询问stata corp.的人得到的回复：

Dear Rex,
Unfortunately, the -gplm- command never made it much past the prototype phase that was the subject of that presentation I made about five years ago.  As such, there is no version of -gplm- suitable for general use.  I should have made the point that -gplm- was only a prototype in the presentation.
I wish I could be more helpful.  As an alternative to a partially linear model, try considering a flexible parametric model based on fractional polynomials; see -help fracpoly- within Stata 10.1 for more information.
Best,
Bobby