eigen on large models

[fmk]

we had thought of trying to deal with user error in that manner. It will deal with error case in which a user adds more than necessary. It is however an incorrect solution for the case where the user provides some incorrect node numbering in the element definitions. So while OpenSees could always provide an answer like these other codes, it is not going to be gauarenteed to be correct. We prefer for user error to be corrected by the user.

[johnnyontheweb]

Dear fmk,
thanks for your reply. I could agree with that, but I cannot share the reasons. I lost more than one month to catch this error and as I said, the comparison with other solvers (such as Abaqus and SAP) didn't help too much because they automatically correct these errors (that, let me say, are triavial in a large mesh!).
From this forum I noticed that the same error can be raised from an incorrect node numbering in element (as you reminded), I wrote a code that checks for counterclockise connectivity but it was not the solution for me (and the OS error in my opinion had to be different from the case of free nodes).

So, in these cases a warning can be useful to help users to know what's happening; otherwise, handling large models can be very difficult in OS.