Waldek Hebisch <[EMAIL PROTECTED]> writes:

> Today I toyed a bit with guessing package and immediatly I
> hit a few quirks:
> 
> 1) I tried to guess polynomial from its values.  So I wrote:
> this looks like off by one error.

Currently, the definitive reference is my article on

http://uk.arxiv.org/abs/math.CO/0702086

or

http://axiom-wiki.newsynthesis.org/GuessingFormulasForSequences

I guess I should add a link into the documentation.  Done. Ideally, it would
turn into a HyperDoc examples page...
 
> 2) plain guess gives:
> 
> guess [myPol(i) for i in 1..15]
>    Compiling function myPol with type PositiveInteger -> Integer
> 
>    (33)
>    [
>                    n         4     3     2                  3     2
>      [function= [[x ]f(x): (x  - 4x  + 6x  - 4x + 1)f(x) + x  - 3x  - 3x - 1= 
> 0]       ,
>       order= 0]
>      ]
>   Type: List(Record(function: Expression(Integer),order: NonNegativeInteger))
>                          Time: 0.004 (IN) + 0.004 (EV) + 0.02 (OT) = 0.02 sec
> 
> Description of guess says that it tries guessRec and guessADE. 
> Description of guessADE says that that it finds equations satified
> by function with given first Taylor coefficients, so technically this
> result is probably OK.  But our polynomial is determined just by
> 4 coefficients, while in the equation we have 9 numbers, so the
> reslut is much more complicated than necessary (and it requires
> a little effort to realize that it really decribes a polynomial).

Yes.  The default values for guess are stupid.  Changed.

> 3) I tried guessExpRat to force the result to be polynomial, but guessExpRat
> seems to a lot of time (I was too impatient to actually see any answer).  I
> wonder what method is used in guessExpRat?

Also described in the article.  Yes, this method actually doesn't really work
for lists longer than about 10 items, unless you have a supercomputer, that is.

> guessRec seem to reasonably quickly find difference equation satisfied by a
> function.  Once we know f(n+1)/f(n) asymptotic at infinity should immediatly
> give us exponential factor.

The hard part is to get the asymptotic at infinity :-)  As far as I know,
functions of the form

(a+b*n)^n*q(n)

(a,b are fractions, q a rational function) does not necessarily satisfy a
difference equation.

Suggestions *very* welcome.

Martin


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"FriCAS - computer algebra system" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/fricas-devel?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to