Split partitioned-heat-conduction case into complex and simple one #148

BenjaminRodenberg · 2021-01-20T13:21:25Z

closes #147

Specification for `partitioned-heat-conduction`:

Dirichlet is always left, Neumann is always right
Coupling Boundary is always straight line parallel to y axis
No time dependent heat flux and rhs
mpirun is still supported (for fenics)
no subcycling. Timestep of both FEniCS solvers is hard-coded to be equal to time window size from precice-config.xml.
Analytical solution is always possible
use scalar-valued flux (important for compatibility with nutils and again easier)

Specification for `partitioned-heat-conduction-complex`:

Dirichlet and Neumann solvers are arbitrary
Coupling Boundary is arbitrary
Time dependent heat flux and rhs may be set via gamma
mpirun is supported (for fenics)
subcycling possible
Analytical solution is possible for straight and circular boundary as long as no subcycling is used.
use vector-valued flux

Open Todos

@IshaanDesai @uekerman Is anything missing or too much in the specification above?
Need to adapt the structure to Restructure tutorials #125
Make sure that solver can be reused with other compatible cases, as well. (Nutils)
Make sure that nutils also recovers exact solution (postponed to Fix nutils-nutils heat transport case #152 & Add nutils solver for heat conduction problem #153)

BenjaminRodenberg · 2021-01-20T13:23:01Z

@uekerman I assume that flow-over-heated-plate is independent from this PR? The FEniCS solver used in flow-over-heated-plate is very similar, but has some differences and cannot easily be reused.

uekerman · 2021-01-20T14:51:21Z

@uekerman I assume that flow-over-heated-plate is independent from this PR? The FEniCS solver used in flow-over-heated-plate is very similar, but has some differences and cannot easily be reused.

yes, flow-over-heated-plate is completely independent of this one.

uekerman · 2021-01-20T15:05:21Z

Specification looks good so far. I guess a real review only makes sense once we have the compatibility with Nutils.
A comment already now: for that we will need a scalar flux and a nearest-neighbor mapping.

BenjaminRodenberg · 2021-01-27T18:25:51Z

Specification looks good so far. I guess a real review only makes sense once we have the compatibility with Nutils.
A comment already now: for that we will need a scalar flux and a nearest-neighbor mapping.

To make sure I understand: We want this case for FEniCS and for nutils. Meaning, according to the new structure we get:

partitioned-heat-conduction/
- solid-fenics/...
  - heat.py
- solid-nutils/...
  - heat.py
- precice-config.xml
- ...

This means, that we need a new tutorial (according to the old structure), precice/tutorials/tree/master/HT/partitioned-heat/nutils-nutils, first. Does this already exist somewhere? The closest one I can find is flow-over-plate/buoyantPimpleFoam-nutils.

I think the cleanest approach would be to first create another PR from develop, where I collect all the necessary parts from partitioned-heat/fenics-fenics and flow-over-plate/buoyantPimpleFoam-nutils to develop a new case partitioned-heat/nutils-nutils. Afterwards I can use the resulting case, add it in this PR and move everything according to the new structure.

uekerman · 2021-02-03T10:00:09Z

This means, that we need a new tutorial (according to the old structure), precice/tutorials/tree/master/HT/partitioned-heat/nutils-nutils, first. Does this already exist somewhere?

Yes. Sources are provided in the issue: #137

To make sure I understand: We want this case for FEniCS and for nutils. Meaning, according to the new structure we get:

The in #137 suggested structure is

- dirichlet-fenics
  - heat.py
- neumann-fenics
  - heat.py
- dirichlet-nutils
  - heat.py
- neumann-nutils
  - heat.py 
- precice-config.xml
- clean.sh
- README.md

Yes, the drawback of this scheme is that it leads to quite some code duplication. Maybe it is simple to understand though.

Alternatively:

- fenics
  - heat.py
- nutils
  - heat.py
- precice-config.xml

together with some command line argument to decide whether the left Dirichlet or the right Neumann sub domain should be computed.

BenjaminRodenberg · 2021-02-03T12:26:42Z

This means, that we need a new tutorial (according to the old structure), precice/tutorials/tree/master/HT/partitioned-heat/nutils-nutils, first. Does this already exist somewhere?

Yes. Sources are provided in the issue: #137

I added the sources to this PR. Currently just copy-paste without any testing or updates of the code.

To make sure I understand: We want this case for FEniCS and for nutils. Meaning, according to the new structure we get:

The in #137 suggested structure is
- dirichlet-fenics
  - heat.py
- neumann-fenics
  - heat.py
- dirichlet-nutils
  - heat.py
- neumann-nutils
  - heat.py 
- precice-config.xml
- clean.sh
- README.md
Yes, the drawback of this scheme is that it leads to quite some code duplication. Maybe it is simple to understand though.

That's probably a good reason not to use this structure here.

Alternatively:
- fenics
  - heat.py
- nutils
  - heat.py
- precice-config.xml 
together with some command line argument to decide whether the left Dirichlet or the right Neumann sub domain should be computed.

I like this idea. Added to this PR 👍

BenjaminRodenberg · 2021-02-03T14:56:50Z

I took care of updating the nutils case as far as I can get. I currently see one problem:

Nutils uses two meshes and FEniCS only one. This does not allow us to use the same configuration for both cases. @uekerman any idea what we can do here?

uekerman · 2021-02-03T16:15:54Z

Nutils uses two meshes and FEniCS only one. This does not allow us to use the same configuration for both cases. @uekerman any idea what we can do here?

We could read and write at the GPs only, that should be simple

BenjaminRodenberg · 2021-02-03T22:07:37Z

The last problem with this PR that I see is that the exact solution cannot be recovered when using nutils. Remaining features are fine.

BenjaminRodenberg · 2021-02-05T15:10:30Z

I decided to postpone fixing the nutils case to #152 and #153. This PR is from my point of view ready to ~~merge~~ review. We can continue working on #153 at a later point in time from my point of view and I don't want with merging the features that are already working until the nutils case is fixed, since I'm a bit stuck here currently and it is hard to estimate how long fixing this will take.

BenjaminRodenberg · 2021-02-05T15:14:07Z

Notes for review:

Are the specifications from above fulfilled?
Are there still any features or files we don't need for the respective cases?
Is the documentation clear (enough)?

IshaanDesai

Looks good 👍 The separation of simple and complex geometry cases makes the FEniCS code much more readable :)

partitioned-heat-conduction/README.md

BenjaminRodenberg · 2021-02-08T14:59:43Z

@uekerman another hint for the review: Currently we don't provide fenics/run.sh, since we need command line input to decide whether we run the Dirichlet or Neumann solver. I see two options here:

provide a fenics/run.sh that accepts command line input
provide a dirichlet-fenics/run.sh and neumann-fenics/run.sh. Put all actual code only into dirichlet-fenics. neumann-fenics only exists for following the general structure and providing a run.sh script.

Any opinion?

uekerman

Overall structure looks good. Minor conventions things and README stuff below.

partitioned-heat-conduction-complex/precice-config.xml

partitioned-heat-conduction/README.md

uekerman · 2021-02-19T09:41:20Z

partitioned-heat-conduction/README.md

-{% include important.html content="We have not yet ported the documentation of the preCICE tutorials from the preCICE wiki to here. Please go to the [preCICE wiki](https://github.com/precice/precice/wiki#2-getting-started---tutorials)" %}
+## Setup
+
+In this tutorial we solve a partitioned heat equation. For information on the non-partitioned case, please refer to [1, p.37ff]. The corresponding FEniCS code can be found on [github:fenics-tutorial](https://github.com/hplgit/fenics-tutorial/blob/master/pub/python/vol1/ft03_heat.py).


Needs probably a bit more information. We have not yet mentioned FEniCS before.

I provided more details on the case in 75e68af. I don't think that we have to explain what FEniCS is here. Or what do you mean?

Right now, "FEniCS" falls a bit from the sky. The setup should actually be independent of the solver.

I did some reordering in 9614c02. I think this makes it a bit better.

partitioned-heat-conduction/README.md

uekerman · 2021-02-19T09:46:50Z

Currently we don't provide fenics/run.sh, since we need command line input to decide whether we run the Dirichlet or Neumann solver. I see two options here:
1. provide a `fenics/run.sh` that accepts command line input

2. provide a `dirichlet-fenics/run.sh` and `neumann-fenics/run.sh`. Put all actual code only into `dirichlet-fenics`. `neumann-fenics` only exists for following the general structure and providing a `run.sh` script.
Any opinion?

I guess we are also fine without a run script here for the moment. If we add some later then option 1.

Co-authored-by: Benjamin Uekermann <[email protected]>

uekerman

Looks good. Did not run it again, but did not see any obvious problems.

BenjaminRodenberg added 2 commits January 20, 2021 14:05

Strip case of everything besides simple heat conduction setup.

c05a2e8

Adds precice-config.xml supporting parallel runs.

c41013e

BenjaminRodenberg requested review from uekerman and IshaanDesai January 20, 2021 13:21

BenjaminRodenberg self-assigned this Jan 20, 2021

BenjaminRodenberg mentioned this pull request Jan 20, 2021

Add complex geometry fenics-fenics tutorial #147

Closed

BenjaminRodenberg added 2 commits February 3, 2021 12:59

Merge branch 'restructure' into i_137

71fcc28

Add documentation and nutils case.

51eb727

BenjaminRodenberg changed the title ~~Simplify partitioned-heat-conduction case~~ Split partitioned-heat-conduction case into complex and simple one Feb 3, 2021

Add complex case. Based on ae77ae5.

91510da

Cleanup and Update to newest version of python bindings.

1ca266d

BenjaminRodenberg added 2 commits February 3, 2021 16:22

Use scalar-valued flux.

0551635

Apply new structure to complex case.

f410eec

This was linked to issues Feb 3, 2021

Flatten solver hierarchy for perpendicular-flap #135

Closed

Add complex geometry fenics-fenics tutorial #147

Closed

Remove subcycling.

6aacd46

uekerman and others added 5 commits February 3, 2021 18:04

Simplify heat nutils to one mesh, make config consistent

77bf103

Fix names.

8dd4258

BCs working.

d2e726c

Minor restructuring

0bcf866

Fix initial condition and output.

679fe8d

BenjaminRodenberg mentioned this pull request Feb 5, 2021

Fix nutils-nutils heat transport case #152

Closed

Mark nutils case as under construction. See #152.

f80f8fd

BenjaminRodenberg force-pushed the i_137 branch from 9c7a8fa to f80f8fd Compare February 5, 2021 14:59

BenjaminRodenberg marked this pull request as ready for review February 5, 2021 15:08

IshaanDesai approved these changes Feb 7, 2021

View reviewed changes

partitioned-heat-conduction/README.md Outdated Show resolved Hide resolved

Add link to Nutils

35dc013

uekerman requested changes Feb 19, 2021

View reviewed changes

BenjaminRodenberg and others added 11 commits February 19, 2021 11:33

Update partitioned-heat-conduction-complex/precice-config.xml

1d01384

Co-authored-by: Benjamin Uekermann <[email protected]>

Update partitioned-heat-conduction-complex/precice-config.xml

4acdad4

Co-authored-by: Benjamin Uekermann <[email protected]>

Update partitioned-heat-conduction-complex/precice-config.xml

4afe280

Co-authored-by: Benjamin Uekermann <[email protected]>

Update partitioned-heat-conduction-complex/precice-config.xml

97cf386

Co-authored-by: Benjamin Uekermann <[email protected]>

Add note for website.

e1d1411

Provide more details and use x_c = 1

75e68af

Cleanup for consistency with non-complex case.

2ad8b53

Fix formatting

9db75e5

Cleanup w.r.t visualization

69c0c46

Merge branch 'i_137' of github.com:precice/tutorials into i_137

cf11057

Add summary.

bf765be

BenjaminRodenberg removed a link to an issue Feb 19, 2021

Flatten solver hierarchy for perpendicular-flap #135

Closed

Remove FEniCS from setup.

9614c02

uekerman approved these changes Mar 5, 2021

View reviewed changes

uekerman merged commit caa5ad0 into restructure Mar 5, 2021

BenjaminRodenberg deleted the i_137 branch September 28, 2023 15:14

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Split partitioned-heat-conduction case into complex and simple one #148

Split partitioned-heat-conduction case into complex and simple one #148

BenjaminRodenberg commented Jan 20, 2021 •

edited

Loading

BenjaminRodenberg commented Jan 20, 2021

uekerman commented Jan 20, 2021

uekerman commented Jan 20, 2021

BenjaminRodenberg commented Jan 27, 2021

uekerman commented Feb 3, 2021

BenjaminRodenberg commented Feb 3, 2021

BenjaminRodenberg commented Feb 3, 2021

uekerman commented Feb 3, 2021

BenjaminRodenberg commented Feb 3, 2021

BenjaminRodenberg commented Feb 5, 2021 •

edited

Loading

BenjaminRodenberg commented Feb 5, 2021

IshaanDesai left a comment

BenjaminRodenberg commented Feb 8, 2021

uekerman left a comment

uekerman Feb 19, 2021

BenjaminRodenberg Feb 19, 2021

uekerman Feb 19, 2021

BenjaminRodenberg Feb 19, 2021

uekerman commented Feb 19, 2021

uekerman left a comment

Split partitioned-heat-conduction case into complex and simple one #148

Split partitioned-heat-conduction case into complex and simple one #148

Conversation

BenjaminRodenberg commented Jan 20, 2021 • edited Loading

Specification for partitioned-heat-conduction:

Specification for partitioned-heat-conduction-complex:

Open Todos

BenjaminRodenberg commented Jan 20, 2021

uekerman commented Jan 20, 2021

uekerman commented Jan 20, 2021

BenjaminRodenberg commented Jan 27, 2021

uekerman commented Feb 3, 2021

BenjaminRodenberg commented Feb 3, 2021

BenjaminRodenberg commented Feb 3, 2021

uekerman commented Feb 3, 2021

BenjaminRodenberg commented Feb 3, 2021

BenjaminRodenberg commented Feb 5, 2021 • edited Loading

BenjaminRodenberg commented Feb 5, 2021

IshaanDesai left a comment

Choose a reason for hiding this comment

BenjaminRodenberg commented Feb 8, 2021

uekerman left a comment

Choose a reason for hiding this comment

uekerman Feb 19, 2021

Choose a reason for hiding this comment

BenjaminRodenberg Feb 19, 2021

Choose a reason for hiding this comment

uekerman Feb 19, 2021

Choose a reason for hiding this comment

BenjaminRodenberg Feb 19, 2021

Choose a reason for hiding this comment

uekerman commented Feb 19, 2021

uekerman left a comment

Choose a reason for hiding this comment

BenjaminRodenberg commented Jan 20, 2021 •

edited

Loading

Specification for `partitioned-heat-conduction`:

Specification for `partitioned-heat-conduction-complex`:

BenjaminRodenberg commented Feb 5, 2021 •

edited

Loading