Hi Daniel,

On 09/04/2018 05:06 PM, Daniel Wheeler wrote:

Hi,

I'd like to try and solve the problem outlined here,

https://pages.nist.gov/pfhub/benchmarks/benchmark4.ipynb/

Is this a problem that Sfepy could handle? I'm assuming the elasticity part of the problem would be straightforward, but what about a fourth order phase field equation? Given that it's tractable are there examples that match the form of the elasticity equation and/or any Cahn-Hilliard type examples for Sfepy?

I have no experience with that kind of equation. However, I have found an article mentioning that the fourth order equation can be solved by splitting into two second order equations [1]. That should be doable with sfepy, talking about assembling the problem and getting the discretized system (if H1 elements are OK). Again, I have no experience with solving such systems.

Cheers, r.

[1] http://lib.dr.iastate.edu/cgi/viewcontent.cgi?article=1237&context=me_pu... or https://www.sciencedirect.com/science/article/pii/S0021999111002361

On 09/05/2018 06:50 PM, Daniel Wheeler wrote:

Thanks for the references. I'll let you know how it goes. I might solve the elasticity part with Sfepy and the CH part with FiPy for now.

What is the best example in Sfepy to understand how to solve multiple PDEs?

Try looking at the examples in examples/multi_physics/ - it is usually elasticity + something.

r.

On Tue, Sep 4, 2018 at 5:33 PM Robert Cimrman cimrman3@ntc.zcu.cz wrote:

...

Hi Daniel,

On 09/04/2018 05:06 PM, Daniel Wheeler wrote:

...

Hi,

I'd like to try and solve the problem outlined here,

https://pages.nist.gov/pfhub/benchmarks/benchmark4.ipynb/

Is this a problem that Sfepy could handle? I'm assuming the elasticity part of the problem would be straightforward, but what about a fourth order phase field equation? Given that it's tractable are there examples that match the form of the elasticity equation and/or any Cahn-Hilliard type examples for Sfepy?

I have no experience with that kind of equation. However, I have found an article mentioning that the fourth order equation can be solved by splitting into two second order equations [1]. That should be doable with sfepy, talking about assembling the problem and getting the discretized system (if H1 elements are OK). Again, I have no experience with solving such systems.

Cheers, r.

[1] http://lib.dr.iastate.edu/cgi/viewcontent.cgi?article=1237&context=me_pu... or https://www.sciencedirect.com/science/article/pii/S0021999111002361 _______________________________________________ SfePy mailing list -- sfepy@python.org To unsubscribe send an email to sfepy-leave@python.org https://mail.python.org/mm3/mailman3/lists/sfepy.python.org/

Hi Daniel,

On 09/12/2018 08:28 PM, Daniel Wheeler wrote:

Hi Robert,

So, I've implemented the elasticity part of the problem using Sfepy, see,

https://pages.nist.gov/pfhub/benchmarks/benchmark4.ipynb/

for the problem definition. My implementation is contained in

https://github.com/wd15/sfepy-bm4/blob/73df8d6a929c9ee70019785aef770cef57938dae/main.py

and

https://github.com/wd15/sfepy-bm4/blob/73df8d6a929c9ee70019785aef770cef57938dae/sfepy_module.py

Basically, I'm using both the dw_lin_elastic and dw_lin_prestress to implement the two terms in the stress / strain equation. Does what I've done make sense to you with respect to how those terms are being used?

If I understand it correctly, dw_lin_elastic computes the elastic strain/stress parts for eta-dependent strain/stress in dw_lin_prestress? It looks OK to me (after a cursory glance).

I'm not currently solving the phase field equation for $\eta$. I'm simply imposing an $\eta$ field. Initially, I'm going to solve the phase field equation using FiPy and then use Sfepy for the elasticity. Eventually, I'd like to learn how to do it all in Sfepy.

The two-equations approach from that article seems to me to be a viable route.

Cheers, r.

PS: I needed to adjust myself a bit to the functional style of your code and I am starting liking it. What was your motivation for using toolz and what practical experience have you got with using it? Just curious :) (I am trying to use a more functional style in another Python-based project.)

Any advice would be greatly appreciated.

Cheers,

Daniel

On Thu, Sep 6, 2018 at 7:58 AM Robert Cimrman cimrman3@ntc.zcu.cz wrote:

...

On 09/05/2018 06:50 PM, Daniel Wheeler wrote:

...

Thanks for the references. I'll let you know how it goes. I might solve the elasticity part with Sfepy and the CH part with FiPy for now.

What is the best example in Sfepy to understand how to solve multiple PDEs?

Try looking at the examples in examples/multi_physics/ - it is usually elasticity + something.

r.

On 09/13/2018 04:10 PM, Daniel Wheeler wrote:

On Thu, Sep 13, 2018 at 8:29 AM Robert Cimrman cimrman3@ntc.zcu.cz wrote:

...

PS: I needed to adjust myself a bit to the functional style of your code and I am starting liking it. What was your motivation for using toolz and what practical experience have you got with using it? Just curious :) (I am trying to use a more functional style in another Python-based project.)

Well, I'm not sure why I got into functional programming, but probably just from a few friends and colleagues who are into it. I ended up learning Haskell for fun and went from there. Regarding Toolz, I've been using it for quite some time. It seems well supported and is the basis for Dask, which seems to be widely used. I think there are alternatives in Python for functional style programming, but haven't tried them. They don't seem as complete as Toolz. There's a list of them here, https://github.com/sfermigier/awesome-functional-python#libraries. I haven't tried the others.

Thanks! The list is also interesting, I was not aware of the number of functional programming packages.

If you'd like to chat about this then we could set up a video chat sometime.

Yeah, but I would need to look at toolz first to have some common ground :) I will ping you when I am ready (hopefully in a finite time).

r.

On Mon, Oct 1, 2018 at 7:27 PM Anil Kunwar romagurung33@gmail.com wrote:

Daniel, How far have you been in coupling linear elasticity with PFM? What are the major challenges in implementing terms for C-H equation in sfepy?

Hi Anil,

Currently, I'm solving the linear elasticity with Sfepy and the phase field with FiPy. I know that is a bit lame, but I'm hoping to solve it all with Sfepy at some point. A link to the code is here, https://github.com/wd15/sfepy-bm4.

The Sfepy part is https://github.com/wd15/sfepy-bm4/blob/master/sfepy_module.py and the FiPy part, https://github.com/wd15/sfepy-bm4/blob/master/fipy_module.py and the code to link them is https://github.com/wd15/sfepy-bm4/blob/master/main.py.

It's almost finished. I need to get the coupling working fully. It will be incredibly slow using this approach, but at least it will give me an answer.

I'll let you know when it's done.

Yes, I'd like to solve a number of the phase field benchmark problems (https://pages.nist.gov/pfhub/simulations/) using Sfepy if possible. Also, with some of the other FE solvers out there such as Moose and Prisims PF. Have you tried solving any of the benchmark problems?

On Tue, Oct 2, 2018 at 6:31 PM Anil Kunwar romagurung33@gmail.com wrote:

Dear Daniel, It is nice to know about your brilliant works. I have already used FiPy (in 2015) in my works and so it will be not difficult to know about the FiPy codes, and currently I am learning sfepy. I am following your works and simply I am inspired with your updated efforts. Please keep on updating. Yes, it will be a nice output if sfepy can have independent codes on phase field equations too (and this is your goal too). Keep it up.

Yours sincerely, Anil Kunwar

On Tue, Oct 2, 2018 at 5:01 PM Daniel Wheeler daniel.wheeler2@gmail.com wrote:

...

On Mon, Oct 1, 2018 at 7:27 PM Anil Kunwar romagurung33@gmail.com wrote:

...

Daniel, How far have you been in coupling linear elasticity with PFM? What are the major challenges in implementing terms for C-H equation in sfepy?

Hi Anil,

Currently, I'm solving the linear elasticity with Sfepy and the phase field with FiPy. I know that is a bit lame, but I'm hoping to solve it all with Sfepy at some point. A link to the code is here, https://github.com/wd15/sfepy-bm4.

The Sfepy part is https://github.com/wd15/sfepy-bm4/blob/master/sfepy_module.py and the FiPy part, https://github.com/wd15/sfepy-bm4/blob/master/fipy_module.py and the code to link them is https://github.com/wd15/sfepy-bm4/blob/master/main.py.

It's almost finished. I need to get the coupling working fully. It will be incredibly slow using this approach, but at least it will give me an answer.

I'll let you know when it's done.

-- Daniel Wheeler

On Wed, Oct 3, 2018 at 6:06 PM anil kunwar romagurung33@gmail.com wrote:

However, I will be solving them (benchmark problems) in nearby future, and most probably with MOOSE software, as I am currently working with the software. Equally, I am interested to compare the results of MOOSE with sfepy + fipy .I have not worked yet with PRISMS , but recently I have read an article on it by Aagesen et al , 2018 (https://link.springer.com/article/10.1007%2Fs11837-018-3079-6); and I am very much motivated to learn the concept of materials genome initiative and ICME via

That's great, please do upload your results when you've finished.