John D'Errico

No help in here. No explanation of how to use it, or even any clear explanation of what it does. As it is, this is just a set of random lines of code, most of which are not that well written.

For example, looking inside, I see the lines:

ind=[1,0,0,0;
0,1,0,0;
0,0,1,0;
0,0,0,1;];

Perhaps the author should learn how to use the tools already in MATLAB, here for example, eye.m.

Without looking deeply into the code, I don't know if the rest of this loopy code can be improved as trivially. I'll bet it can, but I'm not going to invest that time.

I'd suggest the author learn how to comment their code. Learn how to write adequate help.

Alejandro - while I would like to implement that capability one day in the future, that day may be a long time away before I find the time. Until then, all I can offer is to use a finer grid, which will mimic a higher order interpolant.

A long time ago, I wrote a tool that would take an existing spline, and rather than simply evaluate the first or last segments of the spline for extrapolation, I added a new segment that had all desired properties, like monotonicity, concavity, endpoint slope or value constraints, etc. Of course the new segments were fully consistent with the old end points of the spline and the shape at that point. If necessary to meet the specified constraints, I added several segments. The interface for this code was similar to that for SLM, with many possible property/value pairs for any possible shape.

I'll claim that this tool fully met with the SLM philosophy, in that it encouraged the user to explicitly specify information about the shape of the extrapolated curve. While I'd like to provide such a tool, more important in my opinion is to write a GUI wrapper for SLM.

To a large extent you can do that form of extrapolation already, when you first build the curve. Simply specify knots that go out as far as you need the curve to go. The knots need not always be tight up to the end points of the data. (That is the default for SLM, but you can choose your own knots.) This allows you to directly apply any pertinent shape information about that extrapolated region.

Very happy with results. Great Work!