I would argue that they should only be members for SNESVI.
Todd.
> On May 14, 2016, at 10:46 PM, Matthew Knepley wrote:
>
> If I call this on a regular SNES, like newtonls, I get a memory leak
> because it defaults to the VI version, but SNESReset() only uses
> the newtonls
Presumably. Argonne lost power during a thunderstorm and they are trying
to bring the site back up.
Todd.
> On Jul 28, 2016, at 2:35 PM, Matthew Knepley wrote:
>
> Matt
>
> --
> What most experimenters take for granted before they begin their experiments
> is infinitely
> If you did this for the TAO user API then one could argue it should be done
> for TS also. But I don't think it should be done for TS.
>
> Two reasons for not having a separate API for linear and nonlinear problems
> is
> 1) I may want to use a "nonlinear" ODE integrator (that is an
TAO_BCO_TRON, and TAO_BQP_GPCG.
>>>
>>> Thoughts?
>>>
>>> Todd.
>>>
>>>
>>>> On Aug 16, 2016, at 4:32 PM, Oxberry, Geoffrey Malcolm <oxber...@llnl.gov>
>>>> wrote:
>>>>
>>>> Definitel
te:
>
> Definitely flags for linear problems would be helpful for TAO. Once there is
> an example up, I'd be happy to add that to the SQPTR pull request.
>
>
> From: petsc-dev-boun...@mcs.anl.gov [petsc-dev-boun...@mcs.anl.gov] on behalf
&g
talk about
> it to someone not at ANL. As someone who is interested in both contributing
> to TAO and using it in research, this question is very interesting to me.
>
> Best,
> Geoff
>
> From: Munson, Todd [tmun...@mcs.anl.gov]
>
outside the box and there is no guarantee
that zero is within the box.
Todd.
> On Aug 18, 2016, at 9:00 AM, Matthew Knepley <knep...@gmail.com> wrote:
>
> On Thu, Aug 18, 2016 at 8:55 AM, Munson, Todd <tmun...@mcs.anl.gov> wrote:
>
> Just to add some more thoughts to this ar
t of this particular comment that the user claims their
> model is quadratic? Of course evaluating at 0 has no special meaning
> for a general nonlinear model.
>
>>> On Aug 18, 2016, at 11:02 AM, Munson, Todd <tmun...@mcs.anl.gov> wrote:
>>>
>>>>
>&
>> Internally the solvers can do lots of stuff. Most interior point
>> methods, for example, would convert any inequality constraints
>> into equality constraints by adding slack variables and
>> bounding the slacks. That way they only need bounds on
>> variables.
>
> Yes; for instance,
For now, I am not proposing interface changes, but rather answering the
question of what types of problems do we need to support. We can
discuss actual interfaces later.
Note: you really only need one multiplier for the each of the
constraints (maybe interior-point methods are different).
The
>> For now, I am not proposing interface changes, but rather answering the
>> question of what types of problems do we need to support. We can
>> discuss actual interfaces later.
>>
>> Note: you really only need one multiplier for the each of the
>> constraints (maybe interior-point methods
I don't see any reason not to make the change.
Todd.
> On Aug 21, 2016, at 9:40 PM, Barry Smith wrote:
>
>
> Todd,
>
>The TAO solvers that utilize KSPNASH STCG or GLTR have all this horribly
> redundant unneeded code
>
> /* Solve the trust region subproblem
>>> As far as I know, the only information needed in the Hilbert space approach
>>> is matrices denoting inner products and norms.
>>
>> The norms are not a problem, especially just the norm you use to decide if
>> you have "converged". The trickier issue is "the inner products", which
>>
>> I do not understand this. The developer API will be the same for all
>> solvers. What a solver does internally to reformulate is its own
>> problem. I am more worried about providing consistent
>> information to the users.
>
> As far as I understand it, there is no API for developing
> SQPTR needs some work. I wrote it over 14 days; mostly whatever time I could
> spare at ICSP, plus three weekends. Right now, it’s written so that various
> steps of the algorithm could be factored out and tested individually using a
> handwritten test harness. The performance could be
Hi!
The only provable way to test for linearity of a function is using
code analysis.
A quick hack that should work okay is to evaluate the gradient at
two (or more) random points and calculate the distance between
the gradients. Note: this is not a good test for piecewise
linear functions,
>> Please do. I’m interested in contributing methods to TAO because I like
>> using PETSc, it seems like the most expedient way to get a solver my project
>> can use with features we need, and I have great respect for the software you
>> guys put out. If there is anything I can do to expedite
If the user supplied only a g(x) to a solver meant for
quadratic problems, we would call grad[g(0)] [thanks
Matt!] to get the linear term and hess[g(0)] to get
the Hessian matrix. So that is possible as well.
Maybe we should just keep things as they exist now
and not add extra complications.
>
> People are free to use MatShell to create a "matrix" that is actually a
> nonlinear operator. Solvers won't work properly if it's not, but that's
> their problem.
The quadratic programming solvers in our case will happily go and tell
you it solved the problem...however, the problem it
We are in agreement. I should not use double negatives.
This change should be made.
Todd.
> On Aug 21, 2016, at 11:39 PM, Barry Smith <bsm...@mcs.anl.gov> wrote:
>
>
>> On Aug 21, 2016, at 11:33 PM, Munson, Todd <tmun...@mcs.anl.gov> wrote:
>>
>>
I will start making this simplification today in a new branch.
Todd.
> On Aug 22, 2016, at 6:18 AM, Munson, Todd <tmun...@mcs.anl.gov> wrote:
>
>
> We are in agreement. I should not use double negatives.
>
> This change should be made.
>
> Todd.
>
>&
I am confused by Blaise's statement. "selected_real_kind(5)" is well defined
across
compilers; it returns an integer value that can be used in the kind statements
and
selected real numbers with at least 5 digits of precision.
With the exception of -1, -2, and -3 (plus -4 and -5 in Fortran
Cool!
For what its worth, that only works when PETSC_USE_FORTRANKIND is true.
I'm not sure how many of the Petsc builds have this flag set.
Any idea how to do the same thing in C?
Todd.
> On Sep 1, 2016, at 4:13 PM, Lisandro Dalcin wrote:
>
>
> On 1 September 2016 at
>> I'd like to be able to do something like the following in Fortran:
>>
>> PetscReal c
>>
>> c = PetscRealCast(5.78D0)
>>
>> where PetscRealCast is real() for single precision and dble()
>> for double precision. Without the cast, you would get all
>> kinds of warnings of the form
Dear petsc developers,
Please do not get too mad at me...
I have a C and Fortran test example that should be exactly the same
problem. The numerical methods, however, produce different results.
I traced this back to the literals used to define constants in the
test Fortran problem. C
The TS constants are weird in the code. Some of have enough digits for double,
some for quad, and some for quad double. I suspect mathematica output some
of the constants.
Todd.
> On Sep 1, 2016, at 10:29 PM, Jed Brown <j...@jedbrown.org> wrote:
>
> "Munson, Todd&q
One of the common concepts for TAO and SNES is the quasi-Newton approximations.
SNES seems to only use them in SNESQN (for non-symmetric matrices) and TAO uses
them in TAOLMVM and TAOBLMVM (for symmetric matrices). TOA also allows them to
be used as a preconditioner for the Hessian-based
used, I’m not sure it
> makes sense anymore.
>
> Geoff
>
> From: <petsc-dev-boun...@mcs.anl.gov> on behalf of Matthew Knepley
> <knep...@gmail.com>
> Date: Tuesday, August 30, 2016 at 2:02 PM
> To: "Munson, Todd" <tmun...@mcs.anl.gov>
> Cc:
30, 2016, at 9:26 PM, Oxberry, Geoffrey Malcolm <oxber...@llnl.gov>
> wrote:
>
>
>
> On 8/30/16, 12:36 PM, "Munson, Todd" <tmun...@mcs.anl.gov> wrote:
>
>>
>> For preconditioners, I use them as a default for unconstrained and bound
>>
Geoff,
How are they modeling and solving their PDEs?
Todd.
>> € Transforming Additive Manufacturing through Exascale Simulation
>> (TrAMEx), John Turner (ORNL) with LLNL, LANL, NIST
>
> PETSc did not come up in TrAMEx meetings I was involved in, and I do not
> believe it will be used. I
The write statement return type is ssize_t. For whatever reason its a signed
type
because it can return -1 on error.
I would suggest using the ssize_t type for both and changing the "!=" to a "<"
which will abort if write either returns a negative or the number of bytes
written is not
You can set up TAO to solve such a problem.
However, your problem boils down to solving the
linear system
w - V*lambda = 0
V'*w = b
Taking the Schur complement with respect to w, you get
the system
V'*V*lambda = b
You then form and invert V'*V, which is a 5x5 matrix
and
Did we come to a conclusion on this? Is it worthwhile adding macros for C and
Fortran literals to make sure they match the PetscScalar type along with the
appropriate configure tests?
Todd.
> On Sep 3, 2016, at 3:18 PM, Lisandro Dalcin wrote:
>
>
> On 2 September 2016
.
> On Aug 16, 2016, at 7:03 PM, Barry Smith <bsm...@mcs.anl.gov> wrote:
>
>
>> On Aug 16, 2016, at 6:09 PM, Munson, Todd <tmun...@mcs.anl.gov> wrote:
>>
>>
>> Is there a reason to put flags into PETSc for linear or nonlinear equations?
>>
ey Malcolm <oxber...@llnl.gov>
>>> wrote:
>>>
>>> Definitely flags for linear problems would be helpful for TAO. Once there
>>> is an example up, I'd be happy to add that to the SQPTR pull request.
>>>
>>> _________
> On Sep 22, 2017, at 7:35 AM, Matthew Knepley wrote:
>
> On Fri, Sep 22, 2017 at 12:06 AM, Richard Tran Mills wrote:
> Thanks for sharing this, Barry. I haven't had time to read their paper, but
> it looks worth a read.
>
> Hong, since many
Hi,
I am not sure that this linesearch is implemented in PETSc itself. To implement
it, you would need the ability to perform projections onto your feasible region,
which is what I am assuming you mean by undershoot or overshoot.
Once the projections are involved, however, there are lots of
I had a question. There is VecGetArrayRead and VecRestoreArrayRead. However,
the
sameconcept does not appear with IS. Is there a reason for not having
ISGetIndicesRead and ISRestoreIndicesRead?
Thanks, Todd.
> On Nov 8, 2017, at 10:32 AM, Matthew Knepley <knep...@gmail.com> wrote:
>
> On Wed, Nov 8, 2017 at 11:21 AM, Munson, Todd <tmun...@mcs.anl.gov> wrote:
>
> I had a question. There is VecGetArrayRead and VecRestoreArrayRead.
> However, the
> We could consider this, but really IS is intended to be a static object, not
> a dynamic one.
> I think Jed's SegBuffer is supposed to be dynamic, and could underlie a
> dynamic index
> object.
I could be wrong, but my recollection is that the initial reason for the index
sets was the
> On May 8, 2018, at 1:02 PM, Smith, Barry F. <bsm...@mcs.anl.gov> wrote:
>
>
>
>> On May 8, 2018, at 12:30 PM, Munson, Todd <tmun...@mcs.anl.gov> wrote:
>>
>>
>> Hi,
>>
>> I wonder if anyone has experience calling petsc
ot;.
Todd.
> On May 8, 2018, at 2:21 PM, Smith, Barry F. <bsm...@mcs.anl.gov> wrote:
>
>
> Push the branch that does this or send me the code and I'll debug it. I now
> have some vague memory about this behavior.
>
>Barry
>
>
>> On May 8, 2018, at
get started, type one of these: helpwin, helpdesk, or demo.
> For product information, visit www.mathworks.com.
>
>>> taopounders()
>>> exit
> ~/Src/petsc/src/tao/leastsquares/examples/matlab
> (tmunson/tao-pounders-matlab-interface=) arch-matlab
>
>
>> On
Hi,
I wonder if anyone has experience calling petsc from matlab using mex. I have
the
simplest possible mexFunction that just calls PetscInitialize() and
PetscFinalize(). That part seems to work fine, but when I exit
from matlab I get a segmentation violation and a petsc error
message.
I
oving accounts, such as
> Satish's, so testing gets difficult.
>
>But yes, we want to support Titan so have users send
> configure.log/make.log to petsc-ma...@mcs.anl.gov
>
>Barry
>
>
> > On Jul 6, 2018, at 12:45 PM, Munson, Todd wrote:
> >
&
I would be happy with one matrix type and removing the convenience functions
from the code. I am surprised that the dark lord did not mention it when
he approved the pull request...
Matt: the difference in the subtypes is that they are using different formulas
that result in different
There is a bit of code in TAO that allows the user to change the norm to
a matrix norm. This was introduced to get some mesh independent
behavior in one example (tao/examples/tutorials/ex3.c). That
norm, however, does not propagate down into the KSP methods
and is only used for testing
, then another
> implementation that is based on passing in a matrix, maybe one based on
> passing in a vector. Then each solver would have a XXXSetInnerProduct()
> while defaulting to l2.
>
> Barry
>
>
>> On Apr 12, 2018, at 12:21 PM, Munson, Todd <tmun...@mcs.a
on discrete quantities is DM, so my first inclination is that
>> any continuous interpretation of vectors, including inner products and
>> norms, belongs in DM.
>>
>> "Munson, Todd" <tmun...@mcs.anl.gov> writes:
>>
>>> There is a bit of code
supported only for the TAO quasi-Newton method?
Todd.
> On Apr 13, 2018, at 9:11 PM, Oxberry, Geoffrey Malcolm <oxber...@llnl.gov>
> wrote:
>
>
>
> On 4/12/18, 15:06, "petsc-dev on behalf of Munson, Todd"
> <petsc-dev-boun...@mcs.anl.gov on beh
The next milestone that includes a PETSc release is due on
September 30, 2018. Consider yourself warned. :)
Todd.
> On Mar 23, 2018, at 12:21 PM, Jed Brown wrote:
>
> I think it's feasible, but are we ever going to have more than a week's
> notice so that we can have a
I am not sure that VecGetSubVec and MatGetSubMat are needed. These should only
be used
for the active-set methods (I hope) and there should be native PETSc
equivalents used
by the PETSc active-set variational inequality solvers, for example.
Todd.
On Feb 18, 2014, at 9:22 AM, Jed Brown
Sounds like someone needs to look up the definition of customer support.
On Feb 21, 2014, at 1:20 PM, Jeff Hammond jeff.scie...@gmail.com wrote:
Aron told me that I should have said we look forward to your patch
earlier instead of causing you guys to go apoplectic over my different
Can you create a copy of the /usr/bin/env executable in your home directory and
call that?
I wonder if its based on the directory where the executable resides. /usr is
"protected".
Todd.
> On Oct 7, 2015, at 4:36 PM, Jed Brown wrote:
>
> Satish Balay
Satish can you try it from an account that does not have admin privileges?
Todd.
> On Oct 7, 2015, at 4:08 PM, Jed Brown wrote:
>
> Satish Balay writes:
>
>> And it works fine from C code.
>
> Yes, but /usr/bin/python is also C code, so how does the OS
e are other
> issues that I'm not considering?
>
> Il giorno Mer 12 Dic 2018, 18:46 Munson, Todd ha
> scritto:
>
> Yes; the optimization problems do not make sense if the objective function is
> from C^n to C, as there is not a natural ordering for complex numbers. If
> your object
Yes; the optimization problems do not make sense if the objective function is
from C^n to C, as there is not a natural ordering for complex numbers. If
your objective is from C^n to R (and the constraints are from C^n to R^m),
then the problem can be well defined. Then we need to get into
Hi,
Let me include the pets-dev team in the discussion. TAO uses the PETSc
matrices, vectors, and linear solvers our operations. Some of the
operations already have GPU implementations.
In terms of the unconstrained solvers, the Newton methods are based on
the linear conjugate gradient
One question from me would be, will any of the exascale machines support Julia
when they are delivered?
Todd.
> On Mar 11, 2019, at 9:28 PM, Smith, Barry F. via petsc-dev
> wrote:
>
>
> PETSc source code is becoming an unmaintainable, unextendable monstrosity.
> How long until Julia is
Barry F. via petsc-dev
>> wrote:
>>
>>
>> There is no particular reason they shouldn't.
>>
>> In fact, we should ask for an uplift to prepare Julia for exascale if
>> someone hasn't already ;)
>>
>> Barry
>>
>>
>>> On Mar 11, 2019, at 9:35 PM,
Plus no mention of MPI. I wonder if that is an oversight or if we should ditch
MPI as well...
Todd.
> On May 7, 2019, at 10:25 AM, Smith, Barry F. via petsc-dev
> wrote:
>
> PETSc developers,
>
> Note that this machine will apparently not support either of our two
> current GPU
There are some side effects when loading shared libraries, such as
initializations of
static variables, etc. Is something like that happening?
Another place is the initial runtime library that gets linked (libcrt0 maybe?).
I
think some MPI compilers insert their own version.
Todd.
> On
Hi all,
I forgot to send a reminder earlier. Our next tutorial will be October 15 at
2pm central. Barry will be leading a tutorial on pull requests from preparing
the request to what happens during the review and after the pull request is
accepted. Please email Barry directly if you have
Hi,
Thank you to Patrick for giving his tutorial on sphinx. The slides should soon
be available at:
https://docs.petsc.org/en/latest/developers/documentation/#sphinx-documentation
Our next tutorial will be October 15 at 2pm central. Barry will be leading a
tutorial on pull requests from
Hi all,
Patrick Sanan will be giving a tutorial on sphinx for the petsc developers on
October 1 at 2pm central. The information to join the call is below. This is
timely as the petsc documentation is being moved to use sphinx. Hope to see
those interested at the tutorial.
Thanks, Todd.
Hi all,
A reminder that Patrick Sanan will be giving a tutorial on sphinx for the petsc
developers today, October 1 at 2pm central. The information to join the call
is below. This is timely as the petsc documentation is being moved to use
sphinx. Hope to see those interested at the
Dear all,
A reminder that our next PETSc tutorial will be by Scott Kruger who will be
talking about the test harness, which will occur on October 29 at 2pm central.
Please email Scott directly if you have questions that you would like him to
address in the tutorial. The information for the
Hi Pierre,
I would say the answer to that question is "no, its not expected".
We will look into fixing it. It seems like the default tolerances are being
set to zero and the result is an inability to satisfy the constraints or
gradient of the Lagrangian to that small of a tolerance.
Hi all,
Due to supercomputing workshops and many conflicts, including Thanksgiving, we
will be cancelling the normal biweekly meeting for Thursday November 12 and 24.
Our next meeting will be December 10 at 2pm central.
Thanks, Todd.
Dear all,
Our next PETSc tutorial will be by Scott Kruger who will be talking about the
test harness, which will occur on October 29 at 2pm central. Please email
Scott directly if you have questions that you would like him to address in the
tutorial. The information for the call is below.
I want to note that crusher is early access hardware, so we should expect
performance to not be great right now. Doing what we can to help identify the
performance issues and keeping OLCF informed would be the best.
Note that we cannot make any of the preliminary results publicly available
I guess one could use tags to indicate the complexity of the review?
I personally would like a weekly digest of MRs that I have been requested to
review with clickable links. The dashboard would also be helpful.
All the best, Todd.
From: petsc-dev on behalf of Matthew Knepley
Date:
Hi all,
We are seeking postdoctoral candidates to work at Argonne on research and
development efforts in the areas of robust and efficient algebraic solvers,
time integrators, numerical optimizers, or related technologies in the
PETSc/TAO library that target exascale supercomputers and other
73 matches
Mail list logo