Mikhail Drugov

123 Reputation

5 Badges

15 years, 101 days

MaplePrimes Activity


These are replies submitted by Mikhail Drugov

@tomleslie  You are right, of course! I did not notice the third dot. I am very sorry for posting such a trivial question.

@Joe Riel You are right, of course! I did not notice the third dot. I am very sorry for posting such a trivial question.

@mmcdara 

Thank you!

And one question: In both files you use

dist := Distribution(PDF = (t -> PDF(U, t)), CDF = (t -> CDF(U, t))):
Y    := RandomVariable(dist):


But wouldn't using either PDF or CDF be enough?

@mmcdara 

Thank you!

I think I came across Gamma Distribution and probably some others where the choice of parametrization in Maple was not the most obvious.

@mmcdara 

Yes, a bug. Returning zero instead of unevaluated, error, whatever, is very dangerous.

@mmcdara 

This is extremely helpful, thank you very much!

So, this is not a bug but a very non-intuitive behaviour. I would imagine that either Maple considers Y to be a random variable and then all functions should work, or the opposite, and none should work. But this partial work (PDF, CDF, Variance, etc work) is quite confusing.

It also means that defining a random variable as a function of other random variables is not very straighforward. And also that  Maple help is sometimes confusing. Say, here https://de.maplesoft.com/support/help/maple/view.aspx?path=Statistics%2fRandomVariables in (15) any reasonable person would think that X/Y is Cauchy(0,1) random variable, not some strange object of which function PDF gives the same answer (and again, HazardRate(X/Y,t) always returns zero).

On your last point: "To return to your answer to vv, even if Statistics is a good package (I agree with vv), it still suffers from several flaws (I have already pointed out some of them in the past)."

could you please provide links? I am a heavy user of the Statistics package, and this will be very useful.

Thank you!

 

 

@vv 

Sure, in this particular case I noticed it does not work while writing explicitly f/(1-F) does. But now I wonder which other functions of the Statistics package might have this problem? Or suppose somebody uses it without realizing (maybe, in an expression, maybe, a student),

If this is a bug, it should be reported (I don't know how to do it).

@Carl Love 

This is very good, thank you! I also need other combinations (like 1/4*X1+3/4*X2) and they all work when in this simplest form fraction.

 

@tomleslie 

Thank you! This helps! Though if I do for Beta(5,5) then again there is negative pdf in the right tail but then setting Digits:=20: helps. And for Beta(10,10) I need about 40 digits.

@Daniel Skoog 

Thanks! 

I just need to get the 2018 version then...

 

 

@rlopez 

Thank you!

I guess I could also create an image from the text which then can be rotated. It is a pity there is no simple solution...

Any progress on this?

Thank you all for the tips! They are very helpful.

I would be nice, of course, if the Quantile function just worked - in the next version?

1 2 Page 1 of 2