Numerical models and crystal balls

I attend a lot of talks. I sometimes hear talks presenting insights and conclusions solely based on the results of numerical analysis (usually FEA) which include few details on the simulation’s basis, assumptions, limitations, or how it was validated. Important principle:

When all you say is, “We built a model and here is what we learned”, you might as well say “We got these numbers from our crystal ball!”

Crystal ball

Not the image you want to evoke during your technical talk.

There is a real skepticism toward numerical analysts because of talks and papers with insufficient and irreproducible detail. The problem is not that you neglected certain factors in your analysis—it’s that you didn’t acknowledge it. Let’s look at how to handle this right.

Even if you are limited with time, you should shorten the discussion of your very important results enough to say something about how you got them—scaled appropriately to the expertise of your audience. You can address this issue in three sentences (and perhaps an accompanying slide):

“We simulated the response using a [algorithm name] approach in [software name] with special attention given to [key components of the model]. Our simulation neglected [key things you forgot neglected] but was validated against [other accepted numerical methods or experimental data] and shown suitable within [whatever bounds]. I’m happy to provide further details about any aspect of our methodology to anyone interested.”

There. Because you spent thirty seconds saying that, you have now:

Other actual experts in the crowd won’t mind simplifying assumptions and limitations. They know that a big part of research is figuring out which factors matter and which don’t. And if they challenge something in the Q&A;, you probably stand to learn a lot from the discussion. If you appear evasive, however, you can expect the feistier academics in attendance to try to make a point out of it publicly—this is not what you want.

These types of omissions also make it uncomfortably obvious when a grad student’s advisor is essentially driving the bus on a project.

Pro tip: A little honest disclosure goes a long way.