At this moment I am testing the R-implementation of the relevant codes and I am finding that R is very slow (at least on my windows pc).

So I have to ask you about the size of your data:

What is the number n?

If n is large enough, then only FORTRAN can run fast.

From what I know, R is slower than Fortran or C (which I don't know).

Moreover, R under windows cannot exploit power of multi-core stations because R processus (.exe) runs only on one core. There are some packages for mac and linux that enable multi-core. I saw also some beta version for windows but I never tested. And since, maybe there are some stable versions

look at this link for fresh news:

http://cran.r-project.org/web/views/HighPerformanceComputing.html

My dataset has 80 units. 8 replicates at 10 sites, which are distributed along a gradient.

This is not large I imagine.

What do you think?

Thanks

Your problem is tiny and will not cause any need to link to Fortran or other external code from R.

Dear Pierre, you can now find the R-implementation of methods ESE, EDE as supplementary material to the publication, here in researchgate.

Any problem that has a sigmoid pattern can be solved easily and fast with those methods.

We can find the inflection point without any prior assumption for the underlying sigmoid function.

Please give me your feedback or send me sigmoid data from your research.

Thank you Demetris,

I will have a look to your codes soon, and I 'll let you know by emails.

Thanks again,

Pierre

Pierre, I have good news for you.

The new package inflection is now on CRAN and you can install and use it for analyzing your data.

Remember just one think: if the inflection point seems to be near the right edge of the data x-range [a,b], then probably your data is simply convex or the inflection point has not occured yet.

I wish you a good and efficient research.

http://cran.r-project.org/web/packages/inflection/index.html