DMPlexCreateSection expects that the boundary index sets its given are
non-null. However, if these are retrieved via DMPlexGetStratumIS from
a label generated by DMPlexMarkBoundaryFaces, the index sets will come
out null, causing the following error.
In my case this was caused by a bug: my DMPlex
Geoffrey Irving writes:
> Was there any particular reason for making DMPlexMarkBoundaryFaces a
> quadratic time algorithm? I realize it's hard to write a subquadratic
> time algorithm on top of DMLabelSetValue; maybe PETSc needs some basic
> integer hash tables?
We have actually hash tables and
Matthew Knepley writes:
> Why is it quadratic time? I just looked again, and it seems to be linear
> time to me.
Look at DMLabelSetValue(). If the point is not labeled yet
(PetscFindInt returns negative) then you call PetscMemmove(), which is
linear complexity in general. It's not really quadra
On Wed, Nov 20, 2013 at 6:44 PM, Matthew Knepley wrote:
> On Wed, Nov 20, 2013 at 8:38 PM, Geoffrey Irving wrote:
>>
>> On Wed, Nov 20, 2013 at 6:30 PM, Matthew Knepley
>> wrote:
>> > On Wed, Nov 20, 2013 at 7:22 PM, Geoffrey Irving wrote:
>> >>
>> >> Was there any particular reason for making
On Wed, Nov 20, 2013 at 6:22 PM, Matthew Knepley wrote:
> On Wed, Nov 20, 2013 at 6:58 PM, Geoffrey Irving wrote:
>>
>> On Wed, Nov 20, 2013 at 3:49 PM, Matthew Knepley
>> wrote:
>> > On Tue, Nov 19, 2013 at 4:35 PM, Geoffrey Irving wrote:
>> >>
>> >> The documentation for DMPlexGetStratumIS he
On Wed, Nov 20, 2013 at 8:38 PM, Geoffrey Irving wrote:
> On Wed, Nov 20, 2013 at 6:30 PM, Matthew Knepley
> wrote:
> > On Wed, Nov 20, 2013 at 7:22 PM, Geoffrey Irving wrote:
> >>
> >> Was there any particular reason for making DMPlexMarkBoundaryFaces a
> >> quadratic time algorithm? I realiz
On Wed, Nov 20, 2013 at 6:30 PM, Matthew Knepley wrote:
> On Wed, Nov 20, 2013 at 7:22 PM, Geoffrey Irving wrote:
>>
>> Was there any particular reason for making DMPlexMarkBoundaryFaces a
>> quadratic time algorithm? I realize it's hard to write a subquadratic
>> time algorithm on top of DMLabe
On Wed, Nov 20, 2013 at 7:22 PM, Geoffrey Irving wrote:
> Was there any particular reason for making DMPlexMarkBoundaryFaces a
> quadratic time algorithm? I realize it's hard to write a subquadratic
> time algorithm on top of DMLabelSetValue; maybe PETSc needs some basic
> integer hash tables?
>
On Wed, Nov 20, 2013 at 6:58 PM, Geoffrey Irving wrote:
> On Wed, Nov 20, 2013 at 3:49 PM, Matthew Knepley
> wrote:
> > On Tue, Nov 19, 2013 at 4:35 PM, Geoffrey Irving wrote:
> >>
> >> The documentation for DMPlexGetStratumIS helpfully explains that the
> >> "value" argument is
> >>
> >> -
Was there any particular reason for making DMPlexMarkBoundaryFaces a
quadratic time algorithm? I realize it's hard to write a subquadratic
time algorithm on top of DMLabelSetValue; maybe PETSc needs some basic
integer hash tables?
Note: boundaries are not necessarily sublinear in the case of stro
On Wed, Nov 20, 2013 at 3:49 PM, Matthew Knepley wrote:
> On Tue, Nov 19, 2013 at 4:35 PM, Geoffrey Irving wrote:
>>
>> The documentation for DMPlexGetStratumIS helpfully explains that the
>> "value" argument is
>>
>> - value - The stratum value
>
> This is exactly right. A DMLabel is a map f
On Wed, Nov 20, 2013 at 10:38 AM, Geoffrey Irving wrote:
> On Tue, Nov 19, 2013 at 10:32 PM, Jed Brown wrote:
> > Geoffrey Irving writes:
> >> I was missing the obvious INSERT_BC_VALUES parameter in the call to
> >> DMPlexProjectFunctionLocal. It's a shame one has to loop over the
> >> interio
On Tue, Nov 19, 2013 at 5:27 PM, Geoffrey Irving wrote:
> This question isn't about changing ex12, I'm just making sure I
> understand what's going on:
>
> SetupSection in snes ex12 calls DMPlexLabelComplete on the "marker"
> label if Dirichlet conditions are chosen (not if Neumann is chosen).
>
On Tue, Nov 19, 2013 at 4:35 PM, Geoffrey Irving wrote:
> The documentation for DMPlexGetStratumIS helpfully explains that the
> "value" argument is
>
> - value - The stratum value
>
This is exactly right. A DMLabel is a map from value --> {set of integers}.
There
are no builtin semantics on
Rob,
Thanks.
But since SuperLU_dist is freely distributed doesn’t this mean one could
always just get SuperLU_Dist whenever they wanted to use MC64AD? Note the
SuperLU_Dist license
http://crd-legacy.lbl.gov/~xiaoye/SuperLU/License.txt states that SuperLU_Dist
can be included in o
Hi Barry,
We've already asked Sherri about this, and she says she has permission to use
it in DSuperLU. It's in hypre because we currently bundle DSuperLU in our
release.
-Rob
> -Original Message-
> From: Barry Smith [mailto:bsm...@mcs.anl.gov]
> Sent: Wednesday, November 20, 2013 2:2
On Nov 20, 2013, at 4:08 PM, Matthew Knepley wrote:
> On Wed, Nov 20, 2013 at 4:05 PM, Barry Smith wrote:
>
> Matt,
>
>Since it “is in” SuperLU then you don’t need to make a package for it,
> simply require the user link in with superlu to use those functions.
>
> I f2c'd it so I cou
Sherri and Rob,
I noticed both of you guys distribute MC64AD with your libraries which
includes the following copyright/license statement. Did you receive permission
to distribute this source code from Iain or Jacko and how general is that
permission, just for you or for other groups a
Matthew Knepley writes:
> 1) It is in "some branch" but pushing to next is exactly how we test.
'next' is for things that we think will go to 'master', but this is
something that should _never_ go to 'master' lest our repository history
forever distributes something that we don't have a license
On Wed, Nov 20, 2013 at 4:05 PM, Barry Smith wrote:
>
> Matt,
>
>Since it “is in” SuperLU then you don’t need to make a package for it,
> simply require the user link in with superlu to use those functions.
>
I f2c'd it so I could actually see what was happening.
>For testing you sh
Matt,
Since it “is in” SuperLU then you don’t need to make a package for it,
simply require the user link in with superlu to use those functions.
For testing you should have just put it in some branch, not in next. BTW: it
generates nightly build errors that I don’t want to deal with
Matthew Knepley writes:
> This http://www.hsl.rl.ac.uk/licencing.html tells me that I am not doing
> anything wrong until we release :)
It says "distribute", which the public repositories and nightly tarballs
would fall under.
> I want this there for testing. Since it is already in SuperLU, I w
On Wed, Nov 20, 2013 at 2:08 PM, Barry Smith wrote:
>
>Matt,
>
> You have included a bunch of HSL code in PETSc with uncertain
> copyright information; it is not clear that we can include it in PETSc. In
> hslmc64.c is the following incomplete fragment of copyright/license
> information:
Matt,
You have included a bunch of HSL code in PETSc with uncertain copyright
information; it is not clear that we can include it in PETSc. In hslmc64.c is
the following incomplete fragment of copyright/license information:
/* COPYRIGHT (c) 2007 Science and Technology Facilities Counc
On Tue, Nov 19, 2013 at 10:32 PM, Jed Brown wrote:
> Geoffrey Irving writes:
>> I was missing the obvious INSERT_BC_VALUES parameter in the call to
>> DMPlexProjectFunctionLocal. It's a shame one has to loop over the
>> interior just to set the boundary, but it does seem like the code
>> produce
25 matches
Mail list logo