@Axel Vogt  Many thnaks for your interest and help. I have the Brent's book which I think you metioned.

@Carl Love First, many thnaks for your suggestion, I have the book R. Brent, Algorithms for minimization without derivatives, Prentince Hall Inc. 1973

However, I don't undestand at all such code. In this URL https://people.math.sc.edu/Burkardt/f_src/brent/brent.f90 is teh code in fortra99, which seems to can be easily adapted to maple (almost a coy/paste, i think).

@C_R  many thanks for your useful help. With slight modifications, the code seems to work ok.

brent.mw

I reiterate my gratitude for your help.

@acer thanks!

@vv Thanks!! Nice, really nice code!

@mmcdara  Thanks for your code!

Hi!!

It seems that I have solved the problem: In the file IFF_v2.mw, I have changed  global NunFun:=n-1 by global NunFun:=n  in the procedure "CreateDataSet".

@acer many thanks! Now it works.

Great @Rouben Rostamian  !!! It seems that your code works fine..Thanks!! 

@vv thanks.

As you say, "t it is easy to implement a Lebesgue filling curve in any dimension (just a few lines in Maple". In fact, you are a post with this issue:

https://www.mapleprimes.com/questions/220980-Convert-Peano-Curve-Into-A-Function

@vv thanks for your reply. Your code is for Maple 2020, right? In the paper linked in my post, the code (in C) is for the n-dimensional Hilbert curve, and for its  "pseudo-inverse"

@Carl Love The version is  Maple 17.00

@acer You are right, I am not using the 17v, sorry by the confussion.

@tomleslie Tanks for your reply. In Maple 17 your code returns me the error

Error, `Fractals` does not evaluate to a module
 

However, the attached txt contains procedures to (numerically) compute H(t), H being the (approximation of the ) Hilbert curve and t a point in [0,1], as well as the pseudo inverse of H.

Regards.

@Carl Love Mmm....I am not sure at all, but I would have to review the article in more detail but it is likely that I am wrong.

In the paper entitled "One-Dimensional Global Optimization for Observations with Noise" by Calvin and Zilinskas, says that for the global optimization fo a single vairable function with noise (normal 0,1)  "...a Wiener process is accepted as a statistical model of the objective function."