Well, this entry is a bit delayed. April was again an interesting month due to the coronavirus pandemic. It has certainly been interesting attending classes online – I’m not sure that I like it, nor that I’ll ever get used to it. I hope we will be able to get back to some level of normalcy before too long.
A prime focus in April was my optimization class project. I was joined on the project by two other graduate students – one of whom (Chris) is also a Coreform employee. The primary objective of our project was to wrap Sandia’s DAKOTA optimization code around Coreform’s smooth-spline FEM suite of tools. Due to the nascent status of Coreform software stack, and because we were DAKTOA neophytes, we chose to address a problem that was simple to solve with FEM and that lent itself to analytic verification (i.e. we knew what the answer should be). The title of our project was:
Determination of optimal support locations to maximize fundamental frequency of printed circuit boards
The course required the report to be formatted in a similar fashion as one would expect in a journal article and that the reports to be uploaded to an Open Access repository that would provide a DOI for our submission. We chose to publish our paper to Zenodo and you can find our report here:
As I mentioned, we used DAKOTA to wrap Coreform’s Crunch solver via Dakota’s “Black-box interface” functionality. We had to write a fairly “sophisticated” set of Python scripts to automate everything, but in the end everything worked fairly smoothly.
That’s mostly it for this post – the report goes into quite a bit more detail regarding our efforts. In June I’ll be virtually attending the online NAFEMS CAASE conference and am particularly interested to attend this interesting presentation:
Sounds interesting, no?