@Thomas Richard 

Thank you Richard.

To answer your last phrase I'm working on a coupling between Maple and R to delegate the latter some specific calculations which would be too complex to recode in Maple. The R code generated in Maple is saved in a text file "MyFile.R"   that must me made executable before launching the command ssystem("..../bin/Rscript MyFile.R").

 

@Carl Love 

You're right, I'd forgotten ths Welch's test and its "equivalent" number of degrees of freedom.
Your argument is undisputable.
Consider my question as null and void

@Carl Love 

I will rectify this as soon as I get home.

@Carl Love 

Right, once I sent my answer, I realized that I should have mentioned that the constructor || makes the variable global (something that caused me serious problems the first time I used this in a procedure...)
 

@Thomas Richard 

Thanks for your quick reply, here is a toy problem:

Toy_Problem.mw

@vv 

sorry for this late answer...

Thanks you vv

@dharr 

I do not have Maple 2017.
In Maple 2018.0, Windows 7 version, ID 1298750, using names returns an explicit error saying that strings are expected.

Maybe a slight change from one version to the next one...

@Carl Love @vv

You're 100% right.
I redid the exercice this morning, mind calm , and I clearly fooled myself yesterday.
I think it's not appropriate to chat on mapleprimes while watching a soccer's game

@tomleslie 

which is a real pity for loops are very common in probabilistic graphical models.
Of course you can always cheat by writting {1, x}, {x, 1} instead of {1, 1} and give ad hoc transitions probabilities to simulate a 1 to 1 transition, but it would not be smart at all.
Let me immediately moderate these comments: what is really intersting from a practical point of view is the transition matrix (which could be the adjencicy matrix in a zero-loop graph)... indeed, the graph here is more an educational suport than a computational tool.

@acer 

Now from the office...

Worksheet Tab3.mw, after a few personal modifications, works perfectly well.

Thank you very much acer

@Carl Love 

If you write sol := dsolve(...), isn't it the value of sol('numfun')  ?

Maybe you should take a look at the "LinearFit" help page.
This procedure can return a lot of different ouputs, among them there could be what you are looking for.
If not, could you be more specific about your problem?

@Preben Alsholm 

(mmcdara from professional account)

Thanks for your reply, it will prove very useful in the future.
The only thing that makes me reluctant to use assuming instead of assume is that you must repeat "assuming ..." everywhere it's necessary.


 

@tomleslie 
 

Sorry to reply from the office with the login sand15.

Mistake ot not mistake ?
I would even say  mistakes.

You wrote "I simply looked at the graph the OP wanted to fit and "eyeballed" a few obvious points through which the graph passed."
As a rule it's already a very bad idea, specially when you already have numerical values at hand (the one the OP used).
Next, it's not your job, nor mine, not the job of anyone else to replace, as you did, data by one's own data. If you were suspicious about the OP's for they did not match the graph, you would have point this fact to him before giving adetailed answer.
And finally, what makes you so sure that the data you used after "eyeballed a few obvious points" are good: as you have probably noticed, the graph the OP gave does not mention the scales of the axis.

To end this, susceptibility is poor counselor.
You wrote "You seem to believe that the OP's selected points are important..." : NO, I just used your own words you used in your first answer "The overall strategy is to identify about six "important" points".
Please avoid rewritting the past

@acer 

I would have payed more attention to your example.
Thanks for the details.