> As for the function +/@:(m&p.)@:i , the solution can be found 
> by using the method described in sections 11B and 11C of Math 
> for the Layman, only if m&p. is a first degree polynomial. 
> I am looking at another solution for higher order polynomials. 

The solution for +/@:(m&p.)@:i. is actually quite straightforward:
Since   
   (+/ c * f i. n) = c * +/ f i. n
   (+/ (f + g) i. n) = (+/ f i. n) + (+/ g i. n)
it is necessary only to solve   +/(i.n)^k .  But that is 
a polynomial of degree 1+k.  Use linear regression to find the 
coefficients of the polynomial.

csi=: 3 : 0
 (+/\_1|.!.0 t^y) %. ^/~t=. i.2x+y
)

   csi 1
0 _1r2 1r2
   csi 2
0 1r6 _1r2 1r3
   csi 3
0 0 1r4 _1r2 1r4

   ((i.10x)^/i.2+k) +/ .* csi k=: 1
0 0 1 3 6 10 15 21 28 36
   +/@(^&k)@i."0 i.10
0 0 1 3 6 10 15 21 28 36

   ((i.10x)^/i.2+k) +/ .* csi k=: 2
0 0 1 5 14 30 55 91 140 204
   +/@(^&k)@i."0 i.10
0 0 1 5 14 30 55 91 140 204

   ((i.10x)^/i.2+k) +/ .* csi k=: 3
0 0 1 9 36 100 225 441 784 1296
   +/@(^&k)@i."0 i.10
0 0 1 9 36 100 225 441 784 1296

   ((i.10x)^/i.2+k) +/ .* csi k=: 4
0 0 1 17 98 354 979 2275 4676 8772
   +/@(^&k)@i."0 i.10
0 0 1 17 98 354 979 2275 4676 8772



----- Original Message ----- 
From: "Paul" <[EMAIL PROTECTED]>
To: "Programming forum" <[email protected]>
Sent: Saturday, May 06, 2006 1:34 PM
Subject: Re: [Jprogramming] K. E. Iverson's Math for the Layman, section 
11E,exercise 3

Roger,

Sorry for the confusion. I did not mean to imply that J should provide the 
solution to the problems I mentioned. I was so excited to discover that J 
could express +/@i.n as a 2nd degree polynomial with the t. operator, that I 
imagined it would be able to do the same for the other examples and exercises 
of section 11C and 11E of Math for the Layman (by the way, the exercise that 
started all this is #3 of section 11C, not 11E as indicated in the the 
Subject of the messages). I just wanted to know if there is a way to define 
f, g and h below, in such a way that t. would be able to process them.

After trying the polynomial expressions of f, g and ), and based on your 
answer below, I conclude that  +/@:f@:i. is the only member of an infinite 
class recognized by t. ... which is fine with me. The examples you give are 
probably impossible to express in polynomial form, anyway. As for the 
function +/@:(m&p.)@:i , the solution can be found by using the method 
described in sections 11B and 11C of Math for the Layman, only if m&p. is a 
first degree polynomial. I am looking at another solution for higher order 
polynomials. 

For a little more than a month now, that I have downloaded and started 
studying J, it has inspired me a renewed interest in Math and computer 
programming. It is mainly due to the possibility J gives to experiment with 
advanced math concepts, as shown by K.E. Iverson's article. The article 
itself is a great review of these concepts that helps me understand them 
better now than when I studied them 30 year ago. I have to thank the 
designers and maintainers of this unique and fascinating package for the 
making it available on the net.



----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to