In the past month I recall reading an article by Prof. W. Kahan about errors of numeric integration that begins with a scenerio of a colleague bringing him a function to integrate with a calculator. It seems straight forward, he wonders if it is a trick, and of course it turns out to illustrate the limitations of all calculator algorithms when applied to misbehaving functions (misunderstood is probably a better word; they can be handled just fine after conditioning).
At least, I think that is an article I remember. Does anyone have the link?
I've poured over the MoHPC DVD as well as Prof. Kahan's website, and cannot find it.
Thanks!