c Birkhäuser Verlag, Basel, 2005

Annals of Combinatorics 9 (2005) 363-377

0218-0006/05/040363-15 Annals of Combinatorics
DOI 10.1007/s00026-005-0264-4

Binomial Determinant Evaluations

Wenchang Chu and Leontina Veliana di Claudio
Dipartimento di Matematica, Università degli Studi di Lecce, Lecce-Arnesano P. O. Box 193,
73100 Lecce, Italia
[email protected], [email protected]

Received February 24, 2005

AMS Subject Classification: 15A15, 05E05

Abstract. By means of matrix decomposition and the partial fraction method, we establish sev-
eral determinant evaluation formulas, which can be considered as generalizations of the Vander-
monde and Cauchy determinants.

Keywords: Vandermonde determinant, Cauchy determinant, binomial coefficient, hypergeometric

series, symmetric function

1. Introduction and Notation

Letting {λi }ni=0 be a complex number sequence, the Vandermonde determinants of order
n + 1 are evaluated as follows:

∏
j
h i
∆(λ) = det λi = (λ j − λi ), (1.1)
0≤i, j≤n 0≤i< j≤n

∏
n− j
h i
∇(λ) = det λi = (λi − λ j ), (1.2)
0≤i, j≤n 0≤i< j≤n

which are related by the relation

n+1
∆(λ) = ∇(λ) × (−1)( 2 ). (1.3)

Another important determinant evaluation due to Cauchy reads as

1 ∆(λ)∆(µ)
 
det = n . (1.4)
λi + µ j 0≤i, j≤n ∏i, j=0 (λi + µ j )

By means of matrix decomposition and the partial fraction method, we will establish
several determinant evaluation formulas, which can be seen as generalizations of the

363
364 W.C. Chu and L.V. di Claudio

Vandermonde and Cauchy determinants. Several most determinant identities due to

Amdeberhan & Zeilberger [1] are recovered as consequences.
Throughout the paper, the square matrix Ci j 0≤i, j≤n will be abbreviated for sim-


plicity as Ci j when there is no risk of confusion.

 

2. Determinants in Factorial Fractions

For a complex number x, define the shifted factorial of order n by

(x)0 = 1, (x)n = x(x + 1) · · · (x + n − 1) for n = 1, 2, . . . . (2.1)

Then we have the following determinant identity.

Theorem 2.1 (Determinant in factorial fractions). Let a, c and {λi }ni=0 be complex
numbers. Then we have the following determinant identity:

(a) j+λi
 n (c − a) (a)
k λk
det = ∆(λ) ∏ . (2.2)
(c) j+λi 0≤i, j≤n k=0 (c) n+λk

Proof. It is trivial to see that

n (a)
(a + λi) j
 
λk
of (2.2) = det ×∏ .
(c + λi) j k=0 (c) λ k

Recalling the Chu-Vandermonde formula (cf. [2, §1.3])

j
(a + λi) j − j, c − a (c − a)k (− j)k
 
= 2 F1 1 =∑
(c + λi ) j c + λi k=0 (c + λi )k k!

and then applying the Binet theorem, we derive

(a + λi) j (− j)i (c − a) j
     
det = det × det .
(c + λi ) j i! (c + λi ) j

Noticing that (− j)i /i! is a lower triangular matrix with diagonal entries equal to ±1,
 

we have
(− j)i
 
n+1
det = (−1)( 2 ) .
i!
While the other determinant can be reduced, by extracting common column-factors, as
follows:
(c − a) j  n (c − a)k
 
det = det (c + j + λi)n− j ∏

.
(c + λi) j k=0 (c + λk )n

Writing
n
(c + j + λi)n− j = ∑ λin−k αk j
k= j
Binomial Determinant Evaluations 365

where [αi j ]0≤i, j≤n is a lower triangular matrix with diagonal entries equal to one, we
get the following factorization
 n− j 
det (c + j + λi)n− j = det λi × det αi j = ∇(λ).
   

Summing up, we have, in view of (1.3), established

n
(a + λi) j (c − a)k
 
det = ∆(λ) ∏ , (2.3)
(c + λi ) j k=0 (c + λ k )n

which is equivalent to (2.2). This completes the proof of the theorem.

Letting c → ∞ and a → ∞, we derive from (2.2) respectively the following determi-

nant identities:

Corollary 2.2 (Determinants in shifted factorials).

n
det (a) j+λi 0≤i, j≤n = ∆(λ) ∏ (a)λk
h i
(2.4)
k=0

1 ∇(λ)
 
det = . (2.5)
(c) j+λi 0≤i, j≤n ∏nk=0 (c)n+λk

Putting a := 1 + z, c := 1 and a := −y, c := 1 in (2.2), we get respectively the

following determinant evaluations.

Corollary 2.3 (Ostrowski [8]; cf. [4, Equations 13-14]).

n
z + λi + j (k + λk )! z + λk
   
det = ∇(λ) ∏ , (2.6)
λi + j 0≤i, j≤n k=0 (n + λk )! k + λk
n
y (k + λk )! y + k
     
det = ∇(λ) ∏ . (2.7)
λi + j 0≤i, j≤n k=0 (n + λk )! k + λk

The last determinant evaluation can also be derived directly as follows. Recalling
the Chu-Vandermonde convolution formula
j
y j y+ j
    
∑ λi + k j−k
=
λi + j
k=0

we can rewrite

y+ j y j
     
det = det × det .
λi + j λi + j j−i

j

Noticing that is a lower triangular matrix with diagonal entries equal to one, we

j−i
366 W.C. Chu and L.V. di Claudio

can proceed as follows:

y+ j j
   
LHS of (2.7) = det det
λi + j j−i
(1 + y − λi)λi (1 + y) j
 
= det
(λi + j)!
 n
1

= det ∏ (1 + y − λk)k+λk .
(λi + j)! k=0

This leads us to (2.7) in view of the case c = 1 in (2.5).

Example 2.4 (Amdeberhan and Zeilberger [1, Equation 3]).

n
2x + 2ai + 2 j (2k)! (2x + 2ak)!
 
n+1
det = a( 2 ) ∏ .
x + ai + j 0≤i, j≤n k=0 (x + ak)! (x + ak + n)!

Proof. Expressing the binomial coefficient in terms of shifted factorials

2x + 2ai + 2 j
 
(1/2)x+ai+ j x+ai+ j
= 4 ,
x + ai + j (1)x+ai+ j

and then applying (2.2) with λi := x + ai, a := 1/2 and c := 1, we get immediately the
formula stated in the example where some trivial simplification has been performed.

Putting a := x, c := x + r + 1 and λi := ai in (2.2), we can deduce the following:

Example 2.5 (Amdeberhan and Zeilberger [1, Equation 13]).

n
(x + ai + j − 1)! k! (r + 1)k
 
n+1
det
(x + r + ai + j)!
= a( 2 )
∏ (x + ak)n+r+1 .
0≤i, j≤n k=0

Example 2.6 (Amdeberhan and Zeilberger [1, Equation 15]).

n
1 k! (k + 1)!
 
det =∏ .
(r + i + j)2 0≤i, j≤n k=0 (k + r)n+2

This follows easily from (2.3) by specifying a := r, c := 2 + r and λi := i.

n
Theorem 2.7. Let c and λk , µk k=0 be complex numbers. Then there holds


(λi + µ j ) j n
(c − µk )k
 
det = ∆(λ) ∏ . (2.8)
(λi + c) j 0≤i, j≤n k=0 (c + λ k )n

Proof. This can be considered as an extension

 of (2.2) and (2.3) where the parameter a
n
has been replaced by a complex sequence µ j j=0 .
Binomial Determinant Evaluations 367

For a complex number a, recalling the Saalschütz formula [2, §2.2], we have
(c − a) j (λi + µ j ) j − j, c − µ j, a + λi
 
=3 F2 1
(c + λi) j (µ j − a) j 1 + a − j − µ j , c + λi
j
(a + λi )k (− j)k (c − µ j )k
= ∑ (c + λi)k k!(1 + a − j − µ j)k ,
k=0

which yields the determinant factorization

(c − a) j (λi + µ j ) j (− j)i (c − µ j )i (a + λi) j
     
det = det × det .
(c + λi) j (µ j − a) j i!(1 + a − j − µ j)i (c + λi ) j
Evaluate the right hand side by (2.3) and
n
(− j)i (c − µ j )i (c − µk )k
 
det =∏
i!(1 + a − j − µ j)i k=0 (µ k − a)k

for the matrix concerned is lower triangular. We have the following closed form:
n
(c − a) j (λi + µ j ) j (c − a)k (c − µk )k
 
det = ∆(λ) ∏ .
(c + λi) j (µ j − a) j k=0 (c + λk )n (µk − a)k

Extracting the common row and column factors, we obtain

(λi + µ j ) j (c − a) j (λi + µ j ) j n (µk − a)k
   
det
(λi + c) j
= det ∏ (c − a)k
(c + λi) j (µ j − a) j k=0
n
(c − µk )k
= ∆(λ) ∏ ,
k=0 (c + λk )n

which is the formula stated in the theorem.

Letting c → ∞ in (2.8), we get the following limiting case.
Corollary 2.8 (Determinants in shifted factorials).
h i
det (λi + µ j ) j = ∆(λ).
0≤i, j≤n

Note that this determinant is independent of the sequence {µk }nk=0 .

3. Determinants in Symbolic Factorials

For a complex number x, its shifted factorial of order n can be generalized, via an
arbitrary complex number sequence {εk }∞
k=0 , as

n−1
(x|ε)0 = 1, (x|ε)n = ∏ (x + εk ) for n = 1, 2, . . . . (3.1)
k=0
368 W.C. Chu and L.V. di Claudio

Denote by hmi the set {0, 1, 2, . . . , m − 1}. The k-th elementary symmetric function
in variables xι with ι ∈ hmi is defined by

e0 (x|hmi) ≡ 1 and ek (x|hmi) ≡ 0 for k > m, (3.2)

ek (x|hmi) = ek (x0 , x1 , . . . , xm−1 ) (3.3)

= ∑ ∏ xι for 1 ≤ k ≤ m. (3.4)
σ⊂hmi : |σ|=k ι∈σ

Lemma 3.1 (Extension of the Vandermonde determinant). Let {λ k , εk }nk=0 be complex

numbers. Then there holds

det (λi |ε) j 0≤i, j≤n = ∆(λ). (3.5)

 

Proof. Writing the product (λi |ε) j in terms of elementary symmetric functions

j
(λi |ε) j = ∑ λki e j−k (ε|h ji)
k=0

and then applying the Binet theorem, we have

 j
det (λi |ε) j = det λi × det e j−i (ε|h ji) 0≤i, j≤n .
   

One determinant on the right hand side is Vandermonde and the other reduces to one
because the matrix is upper triangular with the diagonal entries equal to one. This
concludes the proof of the lemma.

Example 3.2 (Amdeberhan and Zeilberger [1, Equations 1-2]).

n
1  x + ai + j  (x + ak)!
 
n+1
det
i! y + ai − j
=a( 2 )
∏ (y + ak)!(x − y + 2k)!
0≤i, j≤n k=0

× ∏ {1 + x + y + (i + j)a},
0≤i< j≤n

n
1 x + ai + j −1 (x − y + 2k)!(y + ak − n)!
  
n+1
det =(−a)( 2 ) ∏
i! y + ai − j 0≤i, j≤n k=0 (x + ak + n)!

× ∏ {1 + x + y + (i + j)a}.
0≤i< j≤n

Proof. The general entries of the first determinant can be rewritten as

1  x + ai + j  (x + ai)!
= × (1 + x + ai) j(1 + y + ai − j) j,
i! y + ai − j i!(y + ai)!(x − y + 2 j)!
Binomial Determinant Evaluations 369

where the mixed (i, j)-factors can be expressed as

j−1
(1 + x + ai) j (1 + y + ai − j) j = ∏ (1 + x + ai + k)(y + ai − k)
k=0

j−1
= ∏ {(1 + x + ai)(y + ai) − k(1 + x − y + k)}.
k=0

Then the first determinant formula in Example 3.2 follows from the setting
λi : = (1 + x + ai)(y + ai),
εk : = −k(1 + x − y + k)
in (3.5) after some routine simplifications.
For the second determinant, we can reformulate the general matrix entries as
1  x + ai + j −1 (x − y + 2 j)!(y + ai − n)!
= × (1 + x + ai + j)n− j(1 + y + ai − n)n− j,
i! y + ai − j i!(x + ai + n)!
where the mixed (i, j)-factors on the previous line can be reformulated as
n− j−1
(1 + x + ai + j)n− j(1 + y + ai − n)n− j = ∏ (x + ai + n − k)(1 + y + ai − n + k)
k=0

n− j−1
= ∏ {(x + ai + n)(1 + y + ai − n)
k=0

+ k(x − y + 2n − k − 1)}.
Then the second determinant formula in Example 3.2 follows from reversing the column
order j → n − j and setting
λi : = (x + ai + n)(1 + y + ai − n),
εk : = k(x − y + 2n − k − 1)
in (3.5) after some trivial simplifications.
Theorem 3.3 (Generalized Vandermonde determinant). Let εk = (ε0k , ε1k , . . . , εk−1
k ) be
complex sequences for 0 ≤ k ≤ n. Then we have the following generalized Vandermonde
determinant formula
det (λi |ε j ) j 0≤i, j≤n = ∆(λ). (3.6)
 

Proof. This can be considered as a generalization of Lemma 3.1. Similar to the proof
of Lemma 3.1, the general entries of the determinant can be restated as
j
∑ λki e j−k (ε0j , ε1j , . . . , ε j
j−1
(λi |ε j ) j = ).
k=0

Then the same reason as Lemma 3.1 leads us to the conclusion of the theorem.
370 W.C. Chu and L.V. di Claudio

Example 3.4 (Amdeberhan and Zeilberger [1, Equation 14]).

x + ai + c j y − ai + c j
   n+1
= (−a)( 2 ) ∏

det j! x − y + (i + j)a .

j j 0≤i, j≤n 0≤i< j≤n

Proof. The general matrix entries reads as

x + ai + c jy − ai + c j 1
j! = (−x − ai − c j) j (−y + ai − c j) j
j j j!

1 j−1 
= ∏ (x + ai + c j − k)(y − ai + c j − k)
j! k=0

1 j−1 
= ∏ (x + ai)(y − ai) + c j(x + y + c j)
j! k=0

− k(x + y + 2c j − k) .
Putting in Theorem 3.3
λi = (x + ai)(y − ai),

εkj = c j(x + y + c j) − k(x + y + 2c j − k),

and then performing some simplifications, we get the determinant formula stated in the
example.
In Theorem 3.3, letting εkj = µ j + k for 0 ≤ k < j, we recover Corollary 2.8.
In Theorem 3.3, letting εkj ≡ µ j for 0 ≤ k < j, we have the following special case.
Proposition 3.5 (Extension of the Vandermonde determinant). Let {λ k , µk }nk=0 be com-
plex numbers. Then we have the determinant identity
det (λi + µ j ) j 0≤i, j≤n = ∆(λ). (3.7)
 

In Theorem 3.3, letting εkj = µ j − k for 0 ≤ k < j, we have another special case.
Proposition 3.6 (Extension of the Vandermonde determinant). Let {λ k , µk }nk=0 be com-
plex numbers. Then we have the binomial determinant identity
λi + µ j  ∆(λ)
 
det = n . (3.8)
j 0≤i, j≤n ∏ k=0 k!

Naturally, both propositions can also be proved directly by means of the binomial
theorem and the Vandermonde convolution formula on binomial coefficients.
Theorem 3.7 (Determinant in symbolic factorials). Let ε = {εk }k≥0 be a complex se-
quence and {λκ , µκ }nκ=0 complex numbers. Then we have

  0, 0 ≤ m < n,
det (λi + µ j |ε)m = (3.9)
∆(λ)∇(µ) ∏n n
m n.


0≤i, j≤n k=0 k , =
Binomial Determinant Evaluations 371

Proof. It is curious to notice that the result is independent of ε-sequence.

Expanding the general matrix entries in terms of elementary symmetric functions
m−1
(λi + µ j |ε)m = ∏ (λi + µ j + εk )
k=0
m
= ∑ λki em−k (ε + µ j |hmi),
k=0
and then applying the Binet theorem, we have
j
h i h i
LHS of (3.9) = det λi × det em−i (ε + µ j |hmi)
0≤i, j≤n
h i
= ∆(λ) × det em−i (ε + µ j |hmi) .
0≤i, j≤n
If 0 ≤ m < n, the last determinant just displayed vanishes since all the entries appearing
in its last row are equal to zero. For m = n, reversing the rows of the determinant leads
us to h i n+1
h i
det em−i (ε + µ j |hmi) = (−1)( 2 ) × det ei (ε + µ j |hni) .
The further expansion in terms of elementary symmetric functions
|σ|
∏ ει + µ j = ∑ µ kj e|σ|−k (ε|σ)


ι∈σ k=0
enables us to manipulate the determinant entries in the following manner
ei (ε + µ j |hni) = ∑ ∏ ει + µ j


σ⊂hni : |σ|=i ι∈σ

i
= ∑ ∑ µ kj ei−k (ε|σ)
σ⊂hni : |σ|=i k=0

i
= ∑ µ kj ∑ ei−k (ε|σ),
k=0 σ⊂hni : |σ|=i

which allows us to simplify further the determinant in (3.9) as follows:

" #
n+1
LHS of (3.9) = (−1)( 2 ) ∆(λ)∆(µ) × det ∑
ei− j (ε|σ) .
σ⊂hni : |σ|=i 0≤i, j≤n
The last matrix on the right hand side is in fact lower triangular with diagonal entries
n
 
equal to for 0 ≤ k ≤ n. This can be reproduced as
k
n  
n+1 n
LHS of (3.9) = (−1)( 2 ) ∆(λ)∆(µ) × ∏
k=0
k
n
n
= ∆(λ)∇(µ) × ∏
 
,
k=0
k
372 W.C. Chu and L.V. di Claudio

which completes the proof of the theorem.

The special case εk = −k of Theorem 3.9 reads as:
Proposition 3.8 (Determinant in binomial coefficients). Let {λk , µk }nk=0 be complex
numbers. Then we have the following binomial determinant identity


λi + µ j
 0,
 0 ≤ m < n,
det = ∆(λ)∇(µ) (3.10)
m 0≤i, j≤n  n
 , m = n.
∏k=0 (k!) 2

Proof. It can also be proved directly as follows.

By means of the Vandermonde formula, the determinant can be rewritten as
 m  
λ µj

LHS of (3.10) = det ∑ i
k=0
k m − k 0≤i, j≤n

λi µj
   
= det × det .
j 0≤i, j≤n m − i 0≤i, j≤n

When 0 ≤ m < n, the last determinant on the right hand side reduces to zero, which
corresponds to the first part of (3.10). For m = n, we have the following binomial
determinant formulas
λi ∆(λ)
 
det = n ,
j ∏k=0 k!
µj ∆(µ)
 
det = n ,
i ∏k=0 k!
in view of Lemma 3.1. Combining both evaluations, we get

λi + µ j λi µj
     
det = det × det
n j n−i

∆(λ) µj
 
n+1
= (−1) ( 2 ) det
∏nk=0 k! i

= (−1)( 2) ∆(λ) ∆(µ) ,

n+1

∏nk=0 k! ∏nk=0 k!
which is equivalent to the second part of (3.10).
Another special case εk = 0 of Theorem 3.9 reads as:
Proposition 3.9 (Determinant in binomial polynomials). Let {λk , µk }nk=0 be complex
numbers. Then we have the following determinant identity

0, 0 ≤ m < n,
det (λi + µ j )m 0≤i, j≤n = (3.11)
 
∆(λ)∇(µ) ∏n n
m n.


k=0 k , =
Binomial Determinant Evaluations 373

Proof. A direct demonstration can be analogously produced as follows:

Combining binomial theorem and Binet’s theorem, we have
 m 
m k m−k

LHS of (3.11) = det ∑ λi µ j
k=0
k 0≤i, j≤n
hm
j
h i i
= det λi × det µm−i .
0≤i, j≤n i j 0≤i, j≤n

Now if 0 ≤ m < n, the last determinant reduces to zero. Instead for m = n, both determi-
nants on the right hand side can be expressed in terms of the Vandermonde determinant,
which lead us easily to the result stated in the proposition.
Example 3.10 (Amdeberhan and Zeilberger [1, Equations 8-9]).

h i 0, 0 ≤ m < n,
det (x + ai + b j)m =
0≤i, j≤n (n!)n+1 (−ab)(n+1
2 ), m = n,

det (x + ai + b j)n(y + ai − b j)n 0≤i, j≤n

 

n+1
= (n!)n+1 (ab)( )
∏ (x + y + ai + a j)(x − y + bi + b j) .

2
0≤i< j≤n

They are special cases of (3.11) respectively by specifying the parameters in the
following manner:
 λ = x + ai,  λ = xy + ai(x + y + ai),
i i
&
µ j = b j, µ j = b j(y − x − b j).

4. Determinants in Symbolic Fractions

n
Theorem 4.1 (Determinant in factorial fractions). Let a, c, d and λk , µk k=0 be com-


plex numbers. Then there holds determinant evaluation:

(aλi + µ j ) j ∏0≤i< j≤n a(d + j) − c(i + µ j )
  
det = ∆(λ) . (4.1)
(cλi + d)1+ j 0≤i, j≤n ∏nk=0 (cλk + d)n+1
Proof. Considering λi as a variable, we have the following partial fraction expansion
j
(aλi + µ j ) j ωk j
=∑ ,
(cλi + d)1+ j k=0 cλi + k + d
where the connected coefficients are determined by
(aλi + µ j ) j
ωk j = lim (cλi + k + d)
λi →−(d+k)/c (cλi + d) j+1

(−1)k  j 
× µ j − a(d + k)/c j .


=
j! k
374 W.C. Chu and L.V. di Claudio

Therefore we have the following determinant decomposition:

(aλi + µ j ) j 1
   
det = det ωi j 0≤i, j≤n × det
 
.
(cλi + d)1+ j 0≤i, j≤n cλi + j + d 0≤i, j≤n

The matrix of the first determinant on the right hand side is upper triangular and its
determinant is equal to the product of its diagonal elements:
n n
n (−1)k
det ωi j = ∏ ω j j = ∏ µ j − a(d + j)/c j ∏
 
j=0 j=0 k=0 k!

n
n+1 1
= c −( )
∏ a(d + j) − c(i + µ j) ∏ k! .

2
0≤i< j≤n k=0

While the second determinant on the right side can be evaluated by the Cauchy deter-
minant (1.4) as follows:

1 n+1 ∏0≤i< j≤n (λi − λ j )(i − j)

 
det = c( 2 )
cλi + j + d ∏ni, j=0 (cλi + j + d)
n
n+1 k!
= c( 2 ) ∆(λ)
∏ (cλk + d)n+1 .
k=0

Putting both determinant evaluations, we get (4.1). This completes the proof of Theo-
rem 4.1.

Example 4.2 (Amdeberhan and Zeilberger [1, Equations 4-5-6]).

x + ai + c j y + bi − j
   
det i! (4.2)
j n− j 0≤i, j≤n

∏ (y − j + 1)a − (x + c j − i)b ,

=
0≤i< j≤n

x + ai + jy + bi + c j
  
det i! (4.3)
j n− j 0≤i, j≤n

∏ (y + (n − i − 1)c − j + 1)a − (x + n − j + 1)b ,


=
0≤i< j≤n

1 x + ai + c jy + bi + j−1
  
det (4.4)
i! j j 0≤i, j≤n

n
(y + bk)!
=∏ ∏ (y + j)a − (x + c j − i)b .

k=0 (y + bk + n)! 0≤i< j≤n

Proof. Separating i-factors and j-factors, we can reformulate the general matrix entries
Binomial Determinant Evaluations 375

displayed in (4.2) and (4.4) respectively as


x + ai + c j y + bi − j
  i! (−1 − y − bi)n+1 (−x − ai − c j) j
i! = (−1)n × ,
j n− j j! (n − j)! (−1 − y − bi) j+1
1 x + ai + c jy + bi + j−1 y + bi (−x − ai − c j) j
= (−1) j × ,
i! j j i! (y + bi) j+1
which lead us to the desired formulas in view of Theorem 4.1 under the following
parameter replacements:
λi → i λi → i
 

 

 µ j → −x − c j  µ j → −x − c j

 


 

a → −a & a → −a
c → −b c →b

 


 

 
d → −y − 1 d → y.

 


For determinant (4.3), it can be restated by reversing the column-indices as

x + ai + j y + bi + c j
   
det i!
j n− j 0≤i, j≤n

y cn bi c j x n ai j
  
n+1 + + − + + −

( )
= (−1) 2 det i! .
j n− j 0≤i, j≤n

Under the following parameter specification

a→b
x → y + cn
b→a
y → x + n,
c → −c
the determinant (4.2) becomes equivalently (4.3). This completes the verification of all
the determinant identities in the example.
n
Theorem 4.3 (Generalized Cauchy determinant). Let a, c, d and λk , µk , εk , δk k=0 be


complex numbers. Then there holds the following symbolic determinant identity:
(aλi + µ j |ε) j ∏0≤i< j≤n a(d + δ j ) − c(εi + µ j )
  
det = ∆(λ) . (4.5)
(cλi + d|δ)1+ j 0≤i, j≤n ∏nk=0 (cλk + d|δ)n+1
Proof. This can be considered as a symbolic version of Theorem 4.1. The Cauchy
determinant is its very special case a = c = 1, d = µ j ≡ 0 and εk = δk for k = 0, 1, 2, . . ..
Expanding the rational function in partial fractions, we have
j−1
(aλi + µ j |ε) j ∏ (aλi + µ j + ει )
= ι=0
j
(cλi + d|δ)1+ j ∏ι=0 (cλi + d + δι )
j
Ωk j
= ∑ cλi + d + δk ,
k=0
376 W.C. Chu and L.V. di Claudio

where the connected coefficients are determined by the following limit relation
j−1
∏ι=0 (aλi + µ j + ει )
Ωk j = lim (cλi + d + δk ) j
λi →−(d+δk )/c ∏ι=0 (cλi + d + δι )
j−1 
∏ι=0 a(d + δk ) − c(ει + µ j )
= j
.
c j ∏ι=0 : ι6=k (δk − δι )

This leads us to the following determinant factorization

(aλi + µ j |ε) j 1
   
det = det Ωi j 0≤i, j≤n × det
 
.
(cλi + d|δ)1+ j 0≤i, j≤n cλi + d + δ j 0≤i, j≤n

Noting that the matrix Ωi j is upper triangular, we can calculate its determinant through
 

the product of its diagonal elements as follows:

j−1 
n n
∏ι=0 a(d + δ j ) − c(ει + µ j )
det Ωi j = ∏ Ω j j = ∏
 
j−1
j=0 j=0 c j ∏ι=0 (δ j − δι )

∏0≤i< j≤n a(d + δ j ) − c(εi + µ j )


−(n+1)
=c 2 .
∏0≤i< j≤n (δ j − δi )
While the second determinant can be evaluated by the Cauchy determinant as

1 n+1 ∏0≤i< j≤n (λi − λ j )(δi − δ j )

 
det = c( 2 )
cλi + d + δ j ∏ni, j=0 (cλi + d + δ j )
n+1 ∆(λ) ∆(δ)
= c( 2) .
∏ni, j=0 (cλi + d + δ j )

Their combination yields the determinant identity (4.5). This completes the proof of
Theorem 4.3.

References

1. T. Amdeberhan and D. Zeilberger, Determinants through the looking glass, Adv. Appl. Math.
27 (2001) 225–230.
2. W.N. Bailey, Generalized Hypergeometric Series, Cambridge University Press, Cambridge,
1935.
3. L. Carlitz, Some determinants of q-binomial coefficients, J. Reine Angew. Math. 226 (1967)
216–220.
4. W. Chu, On the evaluation of some determinants with q-binomial coefficients, Systems Sci.
Math. Sci. 8 (4) (1988) 361–366.
5. W. Chu, Binomial convolutions and determinant identities, Discrete Math. 204 (1-3) (1999)
129–153.
6. W. Chu, Generalizations of the Cauchy determinant, Publ. Math. Debrecen 58 (3) (2001)
353–365.
Binomial Determinant Evaluations 377

7. C. Krattenthaler, Advanced determinant calculus, Séminaire Lotharingien Combin. 42

(1999) Article B42q.
8. A.M. Ostrowski, On some determiants with combinatorial numbers, J. für Math. 216 (1963)
25–30.
9. T. Muir, A Treatise on the Theory of Determinants, Dover Publications, New York, 1960.