Hi all!
I'm running my FE code in matlab and now have a few questions to ask you guys.
I know there are algorithms to minimize stiffness matrix bandwidth. But I'm not using that in my code, and the bandwidth turned out to be nearly 1/4 of the matrix size n. Will this affect my calculation speed dramatically?
I'm doing this as a part of a project and will show the result later on. So If I want to show the computational work needed to calculate the inverse of stiffness matrix, can I use Cholesky method? (I wanna use Cholesky because I can calculate the exact number of operations involved.) I know other algorithms like Gauss and CG, but it's hard to evaluate the number of operations, as the number of iterations depend on the properties of stiffness matrix(like condition number, etc.). So any suggestions about what I should I should use in the report I'm writing?
Thanks all!
Best,
P
I'm running my FE code in matlab and now have a few questions to ask you guys.
I know there are algorithms to minimize stiffness matrix bandwidth. But I'm not using that in my code, and the bandwidth turned out to be nearly 1/4 of the matrix size n. Will this affect my calculation speed dramatically?
I'm doing this as a part of a project and will show the result later on. So If I want to show the computational work needed to calculate the inverse of stiffness matrix, can I use Cholesky method? (I wanna use Cholesky because I can calculate the exact number of operations involved.) I know other algorithms like Gauss and CG, but it's hard to evaluate the number of operations, as the number of iterations depend on the properties of stiffness matrix(like condition number, etc.). So any suggestions about what I should I should use in the report I'm writing?
Thanks all!
Best,
P