On Fri, 14 Jan 2011, yunfei zhu wrote:
> I get warning like this:
> WARNING: Second derivatives are not currently correctly calculated on
> non-affine elements!
>
> I noted that this warning come from function FEBase::compute_map() in
> fe_map.C file.
>
> 00497 #ifdef LIBMESH_ENABLE_SECOND_DERIVA
On Fri, 3 Apr 2009, Roy Stogner wrote:
> I'd prefer to be conservative and just add it to --enable-everything
> for now,
I'm clearly being much too stodgy - looking at the actual code,
--enable-second is already part of --enable-everything and probably
has been for years...
---
Roy
I'd prefer to be conservative and just add it to --enable-everything
for now, but I'll consider myself outvoted; I'll make it default
shortly.
But if we're going for radical changes: we could go even farther and
mark the "call reinit(), but don't request data until afterwards"
idiom as deprecated
Me as well.
Derek
On Apr 3, 2009, at 3:52 PM, Kirk, Benjamin (JSC-EG311) wrote:
> I vote for default...
>
>
>
> - Original Message -
> From: Roy Stogner
> To: Derek Gaston
> Cc: libmesh-users@lists.sourceforge.net >
> Sent: Fri Apr 03 16:21:33 2009
I vote for default...
- Original Message -
From: Roy Stogner
To: Derek Gaston
Cc: libmesh-users@lists.sourceforge.net
Sent: Fri Apr 03 16:21:33 2009
Subject: Re: [Libmesh-users] Second Derivatives
On Fri, 3 Apr 2009, Derek Gaston wrote:
> So as long as you request _somethi
On Fri, 3 Apr 2009, Derek Gaston wrote:
> So as long as you request _something_ before doing reinit() you should
> be ok?
That's exactly right. I wanted to optimize FE::reinit() (which
includes some of the most significant inefficiencies in the library
for some of my problems) as much as possib
On Apr 3, 2009, at 2:50 PM, John Peterson wrote:
>
> Unfortunately, no (fe.C, line 236)
>
> // If the user forgot to request anything, we'll be safe and
> // calculate everything:
> #ifdef LIBMESH_ENABLE_SECOND_DERIVATIVES
> if (!calculate_phi && !calculate_dphi && !calculate_d2phi)
>calcula
On Fri, Apr 3, 2009 at 3:43 PM, Derek Gaston wrote:
> On Apr 3, 2009, at 2:43 PM, John Peterson wrote:
>
>> It should recompute stuff during reinit that you wouldn't otherwise
>> want to recompute.
>
> But only if I asked for it by calling get_d2phi() or associated beforehand,
> right?
Unfortunat
On Apr 3, 2009, at 2:43 PM, John Peterson wrote:
> It should recompute stuff during reinit that you wouldn't otherwise
> want to recompute.
But only if I asked for it by calling get_d2phi() or associated
beforehand, right?
Derek
On Fri, Apr 3, 2009 at 3:35 PM, Derek Gaston wrote:
> Does --enable-second actually slow anything else down (as in... does
> it slow down computations that don't use second derivatives?).
It should recompute stuff during reinit that you wouldn't otherwise
want to recompute.
--
John
---
10 matches
Mail list logo
