Any tips/comments regarding the latest version of the C++ library: Eigen (v. 3.0)?
1. A new version of Eigen (v 3.0 now) is out (on March 23, 2011), and it seems promising. First, a few links:
2. No compilation issues on the Win32 platform: I just downloaded it today, and found that it compiles OK with VC++ 10.0 on Win32 (the BUILD_ALL project). I haven't had the time to try anything further, though.
3. Please share your experience: For whatever reasons, if anyone here eventually decides to use it, or not to use it, or has any comments or tips to offer, I would like to hear from him/them, esp. in respect of the following points:
3.1 For eigenvalue computations, comparative advantages/disadvantages, including speed comparisons, with the existing libraries like ARPACK, Trilinios, etc.
Assume that the system to solve has origins in a typical structural/solid mechanics FEM. If possible, please share your experience for global [A] matrix orders of 3k, 10k, 100k, and bigger, the matrices being the typical FEM ones: sparse, symmetric, etc.
[This note is not really necessary, but just in the interest of clarity: If a matrix has 10 rows and 10 columns, then its order is taken to be 10, not 100.]
3.2 For solving a linear FEM-generated system using a direct solver, speed comparisons with the existing FORTRAN/C++ code as implemented in GOTO (latest version), Taucs, etc. Again, matrix orders go from 3k to 100k and bigger, and the matrices, of course, are sparse, symmetric, etc.
4. Any other tips you care to share would also be appreciated.
Thanks in advance.