June 2020 - firedrake - imperial.ac.uk

MatNest and blockSize
by Karin&NiKo 12 Jul '20

12 Jul '20

Dear Firedrakers, I am solving a 3 fields coupled problem with Firedrake. I would like to use a multiplicative block-preconditioner but PETSc complains about a local size not being correct. [0] Arguments are incompatible [0] Local size 9 not compatible with block size 2 I have noticed that the use of a MatAij solves the problem (see comment in line 132) but I am afraid this is just a side effect. Am I missing something? Thanks in advance, Nicolas

2 6

Firedrake meeting this week 1600 BST Thursday (1500 UTC)
by Ham, David A 22 Jun '20

22 Jun '20

Dear all, This week is a little messed around for some of us as it’s the final week of term at many UK universities. Accordingly, the Firedrake meeting this week will be at 1600 London time on Thursday (1500 UTC). As usual, it will be held on the Zoom link below, and the agenda and minutes will appear here: https://github.com/firedrakeproject/firedrake/wiki/Minutes-and-agenda-of-Fi… The zoom link is: https://durhamuniversity.zoom.us/j/93560939004?pwd=bllyanZBblpsU2R0azVKUnlt… Cheers, David -- David Ham Department of Mathematics, Imperial College London https://www.imperial.ac.uk/people/david.ham https://www.firedrakeproject.org

1 0

Re: [firedrake] Fwd: Point evaluation / PointNotInDomainError
by Andrew McRae 14 Jun '20

14 Jun '20

On Sun, 14 Jun 2020 at 11:43, Rodrigo Piccinini <rbpiccinini(a)gmail.com> wrote: > Rebuilding Firedrake solved the problem. Thanks! > > Em sáb, 13 de jun de 2020 23:30, Rodrigo Piccinini <rbpiccinini(a)gmail.com> > escreveu: > >> Hi, Andrew. Thanks for the help! >> I'll try to rebuild Firedrake. I was avoiding that because it takes some >> time on my computer. >> >> I think the point is in the cell centroid because I've picked coordinates >> that are half the cell spacing in the three directions. >> Isn't that correct? >> > Ah, sorry! I misread how the mesh was defined. Glad to hear it's now working. Cheers, Andrew > >> Em sáb., 13 de jun. de 2020 às 22:46, Andrew McRae < >> andrew.mcrae(a)physics.ox.ac.uk> escreveu: >> >>> Strange, because your script works for me! (both with and without the >>> tolerance arguments, for what it's worth) >>> >>> [image: image.png] >>> >>> I guess you could test increasing the tolerance further, e.g. 1e-6 or >>> 1e-4 >>> >>> If that still doesn't work, all I can suggest are obvious things like >>> checking your firedrake is up to date, etc. >>> >>> [Note that the point you picked (0.05, 0.05, 0.05) is on the boundary >>> between two cells in the y-direction, so it's not completely surprising >>> that this has happened. In particular, if you are using a DG function >>> space, the value there isn't well-defined] >>> >>> Cheers, >>> Andrew >>> >>> >>> On Sun, 14 Jun 2020 at 00:01, Rodrigo Piccinini <rbpiccinini(a)gmail.com> >>> wrote: >>> >>>> Forgot to c/c the mailing list. >>>> >>>> ---------- Forwarded message --------- >>>> De: Rodrigo Piccinini <rbpiccinini(a)gmail.com> >>>> Date: sáb., 13 de jun. de 2020 às 19:40 >>>> Subject: Re: [firedrake] Point evaluation / PointNotInDomainError >>>> To: Andrew McRae <andrew.mcrae(a)physics.ox.ac.uk> >>>> >>>> >>>> Hi, Andrew. >>>> >>>> Thanks for the answer. >>>> I've attached a simplified example of what I'm doing. >>>> >>>> This reproduces the error I still get. >>>> >>>> [image: image.png] >>>> >>>> >>>> >>>> Em sáb., 13 de jun. de 2020 às 17:14, Andrew McRae < >>>> andrew.mcrae(a)physics.ox.ac.uk> escreveu: >>>> >>>>> Hi Rodrigo, >>>>> >>>>> Are you using the "f.at(...)" syntax, per >>>>> https://www.firedrakeproject.org/point-evaluation.html ? >>>>> >>>>> If so, you can pass in an extra argument *tolerance*, such as >>>>> tolerance=1e-8 >>>>> >>>>> Under the hood, there are some calculations to see if the point is in >>>>> a particular cell. If the point requested is on the boundary of the >>>>> domain, or on the boundary between cells, due to numerical precision >>>>> issues, it may happen that all nearby cells think the point is not in that >>>>> cell. There is a default tolerance (of, I believe, 1e-14) to allow for >>>>> this, but sometimes this is not enough. >>>>> >>>>> If increasing this to 1e-8 doesn't work, there's probably something >>>>> else going wrong! >>>>> >>>>> Cheers, >>>>> Andrew >>>>> >>>>> On Sat, 13 Jun 2020 at 20:22, Rodrigo Piccinini <rbpiccinini(a)gmail.com> >>>>> wrote: >>>>> >>>>>> Hello, >>>>>> >>>>>> I'm getting the PointNotInDomainError when trying to compute a point >>>>>> evaluation in a extruded mesh. >>>>>> I've tried different coordinates that are inside a cell, but I keep >>>>>> having the same problem. >>>>>> >>>>>> I'm using a DG scheme of 0 degree, so that values are constant inside >>>>>> a cell. >>>>>> Is there a workaround to get a cell value? >>>>>> >>>>>> Currently, I'm getting it from the vtk exported file, but it takes >>>>>> much more time. >>>>>> >>>>>> Thanks in advance for your help! >>>>>> -- >>>>>> Rodrigo >>>>>> >>>>> >>>> >>>> -- >>>> >>>> >>>> -- >>>> Rodrigo >>>> >>>> >>>> -- >>>> >>>> >>>> -- >>>> Rodrigo >>>> >>> >> >> -- >> >> >> -- >> Rodrigo >> >

1 0

Re: [firedrake] Fwd: Point evaluation / PointNotInDomainError
by Andrew McRae 14 Jun '20

14 Jun '20

Strange, because your script works for me! (both with and without the tolerance arguments, for what it's worth) [image: image.png] I guess you could test increasing the tolerance further, e.g. 1e-6 or 1e-4 If that still doesn't work, all I can suggest are obvious things like checking your firedrake is up to date, etc. [Note that the point you picked (0.05, 0.05, 0.05) is on the boundary between two cells in the y-direction, so it's not completely surprising that this has happened. In particular, if you are using a DG function space, the value there isn't well-defined] Cheers, Andrew On Sun, 14 Jun 2020 at 00:01, Rodrigo Piccinini <rbpiccinini(a)gmail.com> wrote: > Forgot to c/c the mailing list. > > ---------- Forwarded message --------- > De: Rodrigo Piccinini <rbpiccinini(a)gmail.com> > Date: sáb., 13 de jun. de 2020 às 19:40 > Subject: Re: [firedrake] Point evaluation / PointNotInDomainError > To: Andrew McRae <andrew.mcrae(a)physics.ox.ac.uk> > > > Hi, Andrew. > > Thanks for the answer. > I've attached a simplified example of what I'm doing. > > This reproduces the error I still get. > > [image: image.png] > > > > Em sáb., 13 de jun. de 2020 às 17:14, Andrew McRae < > andrew.mcrae(a)physics.ox.ac.uk> escreveu: > >> Hi Rodrigo, >> >> Are you using the "f.at(...)" syntax, per >> https://www.firedrakeproject.org/point-evaluation.html ? >> >> If so, you can pass in an extra argument *tolerance*, such as >> tolerance=1e-8 >> >> Under the hood, there are some calculations to see if the point is in a >> particular cell. If the point requested is on the boundary of the domain, >> or on the boundary between cells, due to numerical precision issues, it may >> happen that all nearby cells think the point is not in that cell. There is >> a default tolerance (of, I believe, 1e-14) to allow for this, but sometimes >> this is not enough. >> >> If increasing this to 1e-8 doesn't work, there's probably something else >> going wrong! >> >> Cheers, >> Andrew >> >> On Sat, 13 Jun 2020 at 20:22, Rodrigo Piccinini <rbpiccinini(a)gmail.com> >> wrote: >> >>> Hello, >>> >>> I'm getting the PointNotInDomainError when trying to compute a point >>> evaluation in a extruded mesh. >>> I've tried different coordinates that are inside a cell, but I keep >>> having the same problem. >>> >>> I'm using a DG scheme of 0 degree, so that values are constant inside a >>> cell. >>> Is there a workaround to get a cell value? >>> >>> Currently, I'm getting it from the vtk exported file, but it takes much >>> more time. >>> >>> Thanks in advance for your help! >>> -- >>> Rodrigo >>> >> > > -- > > > -- > Rodrigo > > > -- > > > -- > Rodrigo >

2 2

Fwd: Point evaluation / PointNotInDomainError
by Rodrigo Piccinini 13 Jun '20

13 Jun '20

Forgot to c/c the mailing list. ---------- Forwarded message --------- De: Rodrigo Piccinini <rbpiccinini(a)gmail.com> Date: sáb., 13 de jun. de 2020 às 19:40 Subject: Re: [firedrake] Point evaluation / PointNotInDomainError To: Andrew McRae <andrew.mcrae(a)physics.ox.ac.uk> Hi, Andrew. Thanks for the answer. I've attached a simplified example of what I'm doing. This reproduces the error I still get. [image: image.png] Em sáb., 13 de jun. de 2020 às 17:14, Andrew McRae < andrew.mcrae(a)physics.ox.ac.uk> escreveu: > Hi Rodrigo, > > Are you using the "f.at(...)" syntax, per > https://www.firedrakeproject.org/point-evaluation.html ? > > If so, you can pass in an extra argument *tolerance*, such as > tolerance=1e-8 > > Under the hood, there are some calculations to see if the point is in a > particular cell. If the point requested is on the boundary of the domain, > or on the boundary between cells, due to numerical precision issues, it may > happen that all nearby cells think the point is not in that cell. There is > a default tolerance (of, I believe, 1e-14) to allow for this, but sometimes > this is not enough. > > If increasing this to 1e-8 doesn't work, there's probably something else > going wrong! > > Cheers, > Andrew > > On Sat, 13 Jun 2020 at 20:22, Rodrigo Piccinini <rbpiccinini(a)gmail.com> > wrote: > >> Hello, >> >> I'm getting the PointNotInDomainError when trying to compute a point >> evaluation in a extruded mesh. >> I've tried different coordinates that are inside a cell, but I keep >> having the same problem. >> >> I'm using a DG scheme of 0 degree, so that values are constant inside a >> cell. >> Is there a workaround to get a cell value? >> >> Currently, I'm getting it from the vtk exported file, but it takes much >> more time. >> >> Thanks in advance for your help! >> -- >> Rodrigo >> > -- -- Rodrigo -- -- Rodrigo

1 0

Point evaluation / PointNotInDomainError
by Rodrigo Piccinini 13 Jun '20

13 Jun '20

Hello, I'm getting the PointNotInDomainError when trying to compute a point evaluation in a extruded mesh. I've tried different coordinates that are inside a cell, but I keep having the same problem. I'm using a DG scheme of 0 degree, so that values are constant inside a cell. Is there a workaround to get a cell value? Currently, I'm getting it from the vtk exported file, but it takes much more time. Thanks in advance for your help! -- Rodrigo

1 0

An issue in activate firedrake.
by Alqhtani, Kholoud K.S. 06 Jun '20

06 Jun '20

Dear Sir/ Lady, I am emailing you to inquire about the way to activate firedrake in my laptop. I did what was in the command message but it gave me the following message below. I would appreciate if you assist me to resolve this problem. Thanks in advance. I am looking forward to hearing from you. Kind regards, Kholoud Traceback (most recent call last): File “/Users/k/Documents/Firedrake/firedrake/src/h5py/h5py/__init__.py”, line 25, in <module> from . import _errors ImportError: dlopen(/Users/k/Documents/Firedrake/firedrake/src/h5py/h5py/_errors.cpython-37m-darwin.so, 2): Library not loaded: @rpath/libhdf5.103.dylib Referenced from: /Users/k/Documents/Firedrake/firedrake/src/h5py/h5py/_errors.cpython-37m-darwin.so Reason: image not foundDuring handling of the above exception, another exception occurred:Traceback (most recent call last): File “<string>“, line 1, in <module> File “/Users/k/Documents/Firedrake/firedrake/src/firedrake/firedrake/__init__.py”, line 59, in <module> from firedrake.checkpointing import * File “/Users/k/Documents/Firedrake/firedrake/src/firedrake/firedrake/checkpointing.py”, line 3, in <module> from firedrake.cython import hdf5interface as h5i File “firedrake/cython/hdf5interface.pyx”, line 2, in init firedrake.cython.hdf5interface File “/Users/k/Documents/Firedrake/firedrake/src/h5py/h5py/__init__.py”, line 29, in <module> raise ImportError(“You cannot import h5py from inside the install directory.\nChange to another directory first.“) ImportError: You cannot import h5py from inside the install directory. Change to another directory first.

1 0

DPG Optimal Test Function Implementation
by Austin Kaul 04 Jun '20

04 Jun '20

Firedrake Community, I am attempting to implement a form of DPG that uses DG optimal test functions that can be solved element by element. Has anyone in the community done this without solving for the optimal test functions over the global problem? i.e. they have been able to use Slate to perform the element-wise computations to compute the optimal test functions that will be used in the global solution as DPG FEM would require. If anyone has ideas how this could be implemented let me know. Regards, -- Austin Kaul

2 1

Bi-material example
by Karin&NiKo 01 Jun '20

01 Jun '20

Dear Firedrakers, I would like to model a bi-material medium. One can think of an elasticity problem where the domain is composed of 2 parts, one being made of steel and the other one of rubber. Then I would like to assign different Lamé coefficients to these parts. I do not find such an example in Firedrake tests. Could you please give me some piece of advice? Best regards, Nicolas

3 5

Adaptive mesh example
by Moulay Hicham TBER 01 Jun '20

01 Jun '20

Hi Everyone, Is there any Firedrake example with adaptive mesh refinement (residual , goal-oriented). Regards

2 1