From: Roy S. <roy...@ic...> - 2011-07-04 22:45:18
|
On Mon, 4 Jul 2011, Robert wrote: > I have some trouble with the accuracy of the displacement gradient in > moving mesh situations. Sadly that may not be the end of your troubles in moving mesh situations; I wrote that ALE code for an application which in the near-term ended up switching to a fixed-mesh model, and I wouldn't be suprised if there are still serious bugs. I'd like to help you fix whatever you encounter, but you ought to be warned ahead of time that that feature isn't expected to be fully stable and working right now. > I have attached a minimal example to reproduce my problem. It looks like the mailing lists stripped your attachment? Would you try again, and Cc: me in case it doesn't make it to the list again? There are a couple places in the code where we "cheat" and use finite differencing for gradients, but IIRC in those cases it's the Hex code that uses precise formulae and the Tet code that cheats. And I don't think we do any of that with the Lagrange FE, which is what you should be using for ALE geometry. --- Roy |