MAD 6206 Combinatorics I, Fall 2011 CRN: 87401

Assignment 6 – suggested solution

Question 1. (Two well-known series in hypergeometric notation) (a) Consider the sum k n+k
P
k
k z . We
know how to find a closed form

from other methods. In particular, the sum follows from the generalized
binomial theorem since n+k −n−1 k = (1 − z)−n−1 . Put the sum into hypergeometric notation,


k = k (−1)
and show that  
a
1 F0 ; z = (1 − z)−a .
−
P 1 k
(b) Do the same for the sum k≥0 k! z .
Solution. (a) Note first that the first term is t0 = n0 = 1. Also,

tk+1 n+k+1
= z.
tk k+1

Hence, the sum is

 
n + 1,
1 F0 ;z .
−

As the sum equals (1 − z)−n−1 , we therefore have

 
a
1 F0 ; z = (1 − z)−a .
−

1 k
= ez , we have
P
(b) For k≥0 k! z
tk+1 1
= .
tk k+1
Thus, in the hypergeometric notation, the sum is
 
−
0 F0 ; z = ez .
−

Question 2. (Reduction to hypergeometric notation) Consider the sum k 2n+2

P
x+k

2k+1 2n+1 .
(a) Reduce the sum into hypergeometric notation.
(b) Assume that n ≥ 0 is an integer. Look up a hypergeometric identity for the sum, and show that

X 2n + 2 x + k   
2(x − n) 2x
= .
2k + 1 2n + 1 2n + 1 2n
k

Solution. Observe that   

2n + 2 x
t0 = .
1 2n + 1
Assume x is such that t0 6= 0. Now

tk+1 (2n + 1 − 2k)(2n − 2k)(x + k + 1)

= .
tk (2k + 3)(2k + 2)(x + k − 2n)

Hence the sum becomes

−n − 12 , −n,
 
x+1
t0 ×3 F2 3 ;1 .
2, x − 2n
We try the hypergeometric identity: If d + e = a + b − c + 1 and c is a positive integer, then

(d − a)c (d − b)c
 
a, b, −c
3 F2 ;1 = .
d, e (d)c (d − a − b)c

Here, we take, and assuming n ≥ 1,

1 3
a = −n − , b = x + 1, c = n, d= , e = x − 2n.
2 2
Then the conditions on the the parameters are satisfied:

3
d+e= + x − 2n = a + b − c + 1.
2
Now
d − a = 2 + n, d − b = −x + 21 , d = 23 , d − a − b = 1 + n − x, c = n,
and our sum becomes
(2 + n)n (−x + 21 )n
t0 .
( 32 )n (1 + n − x)n
Note that

(2n + 1)!
( 23 )n = ,
22n n!
(2n + 1)!
(2 + n)n = (n + 2 + n − 1)n = ,
(n + 1)!
1 (2x)2n
(−x + 12 )n = (−1)n (x − 21 )n = (−1)n ,
22n (x)n
(1 + n − x)n = (−1)n (x − n − 1)n ,
x


where the right hand sides are falling factorials. Since t0 = (2n + 2) 2n+1 , our sum becomes
  2n  
x 2 (2x)2n 2(x − n) 2x
(2n + 2) = .
2n + 1 n + 1 22n (x)n (x − n − 1)n 2n + 1 2n
x


We check that the above expression is good also for n = 0. Hence, assuming that 2n+1 6= 0 and integer
n ≥ 0, we have
X 2n + 2 x + k  2(x − n) 2x
= .
2k + 1 2n + 1 2n + 1 2n
k
x


As usual, the assumption that 2n+1 6= 0 is not needed. (The identity holds for all values of x because
both sides are polynomials in x with degree at most 2n + 1 and we have shown that the identity works for
infinitely many values of x.)

Pm n4n
Question 3. (Gosper’s algorithm) Follow Gosper’s algorithm to find a “closed” form for n=0 .
(2n
n)
Check your answer with Maple.
Solution. The ratio r(n) is

n+1 (n + 1)2 (n + 1)2

r(n) = 4 = .
n (2n + 2)(2n + 1) n(n + 21 )
Obviously,
a(n) = n + 1, b(n) = n + 12 , c(n) = n.
We have deg(x) = deg(b) = 1 and lc(a) − lc(b) = 1. Thus, the degree of x can be

deg(x) = deg(c) − deg(a) + 1 = 1 − 1 + 1 = 1.

Thus, take
x(n) = An + B.
Then the equation a(n)x(n + 1) − b(n − 1)x(n) = c(n) becomes

(n + 1)(An + A + B) − (n − 12 )(An + B) = n,

giving that
A = 25 , 4
B = − 15 .
Thus,
b(n − 1)x(n) 4n
zn = tn = 15(2n
(2n − 1)(3n − 2).
c(n) n)

Hence the sum equals

4m+1 2
zm+1 − z0 = 15(2m+2
(2m + 1)(3m + 1) − .
m+1 ) 15

Note: The next two questions are about interpreting the answers from Gosper’s algorithm.

Question 4. (Gosper’s algorithm) Use Gosper’s algorithm to find the sum:

h
X (m)k
, where h ≥ ℓ and ℓ ∈ Z+ ,
(n)k
k=−ℓ

P5 (60)k
and evaluate k=−5 (100)k .
Solution. As in class, we find that
k − n − 1 (m)k
zk = · .
n + 1 − m (n)k
Thus,
h
X (m)k
= zh+1 − z−ℓ .
(n)k
k=ℓ

For z−ℓ where ℓ ∈ Z+ , we have

−ℓ − 1 − n (m)−ℓ −ℓ − 1 − n (n + ℓ)ℓ 1 (n + ℓ + 1)ℓ+1

z−ℓ = · = · =− · .
n + 1 − m (n)−ℓ n + 1 − m (m + ℓ)ℓ n+1−m (m + ℓ)ℓ

and
h−n (m)h+1 1 (m)h+1
zh+1 = · =− · .
n + 1 − m (n)h+1 n+1−m (n)h
Thus,
h  
X (m)k 1 (n + ℓ + 1)ℓ+1 (m)h+1
= − .
(n)k n+1−m (m + ℓ)ℓ (n)h
k=−ℓ
For h = 5, ℓ = −5, n = 100, and m = 60, we have
5  
X (60)k 1 (106)6 (60)6 116045953
= z6 − z−5 = − = ≈ 30.12641412.
(100)k 41 (65)5 (100)5 3851967
k=−5

Question 5. (Gosper’s algorithm) Consider

X (n − 1)!2
.
n
(n − z)!(n + z)!

Pm (n−1)!2
(a) Assume z is not an integer. Find a closed form for the sum n=1 (n−z)!(n+z)! .
(n−1)!2
(b) Let z = k ∈ Z+ . Find m
P
n=k (n−k)!(n+k)! .
Solution. (a) Let
n2 a(n) c(n + 1)
r(n) = = .
(n + 1 − z)(n + 1 + z) b(n) c(n)
Obviously
a(n) = n2 , b(n) = (n + 1 − z)(n + 1 + z), c(n) = 1.
We have deg(a) = deg(b) = 2 and lc(a) = lc(b) = 1, giving that

deg(x) = deg(c) − deg(a) + 1 = 0,

or (since b(n − 1) = n2 − z 2 )
0−0
deg(x) = = 0.
1
Thus, deg(x) = 0 is the only possibility. Suppose that x(n) = K, and we have

n2 K − (n2 − z 2 )K = 1, =⇒ K = 1/z 2 .

Hence
n−1


b(n − 1)x(n) n2 − z 2 (n − 1)!2 (n − 1)!2 z
zn = tn = = = n−1+z , n ≥ 1.
c(n) z 2 (n − z)!(n + z)! z 2 (n − 1 − z)!(n − 1 + z)!


z2 z

The sum is
m
X (n − 1)!2
= zm+1 − z1 .
(n − z)!(n + z)!
n=1

For z1 , we note that from the definition,

 
r t!
= lim lim ,
z t→r w→z (t − w)!w!

we have  
0 0! sin(πz) sin(πz)
= = , =⇒ z1 = ,
z (−z)!z! πz πz 3
and so
m m


X (n − 1)!2 z sin(πz)
= 2 m+z − , z 6∈ Z.
πz 3


(n − z)!(n + z)! z z
n=1
 π
We can also use (−z)!(z − 1)! = sin(πz) to get

m
X (n − 1)!2 m!2 1
= −
(n − z)!(n + z)! z 2 (m − z)!(m + z)! z 2 (−z)!z!
n=1
m
!
1 Y i2
= −1
z 2 (−z)!z! i2 − z 2
i=1
m
!
sin(πz) Y i2
= − 1 , z 6∈ Z.
πz 3 i2 − z 2
i=1

(b) When z = k ∈ Z+ , each term in the sum is defined, and zm+1 is defined. For zk , we have

k−1


zk = k
= 0, k ∈ Z+ .
k 2 2k−1


k

Thus,
m m


X (n − 1)!2 k
= zm+1 − zk = m+k

.
(n − k)!(n + k)! k2 k
n=k