0% found this document useful (0 votes)
64 views4 pages

On Alternating Subsets of Integers: S.M.Tanny

Paper.

Uploaded by

Charlie Pinedo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
64 views4 pages

On Alternating Subsets of Integers: S.M.Tanny

Paper.

Uploaded by

Charlie Pinedo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

ON ALTERNATING SUBSETS OF INTEGERS

S.M.TANNY
Economic Council of Canada
(Present Address: Dept of Mathematics, University of Toronto, Canada, IV15S 1A1)

A finite set /of natural numbers is to be alternating [1] provided that there is an odd member of / between any two
even members and an even member of / between any two odd members; equivalent^, arranging the elements of/ in
increasing order yields a sequence in which consecutive elements have opposite parity. In this note we compute the
number anj of alternating subsets of | 1, 2, -., n [ with exactly r elements, 0 < r < n.
As a matter of notation we denote an alternating /--subset of \ 1, 2, - , n j by (qu q%/ , qr;n), where
we assume qt <q2 < < qr.
Let Enf (resp. On,A De t n e number of alternating subsets of { 1 , 2, , n } with r elements and with least ele-
ment even (resp. odd). It follows that
(1) an/f - Entr+On,r {1 <r <n).
For reasons which will soon become evident we set En/g - O^Q = 1; hence, antQ = 2 for n > 0. In addition,
set 3 0 , 0 = 1 .
Lemma. For any positive integer m,
Em+hr = m,r; 0 < r < m+ 1.
Proof. The case r = 0 is trivial. If r ~ m + 1, then
Em+1,m+1 - 0 - Om,m+1-
For 1 <r<m consider the correspondence
(Qu Q2, "/ Qrtm + 1) ** fai -1,Q2- I ~, Qr- Vm).
If ql is even then it easily follows that the number of r-subsets of \ 1, 2, , m + 1 J with least element even
equals the number of r-subsets of < 1, 2, , m \ with least element odd, q.e.d.
Proposition 1. For any positive integer m, and 1 < r < m + 1,
a
(2) m+1fr = am,r-i+m-1,r
Proof. The case/7? = 1 is obvious, so assumem > 2. If r = 1 then
a
m+i,i = m + 1 while am,0 = 2, am^i = m- 1 ;
hence (2) holds. For r > 1 we divide the r-subsets of | 1, 2, -,/?? + 1 \ (denoted as usual by (qu q2, -r-,^;
m + 1)) into two groups:
(i) qx = 1. Then (q2/ ., qr;m + 1) is an (r-/j-subset of j 1, 2, - , m + 1J which has an even least element,
so there are Em+^r./ such subsets.
(ii) ^ > 2. Then the correspondence given in the previous lemma shows that the number of such r-subsets is
a
m,r
We thus conclude that
(3) a
m+1,r ~ Em+1,r-1 +a
m,r
whence it follows that
(4) a
m+1,r =
Em+1,r-1 +
Em,r-1 +am-1,r-
325
328 ON ALTERNATING SUBSETS OF INTEGERS [DEC.

Applying the Lemma, Eq. (4) becomes


s
(5) am+1,r Om^t + Em^ + am^j.
Substituting (1) in (5) yields (2), q.e.d.
We remark that (2) holds for m = 0 if we define an/t = 0 if n < 0 or r < 0.
The recurrence (2) can be solved using the standard technique of generating functions [2,3]. We first define

(6 An(x) = E *n.rxr
k=0
Notice that An(x) is a polynomial of degree n sincea n , r ~ 0 f o r r > / ? . Using (2) we deduce that for n > 3,
(7) An(x) = xAn-jtxI + A^M,
while (6) and the boundary conditions on an^ give
A
oM = a 0 ^0 = 1
Ajx) = alf0 +axlx = 2+x
AJx) = a2f0 +a2Ax + a2f2x2 = 2 + 2x+x2
Set
oo

A(y,x) - ] T AnMvn

Then the above initial values together with (7) yield

(8) A(y*)= T-lLt)-2 .


1 - xy-y2
We now derive an explicit representation ofAn(x). To begin, expand 1/(1 -xy-y2) in a formal power series:

Fix any integer n > 0. Then the coefficient of yn in (9) is easily seen to be

(10) B(x) = ( nQ ) x" + ( " ~ 1 ) xn-2 + ...+ ( nfn/2]/21) x 2[ /2}


"' "

It follows that An(x), the coefficient of yn in A(ytx), is given by


tn/2] [fn-V/2]
11
< > U i - E (V)*""**' I] (it-/-*),*-/-*
/n/27-/
* E (-f-')*-**
s0
= Bn (x) + 2Bn i (x) + Bn2U) .
We now determinea n j , which, we recall is the coefficient of xr lnAn(x). We have two cases.
CASE 1. Assume r ==/? (mod 2). Then we can find s > 0 so that n- rs2s, i.e., s - #//? - r). Motice that Bn-](x)
does not contain the term x r . If 5= 0, then r-/7 and

<n " ( 2 ) s
a /;
1975] ON ALTERNATING SUBSETS OF INTEGERS 327

otherwise we can rewrite r as r = (n - 2)- 2(s- 1) and thus both Bn (x) and Bn-2M contain a term in xr\ hence
(io) . _ / n - Vs(n - r)\ ' / (n - 2) - [%(n - r) - 1] \

Simplifying (12) we have that for r = n (mod 2),

H3) Bntr ~ \Kln-r)) \ X(n - r) - 1 ) '


r
CASE 2. Assume r ^n (mod 2). Then the term x appears only in Bn-i(x), so we obtain (in a fashion analogous to
the one above) that
a
n,r ~ ' \ Kln-t -1) / "
That is, for r^ n (mod 2),
+
(14) a =2 (Mn r-1)\

We summarize these results in the following:


Proposition 2. Let an,r be the number of alternating /--subsets of j 1,2, -, n \ .
(i) l f r = /?(mod2),
_ f%(n + r)\. (%(n + r)-1\
an
>r " \%(n-r) / \ Kin -.r) - i t '
(ii) Ifr^/? (mod 2),
s WXfn + r - \

Asa result of this development we obtain an interesting relation between the numbersan/tand th&Fibonacci num-
bers [3]:
Corollary. Let fn be the Fibonacci sequence, i.e., /0 - ^ - 7 and ^+7 = />, * /J7-7- T n e n w e nave

n
s
(15) W X ^>"

Proof. Recall (see [3], p. 89) that the ordinary generating function for the sequence fn is
00

(16) Ffy) = fK" = JL-; .


n=0
It follows from (8) that

A(y, 1) = (1 +y)*F(y) = (fn +2f^i + fn-2)y" .


n=*0

where Lt = L2 = 0. But from (7),

n=0

and

n
An(J) = ,,-,
/=0
328 ON ALTERNATING SUBSETS OF INTEGERS DEC. 1975

whence we conclude that

(17) Yl 3n,r = fn+ ^n-1 + fn-2


r0
Using the recurrence
fn+1 = fn +fn-1,
the right-hand side of (17) simplifies to fn+2, which is the desired result, q.e.d.

REFERENCES
1. F. Hering, "A Problem on Inequalities/' Amer. Math. Monthly, Vol. 78, No. 3, March (1971), pp. 275-276.
2. C. L Liu, Introduction to Combinatorial Mathematics, McGraw-Hill, New York, 1968.
3. J. Riordan, Combinatorial Identities, Wiley, New York, 1968.
[This paper was received June 18,1973; revised August 23, 1973.]
AkAAAkA
[Continued from P. 324.]

TABLE 1
Jacobs Symbols: b = 1

(a/b) (b/a) (a/-b) (-b/a)


-7 1 1 - 1 1
-5 1 1 -1 -1
-3 1 1 -1 1
- 1 1 1 -1 -1
1 1 1 1 1
3 1 1 1 - 1
5 1 1 1 1
7 1 1 1 - 1

TABLE 2
Jacobi Symbols: b = 3
a (a/b) (b/a) (a/-b) (-b/a)
-7 -1 -1 1 -1
-5 1 -1 -1 1
-3 0 0 0 0
-1 -1 1 1 -1

1 1 1 1 1
3 0 0 0 0
5 -1 -1 -1 -1
7 1 -1 1 1

[Continued on P. 330.]

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy