[Coin-lpsolver] Some problems about LP solver

[Lei Chen]

Hi,
    I am currently using CLP as a part of my research project. I have a collection of LP problems and try to solve them as fast as possible. So I try to exploite the relations among LPs tasks to speed up the performance. Now I have the following question:

    1. If there are two LP tasks, The second only has one more row constraint than the first one, how can I speed up the second LP after the first one is done? 
            This question has already been answered by John (Thanks!). We can simply add one more row and use dual algorithm. Then the second LP will be a lot faster.

    2. If I have a collection of LP, all of them shares the same matrix A, the column lower bound, and column higher bound. The only difference are the row lower bound and higher bound. How can I take advantage of this fact to minimize the total cost of all these LP tasks.

    3. Since the solution space of a LP is convex. Is there any fast way to find all the extreme points of the convex solution space?

    I know that some of the question might seen to be amateurish. My research major is in Database rather than Optimization. But I really appreciate if some one can give me any clue about the above question. Thanks in advance.

Lei Chen

        
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/clp/attachments/20030917/f989e834/attachment.html>