- firedrake - imperial.ac.uk

Hi everyone,
by Justin Chang 16 Jul '15

16 Jul '15

I would like to try out the Petrov-Galerkin Enrichment Method (Barrenechea et al 2007) for a mixed-poisson/Darcy equation.Here is the link to the paper: http://www.sciencedirect.com/science/article/pii/S0045782507000059 Basically, this method starts with the P1/P0 combination and enriches both velocity and pressure trial functions with multi-scale functions. Is this easily doable within Firedrake? Thanks, Justin

2 3

Re: [firedrake] possible lazy evaluation issue?
by Andrew McRae 15 Jul '15

15 Jul '15

One way is by environment variable: export PYOP2_LAZY=0 I don't know if there's a parameters[...] setting. On 15 July 2015 at 20:27, Cotter, Colin J <colin.cotter(a)imperial.ac.uk> wrote: > I don't know how to turn off lazy. I see that the problem also goes away > if I call it on u0 twice... > > cheers > --cjc > ------------------------------ > *From:* firedrake-bounces(a)imperial.ac.uk [firedrake-bounces(a)imperial.ac.uk] > on behalf of Andrew McRae [a.mcrae12(a)imperial.ac.uk] > *Sent:* 15 July 2015 20:19 > *To:* firedrake > *Subject:* Re: [firedrake] possible lazy evaluation issue? > > Low-hanging fruit: > > Does it go away if you turn off lazy? > > What happens if you call it on u0 twice in succession (with and without > lazy, I guess), do things change? > > On 15 July 2015 at 20:16, Cotter, Colin J <colin.cotter(a)imperial.ac.uk> > wrote: > >> Dear Firedrakers, >> I encountered some spooky behaviour when doing some things with >> handtooled kernels. It's in the following code: >> https://gist.github.com/colinjcotter/85106d0f11e2f7c93ede >> >> Sorry that there is so much. The point is that the function limit_slope >> (which does several things including executing some handtooled kernels) is >> (I claim) idempotent, however, on line 217, I call it on u0, then assign u0 >> to u1 in line 218. Then I call limit_slope on u1 in line 219. If I comment >> out line 219, I get different behaviour, in particular the bounds that are >> printed out go above 1.0 and below 0.0 on occasion. >> >> It seems like the assign is occurring before some of the kernels (or >> perhaps the final project) are executed in limit_slope. >> >> Any thoughts? >> >> cheers >> --cjc >> >> -- >> http://www.imperial.ac.uk/people/colin.cotter >> >> www.cambridge.org/9781107663916 >> >> >> >

2 1

Re: [firedrake] possible lazy evaluation issue?
by Andrew McRae 15 Jul '15

15 Jul '15

Low-hanging fruit: Does it go away if you turn off lazy? What happens if you call it on u0 twice in succession (with and without lazy, I guess), do things change? On 15 July 2015 at 20:16, Cotter, Colin J <colin.cotter(a)imperial.ac.uk> wrote: > Dear Firedrakers, > I encountered some spooky behaviour when doing some things with > handtooled kernels. It's in the following code: > https://gist.github.com/colinjcotter/85106d0f11e2f7c93ede > > Sorry that there is so much. The point is that the function limit_slope > (which does several things including executing some handtooled kernels) is > (I claim) idempotent, however, on line 217, I call it on u0, then assign u0 > to u1 in line 218. Then I call limit_slope on u1 in line 219. If I comment > out line 219, I get different behaviour, in particular the bounds that are > printed out go above 1.0 and below 0.0 on occasion. > > It seems like the assign is occurring before some of the kernels (or > perhaps the final project) are executed in limit_slope. > > Any thoughts? > > cheers > --cjc > > -- > http://www.imperial.ac.uk/people/colin.cotter > > www.cambridge.org/9781107663916 > > >

2 1

possible lazy evaluation issue?
by Colin Cotter 15 Jul '15

15 Jul '15

Dear Firedrakers, I encountered some spooky behaviour when doing some things with handtooled kernels. It's in the following code: https://gist.github.com/colinjcotter/85106d0f11e2f7c93ede Sorry that there is so much. The point is that the function limit_slope (which does several things including executing some handtooled kernels) is (I claim) idempotent, however, on line 217, I call it on u0, then assign u0 to u1 in line 218. Then I call limit_slope on u1 in line 219. If I comment out line 219, I get different behaviour, in particular the bounds that are printed out go above 1.0 and below 0.0 on occasion. It seems like the assign is occurring before some of the kernels (or perhaps the final project) are executed in limit_slope. Any thoughts? cheers --cjc -- http://www.imperial.ac.uk/people/colin.cotter www.cambridge.org/9781107663916

1 0

Enforcing slip boundary condition strongly
by Justin Chang 13 Jul '15

13 Jul '15

Hi everyone, I am solving a mixed poisson (aka: Darcy) problem using the VMS formalism (Masud and Hughes, CMAME, 2002) which enables one to use equal order lagrangian interpolation. In this formulation, the flux/velocity is prescribed strongly and the pressure is prescribed weakly, like how it's done with RT0 elements. The main difference is that velocity is now a VectorFunctionSpace as opposed to a FunctionSpace. My question is, how would I implement the flux boundary condition in a Dirichlet sense? Say I am working with a unit square for now. I want to enforce v_x = some value(s) for x=0/x=1, and v_y - some values(s) for y = 0/y = 1. How would this be done in Firedrake? Thanks, Justin

3 4

mpi problem
by Cotter, Colin J 13 Jul '15

13 Jul '15

Dear all, Not strictly a Firedrake problem but I hope you'll indulge me. I upgraded to 14.04 over the weekend, and now I get: In [2]: from mpi4py import * --------------------------------------------------------------------------- ImportError Traceback (most recent call last) <ipython-input-2-13eee3001ee9> in <module>() ----> 1 from mpi4py import * ImportError: libmpi.so.0: cannot open shared object file: No such file or directory This is true, there is indeed no libmpi.so.0 anywhere on the system. What should I do? cheers --cjc

1 1

Re: [firedrake] Enforcing slip boundary condition strongly
by David Ham 10 Jul '15

10 Jul '15

Correct. However his functionspace is vector Pn not RT0 On Fri, 10 Jul 2015 at 13:08 Cotter, Colin J <colin.cotter(a)imperial.ac.uk> wrote: > Justin is wanting to enforce the flux condition, he said. That's the > normal component. > > --cjc > > On 10 July 2015 at 10:25, David Ham <David.Ham(a)imperial.ac.uk> wrote: > >> Hi Colin, >> >> That does not appear to be what Justin is asking for. He is asking how >> to set a Dirichlet condition on one component of a VectorFunctionSpace. I'm >> not sure we can do that right now, but I think this should be fixable. I >> want to check what the Dolfin syntax for this is (so we maintain >> compatibility) and then try to fix it. (Of course someone might want to >> pleasantly surprise me and tell me we can do this already!) >> >> David >> >> On Fri, 10 Jul 2015 at 10:14 Cotter, Colin J < >> colin.cotter(a)imperial.ac.uk> wrote: >> >>> Hi Justin, >>> Just to be clear, are you talking about enforcing normal component of >>> the vector RT0 quantity to be zero? >>> >>> all the best >>> --cjc >>> >>> On 9 July 2015 at 21:28, Justin Chang <jychang48(a)gmail.com> wrote: >>> >>>> Hi everyone, >>>> >>>> I am solving a mixed poisson (aka: Darcy) problem using the VMS >>>> formalism (Masud and Hughes, CMAME, 2002) which enables one to use equal >>>> order lagrangian interpolation. In this formulation, the flux/velocity is >>>> prescribed strongly and the pressure is prescribed weakly, like how it's >>>> done with RT0 elements. >>>> >>>> The main difference is that velocity is now a VectorFunctionSpace as >>>> opposed to a FunctionSpace. My question is, how would I implement the flux >>>> boundary condition in a Dirichlet sense? Say I am working with a unit >>>> square for now. I want to enforce v_x = some value(s) for x=0/x=1, and v_y >>>> - some values(s) for y = 0/y = 1. How would this be done in Firedrake? >>>> >>>> Thanks, >>>> Justin >>>> >>>> _______________________________________________ >>>> firedrake mailing list >>>> firedrake(a)imperial.ac.uk >>>> https://mailman.ic.ac.uk/mailman/listinfo/firedrake >>>> >>>> >>> >>> >>> -- >>> http://www.imperial.ac.uk/people/colin.cotter >>> >>> www.cambridge.org/9781107663916 >>> >>> >>> >> _______________________________________________ >> firedrake mailing list >> firedrake(a)imperial.ac.uk >> https://mailman.ic.ac.uk/mailman/listinfo/firedrake >> >> > > > -- > http://www.imperial.ac.uk/people/colin.cotter > > www.cambridge.org/9781107663916 > > >

1 0

Re: [firedrake] Enforcing slip boundary condition strongly
by David Ham 10 Jul '15

10 Jul '15

Hi Colin, That does not appear to be what Justin is asking for. He is asking how to set a Dirichlet condition on one component of a VectorFunctionSpace. I'm not sure we can do that right now, but I think this should be fixable. I want to check what the Dolfin syntax for this is (so we maintain compatibility) and then try to fix it. (Of course someone might want to pleasantly surprise me and tell me we can do this already!) David On Fri, 10 Jul 2015 at 10:14 Cotter, Colin J <colin.cotter(a)imperial.ac.uk> wrote: > Hi Justin, > Just to be clear, are you talking about enforcing normal component of > the vector RT0 quantity to be zero? > > all the best > --cjc > > On 9 July 2015 at 21:28, Justin Chang <jychang48(a)gmail.com> wrote: > >> Hi everyone, >> >> I am solving a mixed poisson (aka: Darcy) problem using the VMS formalism >> (Masud and Hughes, CMAME, 2002) which enables one to use equal order >> lagrangian interpolation. In this formulation, the flux/velocity is >> prescribed strongly and the pressure is prescribed weakly, like how it's >> done with RT0 elements. >> >> The main difference is that velocity is now a VectorFunctionSpace as >> opposed to a FunctionSpace. My question is, how would I implement the flux >> boundary condition in a Dirichlet sense? Say I am working with a unit >> square for now. I want to enforce v_x = some value(s) for x=0/x=1, and v_y >> - some values(s) for y = 0/y = 1. How would this be done in Firedrake? >> >> Thanks, >> Justin >> >> _______________________________________________ >> firedrake mailing list >> firedrake(a)imperial.ac.uk >> https://mailman.ic.ac.uk/mailman/listinfo/firedrake >> >> > > > -- > http://www.imperial.ac.uk/people/colin.cotter > > www.cambridge.org/9781107663916 > > >

3 2

Re: [firedrake] Enforcing slip boundary condition strongly
by David Ham 10 Jul '15

10 Jul '15

I agree. I think that's what I was proposing. Are you on this or shall I do something? On Fri, 10 Jul 2015 at 12:14 Mitchell, Lawrence < lawrence.mitchell(a)imperial.ac.uk> wrote: > > > On 10 Jul 2015, at 11:37, David Ham <David.Ham(a)imperial.ac.uk> wrote: > > > > > > That's quite yuck, however I can see the attraction as we're kind of > sort of not mucking with the PyOP2 interface. > > > > However the generated code is specific to BCs applied to a particular > component, which doesn't satisfy Justin's (very standard) use case. > > > > I think we need to encode *which* component we're changing into the map > value. EG. > > > > val = -val -(2**(30-i)) > > > > where i is the component. This can be decoded to zero the correct entry > in a general purpose way. The downside is that we limit ourselves to about > half a billion nodes per core. I think that's OK. > > OK, so it may be that a node may have more than one component zeroed (e.g. > at a corner). > > So how about this: > > > Reserve top 2 (3?) bits to encode the component: > > x = sum(2**idx for idx in indices) > > val = val + (x << 29) > > Convert to negative > > val = -(val + 1) > > In the generated code we can then unwind that and pull out the appropriate > components. > > This could work. They're all pretty ugly solutions, tbf. > > Lawrence >

2 1

Archer environment
by David Ham 10 Jul '15

10 Jul '15

I notice that https://bitbucket.org/mapdes/archer-environment is empty. Is this intended or has something gone wrong? David

3 2