Repoducing baseline design examples about pycycle HOT 1 CLOSED

I ran your commit and the baseline run file in the repo and got the same converged value for fan diameter. There have been a few minor tweaks to pyCycle and OpenMDAO itself since the publication of the paper and that seems to have affected the converged values a tiny bit.

at the level of accuracy of this analysis 100.6 vs 100.1 isn't a meaningful change overall. Some of the numerical details in pyCycle mean that we're taking the inverse of some VERY VERY poorly conditioned matrices and that causes a small amount of numerical noise. That, combined with tiny changes to algorithms and small refactors of the code will add up to tiny drifts like that.

Your run is matching what Im seeing, and our current code base is passing all of our regression tests. So you're good.

If you really want it to match the paper exactly, you should roll back to the 1.0 version of pycycle. You'll have to roll back to a version of OpenMDAO that works with that as well. 2.10 should do it.

from pycycle.