Demonstrator calibration #186
Conversation
- Stochastic optimization with stochastic constraints implementation - Tested it for "column simulation code". Graphs and other detailts in the shared document - ALso implemented a method to perform stochastic optimisation when design variables directly appear in the objective and it is not differntiable. (Variational Objective VO.py)
…ion part. Note that it is not modular completely. Just for ref.
…onSolver' into Calibration_Optimisation_HydrationSolver
# Conflicts: # environment.yml # lebedigital/simulation/precast_column.py # tests/demonstrator_scripts/test_column_simulation.py
…monstrator scripts updated
# Conflicts: # lebedigital/demonstrator_scripts/beam_design.py # lebedigital/demonstrator_scripts/kpi_from_fem.py # tests/demonstrator_scripts/test_beam_design.py # usecases/optimization_paper/analyze_kpis/analyze_kpis.py # usecases/optimization_paper/optimization_workflow/Inputs/aggregates_volume_fraction.json # usecases/optimization_paper/optimization_workflow/Inputs/geometry.json # usecases/optimization_paper/optimization_workflow/Inputs/loads.json # usecases/optimization_paper/optimization_workflow/Inputs/steel_yield.json
# Conflicts: # lebedigital/demonstrator_scripts/dummy_hydration_parameters.py # lebedigital/demonstrator_scripts/dummy_paste_strength_stiffness.py # usecases/optimization_paper/optimization_workflow/Inputs/material_properties.json
|
Some of the files (in particular the yaml files in the minimum working example) should not be added. |
There was a problem hiding this comment.
Thank you for the PR.
- A small comment for the future, please make sure, before creating a branch to open a new issue and link the branch to it, this helps to keep an overview of the repo.
- There are a lot of scripts where I lost the overview what is happening, but I guess that is okay. However, I think the structure does not follow our current structure. We keep scripts in lebedigital and then call them in usecases. Currently we are working on the optimization_paper, so I would not keep things in "usecases/demonstrator".
- as @joergfunger mentioned please do not include the results of the data extraction in the commit, as well as pngs that are generated.
- Please make sure your scripts are tested individually
- please add the calibration step to the workflow, or make sure in some way that this can be run and tested.
|
|
||
| # 1-$1, the $1 is for the first argument of the sbatch run_jobs 100, where 100 samples total | ||
| # Load Python Module | ||
| source /home/atul/.bashrc |
There was a problem hiding this comment.
is there any way to generalize this?
There was a problem hiding this comment.
I think it would be good to rename the file and function, as its not a placeholder anymore, right?
There was a problem hiding this comment.
I have to be honest, I dont remember what this function was used for. Do we still need it?
There was a problem hiding this comment.
This is a result of the metadata extraction, should not be committed (see Jörgs comment)
There was a problem hiding this comment.
it would be good if these can be created from the workflow, not committed in the repo
| # - https://github.com/artix41/AVO-pytorch/blob/master/avo-poisson.ipynb | ||
| # observartion : 9.11.2022 : This code is working as it should. Accurately recreating the Fig 2 of the paper. | ||
| import sys | ||
| sys.path.extend(['/home/atul/PhD_Tasks/LeBeDigital/ModelCalibration']) # temp fix to add the project path |
There was a problem hiding this comment.
not sure if this is a general solution
There was a problem hiding this comment.
similar question as before, I dont remember anymore what we used the "column" for. is this part of the calibration?
| @@ -362,13 +368,15 @@ rule approx_hydration_parameters: | |||
| from lebedigital.demonstrator_scripts.dummy_hydration_parameters import dummy_hydration_parameters | |||
| #merging contents of both dictionaries and individual variable inputs | |||
|
|
|||
| # had to bypass this pint thing as it was not letting me use numpy. | |||
There was a problem hiding this comment.
This sounds dangerous and is a great source for errors.
Can we move the "removing of pint units" to the script? There we can first make sure the right units are applied and then remove the units. Otherwise this will lead to problems. By using pint_object.magnitude, you can access the value of the object.
|
I merged this branch with main, as Cesary fixed a problem unrelated to the PR. |
Just adding the dummy scripts with distribution and inputs of the snakemake to read them. Not in their final form, but to help testing the updated design problem. I will update how the calibration generates the calibrated results and how the optimization reads them subsequently.