phd-scripts/Biomaterials13/Sample_ParamSweep/readme.txt

36 lines
1.6 KiB
Text

Using the optimization scripts. Contact james.grogan@universityofgalway.ie for further details.
--------------------------------------------------------------
These scripts were developed to perform optmizations with Abaqus and DAKOTA on the ICHEC
system. With small modifications they can also be used to perform optimizations on
windows systems.
Prerequisites:
Abaqus (tested in v6.10)
DAKOTA (tested in v5.2) - needs to be built from source on ICHEC.
Sequence:
1. Launch.pbs
Sends optimization job to ICHEC queue. Loads modules neccessary for DAKOTA and runs the optimization job as
a taskfarm.
2. tasks.inp
Input file for the taskfarm program. Changes to a unique directory for each task an runs DAKOTA with the input file
OptB.in.
3. OptB.in
DAKOTA input file. Tells DAKOTA what sort of optimization to perform. How many parameters to use and what ranges the
parameters fall in. Launchs the preprocessing wrapper script 'PyWrapperB.py' and designates 'Bparams.in' and 'Bparams.out' as the optimization input and output files.
4. PyWrapperB.py
Python wrapper script. Launchs the Abaqus geometry kernel file 'OptB.py'. Launches each abaqus job. Launches the
postprocessor python file 'OptPostB1.py'.
5. OptB.py
Abaqus kernel script. Creates the FE model. Model parameters are read in from DAKOTA through the Bparams.in file.
6. OptPostB1.py
Post-processes the initial simulation. If a design looks promising it launches a corrosion simulation and a second postprocessor OptPostB2.py.
7. OptPostB2.py
Post-processes the corrosion simulation. Returns the objective function value to DAKOTA through the Bparams.out text file.