Page 2 of 2
Posted: Sat Nov 26, 2005 8:53 pm
Matrix #1
3
2
2
2
2
2
2
1

Posted: Sun Nov 27, 2005 8:21 am
thanks Rosti for your quick reply ... my code matches your output too... but still WA ... I am out of idea now

Posted: Mon Nov 28, 2005 7:39 am
I have read the description of mf's algorithm and understand how his example works.

But I still don't get how it resolves the issue of partitions like ...
()..()..() ().

From what I understand, whenever an interval is divided in two, one should consider whether it is possible to just append the last few matrices to the end of a sequence and evaluate those together.

That is, ()* last one,
or () * last two,
or () * last three

and so on and use the one that requires least amount of matrissors and within the capacity of matrissor.

But it doesn't consider multiple partitioning like the one I mentioned above, or does it?

Posted: Thu Dec 08, 2005 1:18 pm
I agree with you. Although I have known the algorithm to make it work, I do not know why it works. I thought of a lot of time.
I would like that someone who understands this problem well to explain it to us. (Why it works, and what is the way of thinking that leads to this solution).
Thanks a lot.