[Coin-discuss] Experiences with paralleling simplex
Matthew Saltzman
mjs at clemson.edu
Sun Jan 7 07:31:43 EST 2007
On Fri, 5 Jan 2007, Sathi wrote:
>
>> Some kind
>> of sparse LU factorization is typical.
>
> In case of Cholesky factorization one could use column ordering heuristics
> like minimum-degree to get a better factorization. Does the LU
> factorization depend upon the column ordering like Cholesky factorization?
In unsymmetric LU factorization, the sparsity pattern and column selection
order can (and usually does) change based on the magnitudes of the
coefficients involved at each pivot. A fixed ordering cannot be computed
in advance. See, e.g, Markowitz ordering. Also, updating a sparse
factorization at each pivot is much faster in general than recomputing it.
In interior-point algorithms, pivot order is stable from iteration to
iteration, but there is no exploitable relationship between the
coefficents of consecutive iterations, so factorizations must be
recomputed from scratch.
>
> Thanks,
> Sathi
>
>
>
> _______________________________________________
> Coin-discuss mailing list
> Coin-discuss at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-discuss
>
--
Matthew Saltzman
Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs
More information about the Coin-discuss
mailing list