@acer @tomleslie

What version are you using?

• It was Maple 2016 (usually I use Maple 2018 but I had have a long simulation running on with it and, in these case, I prefer to open a new worksheet in an older version).
  I guess that If I had done the test with Maple 2018 I would not sent this question

Does it change if you set interface(typesetting=extended): ?

• Yes, it works perfectly well in Maple 2016

Thanks to both of you, sorry for the inconvenience.

@Josolumoh 

In my previous reply, please change

@Josolumoh 

A spreviously quoted by Carl the case d=0 is obvious.
More precisely it corresponds to a negative binomial distribution with parameters r and b/(2+b)
(in Maple  NegativeBinomial(r, b/(2+b))  )

Carl also pointed the meaningless of the case d<0

I describe here the way to sample your  "pseudo distribution" B (which does not sum to 1 and then need to me normalized).
The last plot superimposes the probability function of B for r=3, b=2, d=1 (red line) and the histogram in discrete form.
I considered that only integers values of x have a sense.

For other valuers of r, b, d you will have to verify that the value of the variable Xmax is large enough (in this case the variable ShouldBe1 must have reached a "converged" value; to verify this increase the value of Xmax and look to what happened to ShouldBe1).

Let me know ih this suits you

 

Download B_Distribution.mw

@tomleslie

Sorry, you're right, I forgot the "1/3" in the coefficient of x^3.
 

@tomleslie 

Not applicable: see tomleslie's answer

You got a wrong result with Optimization:-NLPSolve (?)

Given fexact is monotone decreasing over [0, 2] and its values are between [0, 1], given fapp is monotone increasing on [0, 2] with fapp(2) > 6789*2, it is obvious that the the maximum M of abs(fapp-fexact) is reached for x=2.

Then M = subs(x=2, fapp-fexact) =  - sin(2)/2 which is about  30543.54535 and not 2827.54535.

Using instead Optimization:-Maximize( abs(fapp-fexact), x=0..2) returns the good value.

More of this: your two calculations (L2 norm and L1 norm) give strangely the same result: could it be that Optimization:-Maximize(  abs(fapp-fexact), x=0..2, maximize) does not do what you were expecting for?

@tomleslie 

I apreciate these non-gurus explanations.
Thanks Tom

@acer 

Thanks acer for all these technical explanations.

For my understanding: I guess than you writting :-`+`(a, b) is a shortcut for Tolerances:-`+`(a, b) isn't it?
If so, is it not an unsafe way to proceed ifyou use two different packages containing different functions with the same name (I recently met this situation with packages simplex and geom which both contain functions named convexhull)?

i

@Carl Love

I had noticed the `+` returned by the command line  < with(Tolerances); > but I forgot that  `+` does not operate on a list but on a sequence; thus I forgot to use op or [] before running  `+`.
I like your (`+`@op) solution, easier to understand than using subsindets(Z, :-`+`, `+`);
Thanks

It seems to me that you wanted to answer the question EB1000 asked some hours ago?
Unfortunately, your answer turned out to be a question!

@tomleslie

Thanks Tom,
 I do not understand either why "evalr" seems necessary. In any case, thank you for keeping me from going in circles.

@acer 

Thank you acer.

As often there are several ways to tackle a problem and looking for the "optimal" solution seems to be an endless story.

By the way, I used myself the "subs({0="H", 1="T"},...)" method to return a solution in terms of "H" and "T.
Unfortunately writting
UseHardwareFloats := true:
V := floor~(SignalProcessing:-GenerateUniform(numThrows, 0, 2))
does not return integers making the {0="H", 1="T"} inoperative.
I tried the substitution {0.="H", 1.="T"}  but it didn't work neither.

I understand that floor, round, ceil do not return integers when the argument is a harware floating point number.
Is there a simple way to obtain, for instance, the integer part of harware floating point number?

@Carl Love 

Thanks Carl for your very documented answer.
I think I will  try to implement your suggestions, Not that the problem of calculating the probabilities of each sum deserves it, but rather out of personal satisfaction (in "practical" situation one would probably use Central Limit Theorems and the normal approximation when the number of dice exceeds a fex dozens).

By the way, I used in Q3 the package SignalProcessing and, from my student years I remember that convolution is generally be done, for effiency, by using FFT (2 directe + one inverse).
I do not know if SignalProcessing:-Convolution proceeds this way, but if it is the case, do you think its limitations you pointed out also impact FFT?

@radaar 

Have you thought about initializing the seed of the pseudo random generator?

This is a classical trap: executing twice the same worksheet generates exactly the same pseudo random numbers.To avoid this use:

Seed := randomize()

in a separate block justr after your top "restart" line

@Carl Love  Thanks for your remark.

It's rater difficult to compare the results produced by Q3 with the true ones. I did that for 3 and 4 dice and that was all good.
Obviously I was overconfident my assuming it would be same for higher values.

I suppose your observation definitely discredits Q3?

Maybe I'm wrong in my interpretation?
I understant A is a random variable with integer outcomes 1..6 and that each outcome has a different weight B

Would that be okay with you?
restart:
with(Statistics):
A := [$1..6];                   # A = [1, 2, 3, 4, 5, 6]
B := A *~ 10;                 # B = [10, 20, 30, 40, 50, 60]
P := B /~ add(B);           # P = [1/21, 2/21, ....6/21]
X := RandomVariable(ProbabilityTable(P));
N := 100;
Histogram(Sample(X, N), discrete);

# also:
plot(CDF(X, t), t=min(A)..max(A), discont=true)

Remark: if A was another set of outcomes, for instance A:=[1, 2, 3, 5, 8, 13], the plot of the histogram would be wrong (abscissas would be 1, 2, ..6). In this case you should correct the histogram plot by writting
Histogram(Sample(X, N), discrete, tickmarks=[[seq(i=A[i], i=1..6)], default])

Let me know if my answer is off the mark.