The multivariate zeta function, also called the Euler-Zagier multiple zeta function, extends the Riemann zeta function to several complex variables. In descending summation order it is
|
(1)
|
The series is absolutely convergent when for
.
It has a meromorphic continuation to
(Akiyama et al. 2001). At positive integer
arguments with
,
its values are called multiple zeta values, multivariate zeta constants (Bailey et
al. 2006, p. 43), multi-zeta values (Bailey et al. 2006, p. 17),
and multivariate zeta values.
Allowing signs gives the more general Euler
sums
|
(2)
|
(Broadhurst 1996, 1998). For nonzero integer indices , a common notation for the signed sums is
|
(3)
|
(Broadhurst 1996; Bailey et al. 2007, p. 38). In this signed notation, a negative index denotes an alternating factor and a positive exponent in the denominator.
The notation
above uses the arguments directly as exponents.
The notation
(as opposed to
)
is sometimes also used to indicate that a factor of 1 in the numerator is replaced
by a corresponding factor of
.
In addition, the notation
is used in quantum field theory.
In particular, for ,
these correspond to the usual Euler sums
|
(4)
| |||
|
(5)
| |||
|
(6)
| |||
|
(7)
|
(Broadhurst 1996).
Multivariate zeta functions (and their derivatives) also arise in the closed-form evaluation of definite integrals involving the log cosine function (Oloa 2011).
These sums satisfy
|
(8)
|
for , as well as
|
(9)
|
for nonnegative integer and positive integer
(Bailey et al. 2007). These give the special cases
|
(10)
| |||
|
(11)
| |||
|
(12)
|
(Bailey et al. 2007).
A different kind of special case is given by
|
(13)
|
(Borwein and Bailey 2003, p. 26; Borwein et al. 2004, Ch. 2, Ex. 29).
Other special values include
|
(14)
| |||
|
(15)
| |||
|
(16)
| |||
|
(17)
| |||
|
(18)
| |||
|
(19)
| |||
|
(20)
|
(Bailey et al. 2007, pp. 223 and 251). Closed forms are known for the convergent signed sums of weight (Bailey et al. 2006).
For positive integer , Borwein et al. (1997) conjectured the identity
|
(21)
|
where each underbrace denotes repetitions of the indicated pair. Zhao (2010) proved the
identity.
The Euler-Zagier multiple zeta function is implemented in the Wolfram Language as MultipleZeta[{s1, ..., sk}].