Kratos: Ideas Regarding the Stream During Calculation

[AlejandroCornejo]

I consider (my personal opinion) that the stream provided by kratos when computing can be enhanced so it can give more information and less repeated or even known info.
Currently the stream is somethign like this:

 [STEP:: 17  TIME: 1.7000000000000004 ]
RESIDUAL CRITERION:  :: [ Obtained ratio = 0.00333897; Expected ratio = 0.0001; Absolute norm = 3.90516e-06; Expected norm =  1e-09]
RESIDUAL CRITERION:  :: [ Obtained ratio = 0.0017649; Expected ratio = 0.0001; Absolute norm = 2.06418e-06; Expected norm =  1e-09]
RESIDUAL CRITERION:  :: [ Obtained ratio = 0.00133829; Expected ratio = 0.0001; Absolute norm = 1.56522e-06; Expected norm =  1e-09]
RESIDUAL CRITERION:  :: [ Obtained ratio = 0.00106871; Expected ratio = 0.0001; Absolute norm = 1.24994e-06; Expected norm =  1e-09]
RESIDUAL CRITERION:  :: [ Obtained ratio = 0.000874373; Expected ratio = 0.0001; Absolute norm = 1.02264e-06; Expected norm =  1e-09]

which repeats at EACH iteration the expected ratio or the max norm allowed (which has been imposed by the user so he/she already knows).

I would like to develop a new interface that could look like this (please give your personal opinion):

 [STEP:: 17  TIME: 1.7000000000000004 ]
|   ITERATION   |     RATIO      |   ABS NORM    |  RATIO TOL      |  ABS.NORM TOL  |  CONVERGENCE |
|   1           |  0.00333897    |  3.90516e-06  |  0.0001         |  1E-9          |     FALSE    |
|   2           |  0.0017649     |  2.06418e-06  |  0.0001         |  1E-9          |     FALSE    |

What do you think?

Regards

@KratosMultiphysics/design-committee @KratosMultiphysics/implementation-committee
I know that @loumalouomega has done smthg similar in his App, maybe he's aware of some problems that could be interesting

[loumalouomega]

I have pending to implement it properly into the LoggerTable. The thing is than in order to properly do it we need a "LoggerManager", stored on ProcessInfo, in order to share the information between conv.criteria (if you use And or Or criteria, you need to append information)

[AlejandroCornejo]

I was thinking innitially to improve the one regarding the residual_criteria or the displ criteria. But we can generalize it as you say.

[loumalouomega]

I already did that once, but we decided to move to the logger side

[RiccardoRossi]

i love the idea, however how can this be imlemented? recall that we can
jave a lot of different convergence criterias, comnined in arbitrary ways

On Thu, Jun 13, 2019, 2:38 PM Vicente Mataix Ferrándiz <
[email protected]> wrote:

I already did that once, but we decided to move to the logger side

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
https://github.com/KratosMultiphysics/Kratos/issues/5080?email_source=notifications&email_token=AB5PWEJNWJCKWVLGFJRQOP3P2I5VHA5CNFSM4HXZDAAKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXTRU2A#issuecomment-501684840,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AB5PWEM6XQZ4LCPRGP5PACTP2I5VHANCNFSM4HXZDAAA
.

[loumalouomega]

i love the idea, however how can this be imlemented? recall that we can jave a lot of different convergence criterias, comnined in arbitrary ways

I have experience with this, that's why I propose a "LoggerManager" class, to access the different types of loggers

[ddiezrod]

just mentioning here @KratosMultiphysics/altair as we were trying something similar in our side, so we can reach a common solution together

[loumalouomega]

just mentioning here @KratosMultiphysics/altair as we were trying something similar in our side, so we can reach a common solution together

Check ContactStructuralMechanicsApplication ;)

[loumalouomega]

Concerning https://github.com/KratosMultiphysics/Kratos/pull/744

[AlejandroCornejo]

I'll propose a PR with an intermediate solution in order to discuss it with the @KratosMultiphysics/technical-committee

[loumalouomega]

I'll propose a PR with an intermediate solution in order to discuss it with the @KratosMultiphysics/technical-committee

Feel free to use #744 as reference

[AlejandroCornejo]

ping @RiccardoRossi Please, this issue has become stale, could you give us a final aswer?

[AlejandroCornejo]

I'll create a PR regarding this as mentioned before

[pooyan-dadvand]

Please consider that you can use table logger to create a separate output as a table for a given set of labels. Nevertheless, it cannot be used in combination with other outputs in the default ones.

[loumalouomega]

Please consider that you can use table logger to create a separate output as a table for a given set of labels. Nevertheless, it cannot be used in combination with other outputs in the default ones.

That's why I suggested to create a logger manager so it can be use to handle the logging between different parts of the code

[pooyan-dadvand]

Can be an option but then how we pass the loggers to corresponging part of the code?

[loumalouomega]

You commented a good approach in #1926.

El mar., 12 nov. 2019 12:55, Pooyan Dadvand [e-mail]
escribió:

Can be an option but then how we pass the loggers to corresponging part of
the code?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/KratosMultiphysics/Kratos/issues/5080?email_source=notifications&email_token=AEYQZAB5RLX7GXZXSDBDEMTQTKKSPA5CNFSM4HXZDAAKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOED2AIEQ#issuecomment-552862738,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AEYQZAGVHEQGGBIHBYA4GW3QTKKSPANCNFSM4HXZDAAA
.