic is very light, or if it is very, very heavy. However, the very heavy case is also prone to unpredictable behavior, as you might imagine, so it's not that great. So, everyone using these things to change lights will randomize them, and help overall throughput.
We did this stuff at a place I worked called AspenTech (http://www.aspentech.com). I think the product is now called "Aspen Plus Online." When I worked on it it was called RT-Opt (Real Time Optimizer). Of course, there we were optimizing entire chemical plants -- say 3000 independant variables in a 250000 equation system. Heating a house would be a pretty trivial optimization problem. Of course Aspen's software cost, say, a quarter million US for an installation (including consulting to customize the plant model). Out of the range of most homeowners, unfortunately. There's tons of chemical engineering literature on these problems, for those interested.