i have a 100 numbers and want to find the algorithm for them , so how
can i do that ?

---

Request for Question Clarification by siliconsamurai-ga on 31 Jan 2005 05:54 PST

I'm sorry, your question is meaningless as stated. Do you want to find
out how the number string is generated or how the numbers are related?

If so, are they all real numbers? All integers? Bear in mind that this
has to be a very simple question for that price.

---

Clarification of Question by tomg_6887-ga on 07 Feb 2005 01:57 PST

i will ask the quistion by other way :
i have 100 numbers like ( 45125120 )( 84560231 ) .......
and i want to know the relation between them to be able to generate a
new numbers by myself .
can anyone help ?

Thanks

i think the question is incomplete... what do u want to do with the 100 numbers?

This should do the trick, http://www.research.att.com/~njas/sequences/ :)

For the problem, as stated, I understand that you have an
unconstrained sequence of 100 numbers and that you require a method to
guess the pattern which generated that sequence. You are about to be
disappointed.
Given a finite sequence of numbers there exists a trivial algorithm to
reproduce that sequence? place the sequence of numbers into a vector
and iterate over that vector.  I suspect this is not the kind of
solution you envisioned.
I suspect you are thinking of numeric sequences of the kind you
frequently see in intelligence tests where there is a trivial
generation rule ? such as for the sequence 1,1,2,3,5,8,13? where each
number is the sum of the previous two.  If you constrain your problem
by only requiring solutions fitting a finite set of known types of
sequence then we can find algorithms to test for a fit for those
specific cases.  In the general case, where the sequence of numbers
you wish to handle are unconstrained, this is not computable ? i.e.
you will find no algorithm that given a finite sequence of numbers
will return the most simple function to generate the sequence.  The
?Online Encyclopedia of integer sequences?  (referred to by
lticket-ga) takes the opposite approach? it knows about many common
sequences and tried to identify the sequence you entered as ?similar?
to one it has seen before (I presume using offsets and scaling)   It
is important to note that this will only reverse-engineer a very small
proportion of the possible sequences.
If you are looking to predict trends (for example if the sequence of
numbers represented stock prices for example) then thinking of the
sequence suggests that a statistical model is what you really want?
statisticians spend most of their time looking at random data and
using trial and error to try and find simple models which are good
approximations to their sample data.  There are no short cuts.
If this is a real problem (as opposed to a mental exercise) then you
need to re-think the problem definition ? you need to bring to bear
more facts that can be assumed about the specific numeric sequences
you wish to analyse?