Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Plot one variable without reevaluation in different plots
Posted 2024/11/27 5:47 JST Results & Visualization 5 Replies
Please login with a confirmed email address before reporting spam
I calculate average values using several non local couplings. I use these values in different plots in different summations. Each of these plots takes quite a while to create because all of the averages are reevaluated, even if it is the same result data set and they don't change.
Can I somehow evaluate the averages once and then reuse them in the different plots to save computation time? I have tried to use derived values, but can not access the resulting table to do the necessary summations...
Please login with a confirmed email address before reporting spam
Hi Finn,
yes, post processing with nonlocal couplings can be pretty time consuming. The ordinary variables are not part of the solution and thus need to be calculated again and again whenever they are called. A method that worked for me in the past is to write the result from the nonlocal coupling into a global state variable. State variables ARE part of the solution and no recalculation is needed when they get called in the post processing. The state variables do not carry a unit, so you must take care of correct units by yourself.
Cheers Edgar
-------------------Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Thank you I will try to do that!
Please login with a confirmed email address before reporting spam
Hi Finn,
yes, post processing with nonlocal couplings can be pretty time consuming. The ordinary variables are not part of the solution and thus need to be calculated again and again whenever they are called. A method that worked for me in the past is to write the result from the nonlocal coupling into a global state variable. State variables ARE part of the solution and no recalculation is needed when they get called in the post processing. The state variables do not carry a unit, so you must take care of correct units by yourself.
Cheers Edgar
I tried to implement it. but it increased the solution time quiet a lot unless I select one point as geometric entity. Unfortunately I can not evaluate the state variables. When i try to evaluate a variable under derived variables, I says the variable is undefined.
Please login with a confirmed email address before reporting spam
Hi Finn,
I only used global state variables and didn't find a significant increase in computation time even in large models and extensive integrations or averages. If you need to tie the state to a geometrical entity (e.g. a point) you will need to do a respective evaluation e.g. a point evaluation. A global state will be available in a global evaluation.
Cheers Edgar
-------------------Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Please login with a confirmed email address before reporting spam
Hello Edgar,
thanks for answering. I was a bit confused. If using state variables on component level they are bound to a geometric entity like point, boundary, domain. If instead defined globally that is not necessary and the impact on computation time is little.
I was a bit confused how to set up state variables correctly now I use:
variable_name for State
0 for initial value
if(1>0, expression, variable_name) for update expression
I have them updated at end of step.
Now the variable access is lightning fast!
Thank you!