Abstract
The refined Kaneko–Zagier conjecture claims that the algebras spanned by two kinds of “completed” finite multiple zeta values, called \({\widehat{\mathcal {A}}}\)- and \({\widehat{\mathcal {S}}}\)-MZVs, are isomorphic. Recently, Komori defined \({\widehat{\mathcal {S}}}\)-MZVs of general integer (i.e., not necessarily positive) indices, extending the existing definition for positive indices. In view of the refined Kaneko–Zagier conjecture, Komori’s work suggests that these extended values are closely connected to \({\widehat{\mathcal {A}}}\)-MZVs of general indices, which can be defined in an obvious way. In this paper, we show that the generalization of the refined Kaneko–Zagier conjecture for general integer indices is actually deduced from the conjecture for positive indices. The key ingredient is an inductive formula for \({\widehat{\mathcal {A}}}\)-MZVs or \({\widehat{\mathcal {S}}}\)-MZVs of indices which contain at least one non-positive entry.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
For any \(r \in \mathbb {Z}_{\ge 0}\), we call an r-tuple \({\varvec{k}}=(k_1, \dots , k_r)\) of integers an index. Such \({\varvec{k}}\) is said positive if all entries \(k_i\) are positive. In order to emphasize that a given index is not necessarily positive, we often call it a general index. The length r of an index \({\varvec{k}}=(k_1,\ldots ,k_r)\) is called the depth of \({\varvec{k}}\) and, if \({\varvec{k}}\) is positive, the sum of its entries \(k_1+\cdots +k_r\) is called the weight of \({\varvec{k}}\). Let \(\varnothing \) denote the unique index of depth 0, called the empty index. We regard \(\varnothing \) as a positive index of weight 0.
A positive index \({\varvec{k}}=(k_1, \dots , k_r)\) is said admissible if \({\varvec{k}}=\varnothing \) or \(k_r\ge 2\). For an admissible positive index \({\varvec{k}}=(k_1, \dots , k_r)\), the multiple zeta value (MZV) \(\zeta ({\varvec{k}})\) is a real number defined by
MZVs have been extensively investigated from various viewpoints, and several variants of them are also considered. In particular, since the discovery of a surprising conjecture by Kaneko and Zagier [5], two kinds of finite multiple zeta values, which we call \(\mathcal {A}\)- and \(\mathcal {S}\)-MZVs, are actively studied. In this paper, we treat the completed versions of them, called \({\widehat{\mathcal {A}}}\)-MZVs [10, 11] and \({\widehat{\mathcal {S}}}\)-MZVs [4, 8]. We recall the definitions of them below.
Let \({\varvec{k}}=(k_1,\ldots ,k_r)\) be a positive index, not necessarily admissible. Then the first variant, the \({\widehat{\mathcal {A}}}\)-MZV \(\zeta _{{\widehat{\mathcal {A}}}}({\varvec{k}})\), is an element of the \(\mathbb {Q}\)-algebra
in which \(\mathbb {Q}\) is diagonally embedded. We define, for each prime number p, an element of the localization \(\mathbb {Z}_{(p)}\) of \(\mathbb {Z}\) with respect to the prime ideal (p) by
Then \(\zeta _{{\widehat{\mathcal {A}}}}({\varvec{k}})\) is defined by
Next, let \(\mathcal {Z}\) denote the \(\mathbb {Q}\)-subalgebra of \(\mathbb {R}\) generated by all MZVs and set \(\overline{\mathcal {Z}}:=\mathcal {Z}/\pi ^2\mathcal {Z}\) (note that \(\pi ^2=6\zeta (2)\in \mathcal {Z}\)). Then the second variant of MZV, the \({\widehat{\mathcal {S}}}\)-MZV \(\zeta ^{}_{{\widehat{\mathcal {S}}}}({\varvec{k}})\), is a formal power series in an indeterminate t with coefficients in \(\overline{\mathcal {Z}}\) defined by
Here, \(\zeta ^{*}({\varvec{k}}) \in \mathcal {Z}\) denotes the harmonic regularized MZV, which is defined for any positive index \({\varvec{k}}\) (for the definition of the regularized MZVs, refer to [3]). This notion of the \({\widehat{\mathcal {S}}}\)-MZV is equivalent to the \(\Lambda \)-adjoint MZVs introduced by Jarossay [4].
It is conjectured that the \({\widehat{\mathcal {A}}}\)-MZVs and the \({\widehat{\mathcal {S}}}\)-MZVs satisfy exactly the same algebraic relations (the refined Kaneko–Zagier conjecture [4, 8, 10]; see Sect. 4 for more details). Though this conjecture seems still far from resolution, it suggests that some phenomenon observed in \({\widehat{\mathcal {A}}}\)-MZVs should have an analogue in \({\widehat{\mathcal {S}}}\)-MZVs and vice versa.
Now let us observe that the definition of \(\zeta ^{}_{\widehat{\mathcal {A}}}({\varvec{k}})\) given by (1.1) and (1.2) is meaningful for general indices \({\varvec{k}}\) as well. On the other hand, the definition (1.3) of \(\zeta ^{}_{\widehat{\mathcal {S}}}({\varvec{k}})\) does not work for such \({\varvec{k}}\), since the regularized MZVs are defined only for positive indices. This gap was filled by Komori [6], who found a natural definition of \(\zeta ^{}_{\widehat{\mathcal {S}}}({\varvec{k}})\) for general \({\varvec{k}}\). In fact, he introduced a complex analytic function \(\zeta _{\widehat{\mathcal {U}}}(s_1,\ldots ,s_r;t)\), called the unified multiple zeta function. This function is entire with respect to \(s_1,\ldots ,s_r\) and satisfies the interpolation property
for any positive index \({\varvec{k}}\). Then Komori defined \(\zeta ^{}_{\widehat{\mathcal {S}}}({\varvec{k}})\) for general \({\varvec{k}}\) to be \(\zeta ^{}_{\widehat{\mathcal {U}}}({\varvec{k}};t)\) modulo \(\pi i\mathcal {Z}[\pi i]\llbracket t \rrbracket \).
Given these extensions of \({\widehat{\mathcal {A}}}\)- and \({\widehat{\mathcal {S}}}\)-MZVs for general indices, it is natural to ask whether the refined Kaneko–Zagier conjecture is generalized to these extended values. In this paper, we will prove that this generalization can be actually deduced from the original one (the precise statement is given in Theorem 4.4):
Theorem 1.1
If the refined Kaneko–Zagier conjecture concerning the \({\widehat{\mathcal {A}}}\)- and \({\widehat{\mathcal {S}}}\)-MZVs of positive indices is true, then the same is true for those values of general indices.
Indeed, this result is an immediate consequence of the following:
Theorem 1.2
For any index \({\varvec{k}}\in \mathbb {Z}^r\), there is a collection \((c_{{\varvec{k}},{\varvec{l}}}(x))_{\varvec{l}}\) of rational polynomials \(c_{{\varvec{k}},{\varvec{l}}}(x)\in \mathbb {Q}[x]\) indexed by a finite number of positive indices \({\varvec{l}}\) such that
hold simultaneously, where we set \({\varvec{p}}:=\bigl ((p \bmod {p^n})_{p}\bigr )_{n}\in {\widehat{\mathcal {A}}}\). Moreover, the depth (resp. the weight) of these positive indices \({\varvec{l}}\) are at most the number (resp. the sum) of positive entries of \({\varvec{k}}\).
For example, we have
for any integer \(k\ge 3\).
The proof of Theorem 1.2 is based on a formula which reduces the depth r whenever \({\varvec{k}}\) contains a non-positive entry. First we state this formula for \(\zeta ^{}_{\widehat{\mathcal {A}}}({\varvec{k}})\). Let \(B_n\) denote the Seki-Bernoulli number defined by \(ze^z/(e^z-1)=\sum _{n=0}^{\infty }B_n z^n/n!\) and \(\delta _{xy}\) denote the Kronecker delta.
Theorem 1.3
Let \({\varvec{k}}=(k_1, \dots , k_r)\) be an index of depth \(r\ge 2\), and suppose that \(k_i=-k\) for some \(i\in \{1,\ldots ,r\}\) and \(k\ge 0\).
-
(i)
If \(i=1\), we have
$$\begin{aligned} \zeta ^{}_{{\widehat{\mathcal {A}}}}({\varvec{k}})= \frac{1}{k+1}\sum _{j=0}^{k+1}\left( {\begin{array}{c}k+1\\ j\end{array}}\right) B_j\Bigl ((-1)^j\zeta ^{}_{{\widehat{\mathcal {A}}}}(k_2-k-1+j,\ldots ,k_r)&\\ -\delta _{j,k+1}\,\zeta ^{}_{{\widehat{\mathcal {A}}}}(k_2,\ldots ,k_r)&\Bigr ). \end{aligned}$$ -
(ii)
If \(1<i<r\), we have
$$\begin{aligned} \zeta ^{}_{{\widehat{\mathcal {A}}}}({\varvec{k}})= \frac{1}{k+1}\sum _{j=0}^{k+1}\left( {\begin{array}{c}k+1\\ j\end{array}}\right) B_j\Bigl ((-1)^j\zeta ^{}_{{\widehat{\mathcal {A}}}}&(k_1,\dots ,k_{i+1}-k-1+j,\dots ,k_r)\\ \qquad -\zeta ^{}_{{\widehat{\mathcal {A}}}}&(k_1,\dots ,k_{i-1}-k-1+j,\dots ,k_r)\Bigr ). \end{aligned}$$ -
(iii)
If \(i=r\), we have
$$\begin{aligned} \zeta ^{}_{{\widehat{\mathcal {A}}}}({\varvec{k}})= \frac{1}{k+1}\sum _{j=0}^{k+1}\left( {\begin{array}{c}k+1\\ j\end{array}}\right) B_j\Bigl ((-1)^j\zeta ^{}_{{\widehat{\mathcal {A}}}}&(k_1,\dots ,k_{r-1}) {\varvec{p}}^{k+1-j}\\ -\zeta ^{}_{{\widehat{\mathcal {A}}}}&(k_1,\dots ,k_{r-1}-k-1+j)\Bigr ). \end{aligned}$$
The corresponding formula for \(\zeta ^{}_{\widehat{\mathcal {S}}}({\varvec{k}})\) holds, in fact, at the level of the complex analytic function. See Sect. 3 for the definition of the unified multiple zeta function \(\zeta ^{}_{\widehat{\mathcal {U}}}({\varvec{s}};t_+,t_-)\).
Theorem 1.4
Let \({\varvec{s}}=(s_1, \dots , s_r)\) be an r-tuple of complex numbers with \(r\ge 2\), and suppose that \(s_i=-k\) for some \(i\in \{1,\ldots ,r\}\) and an integer \(k\ge 0\).
-
(i)
If \(i=1\), we have
$$\begin{aligned} \zeta _{\widehat{\mathcal {U}}}({\varvec{s}};t_+,t_-)= \frac{1}{k+1}\sum _{j=0}^{k+1}\left( {\begin{array}{c}k+1\\ j\end{array}}\right) B_j\Bigl ((-1)^j\zeta _{\widehat{\mathcal {U}}}&(s_2-k-1+j,\dots ,s_r;t_+,t_-)\\ -\zeta _{\widehat{\mathcal {U}}}&(s_2,\dots ,s_r;t_+,t_-)\,t_+^{k+1-j}\Bigr ). \end{aligned}$$ -
(ii)
If \(1<i<r\), we have
$$\begin{aligned} \zeta _{\widehat{\mathcal {U}}}({\varvec{s}};t_+,t_-)= \frac{1}{k+1}\sum _{j=0}^{k+1}\left( {\begin{array}{c}k+1\\ j\end{array}}\right) B_j\Bigl ((-1)^j\zeta _{\widehat{\mathcal {U}}}&(s_1,\dots ,s_{i+1}-k-1+j,\dots ,s_r;t_+,t_-)\\ -\zeta _{\widehat{\mathcal {U}}}&(s_1,\dots ,s_{i-1}-k-1+j,\dots ,s_r;t_+,t_-)\Bigr ). \end{aligned}$$ -
(iii)
If \(i=r\), we have
$$\begin{aligned} \zeta _{\widehat{\mathcal {U}}}({\varvec{s}};t_+,t_-)= \frac{1}{k+1}\sum _{j=0}^{k+1}\left( {\begin{array}{c}k+1\\ j\end{array}}\right) B_j\Bigl ((-1)^j\zeta _{\widehat{\mathcal {U}}}&(s_1,\dots ,s_{r-1};t_+,t_-)\,t_-^{k+1-j}\\ -\zeta _{\widehat{\mathcal {U}}}&(s_1,\dots ,s_{r-1}-k-1+j;t_+,t_-)\Bigr ). \end{aligned}$$
From Theorem 1.4, we obtain the following:
Corollary 1.5
For any \(r \in \mathbb {Z}_{\ge 0}\) and \({\varvec{k}}\in \mathbb {Z}^r\), we have \(\zeta ^{}_{{\widehat{\mathcal {U}}}}({\varvec{k}}; t_+, t_-) \in \mathcal {Z}[\pi i]\llbracket t_+, t_- \rrbracket \).
The contents of this paper is as follows. In Sect. 2, we prove Theorem 1.3 by using Faulhaber’s formula on power sums. In Sect. 3, we prove Theorem 1.4 along the same line as in the previous section, by considering (a slight modification of) the order on non-zero integers due to Kontsevich. In Sect. 4, we explain the Kaneko–Zagier conjecture and its refinement, and present the precise version of Theorem 1.1. Finally, in Sect. 5, we discuss the values of \(\zeta ^{}_{{\widehat{\mathcal {U}}}}({\varvec{s}}; t_+, t_-)\) at non-positive integer points as an application of our results.
2 Main theorem for \(\zeta _{\widehat{\mathcal {A}}}\)
In this section, we prove Theorem 1.3.
Lemma 2.1
For any non-negative integers k and \(a<b\), we have
Here we use the convention that \(a^{k+1-j}=1\) for \(a=0\) and \(j=k+1\).
Proof
From Faulhaber’s formula
we see that
Then the desired expression follows, since \(B_1=1/2\) and \(B_j=0\) for any odd integer \(j\ge 3\). \(\square \)
Proof of Theorem 1.3
By substituting (2.1) into the definition of \(\zeta _p({\varvec{k}})\), one sees that \(\zeta _p({\varvec{k}})\) satisfies the formula of the same form as given in Theorem 1.3: For example,
Then Theorem 1.3 follows immediately. \(\square \)
Remark 2.2
Given integers \(a_p,b_p\) with \(0\le a_p\le b_p\le p\) for each prime p, we may define a generalization of \(\zeta ^{}_{\widehat{\mathcal {A}}}({\varvec{k}})\) by
where \(a_{\varvec{p}}\) and \(b_{\varvec{p}}\) stand for \(((a_p\bmod p^n)_p)_n,((b_p\bmod p^n)_p)_n\in {\widehat{\mathcal {A}}}\), respectively. Theorem 1.3 has a natural generalization to these values \(\zeta ^{}_{{\widehat{\mathcal {A}}}}({\varvec{k}}; a_{\varvec{p}}, b_{\varvec{p}})\), and the similarity to Theorem 1.4 becomes more apparent. For example, the formula for \(i=1\) is
3 Main theorem for \(\zeta _{\widehat{\mathcal {U}}}\)
In this section, we prove Theorem 1.4. First we recall the definition of the unified multiple zeta function introduced by Komori [6]. For \(t\in \mathbb {C}\setminus [1,\infty )\), we define a version of multiple zeta function of Hurwitz type by
Here the power \((n-t)^s\) is defined by using the principal value of the logarithm \(\log (n-t)\). This series (3.1) converges absolutely and uniformly on compact sets of the region
and meromorphically continued to \((s_1,\ldots ,s_r)\in \mathbb {C}^r\) [7, Theorems 3 and 4]. For \(t_+\in \mathbb {C}\setminus (-\infty ,-1]\) and \(t_-\in \mathbb {C}\setminus [1,\infty )\), we define
where \((-1)^s=e^{\pi is}\) for any \(s\in \mathbb {C}\) (note that we have slightly changed the definition from that given in [6, p. 225]; Komori’s \((t_1,t_2)\) is our \((-t_+,t_-)\)). The function \(\zeta ^{}_{\widehat{\mathcal {U}}}(s_1,\ldots ,s_r;t)\) mentioned in Sect. 1 is defined as
In the following, we often abbreviate the pair of \(t_+\) and \(t_-\) as \(t_\pm \). In [6], it is proved that \(\zeta ^{}_{\widehat{\mathcal {U}}}({\varvec{s}};t_\pm )\) is extended to a holomorphic function on
In particular, \(\zeta ^{}_{\widehat{\mathcal {U}}}({\varvec{s}};t_\pm )\) is an entire function on \(\mathbb {C}^r\) for fixed \(t_\pm \).
The definition (3.2) of \(\zeta ^{}_{\widehat{\mathcal {U}}}({\varvec{s}};t_\pm )\) can be rewritten by using the so-called Kontsevich order. Fixing \(t_\pm \), we define sets \(I_+\), \(I_-\) and I by
and introduce a total order \(\prec \) on the set I by
Then, if \(\Re (s_i)>1\) for all i, we may write (3.2) as
Here, for \(a=-n+t_- \in I_-\), we understand \(1/a^{s}=(-1)^s/(n-t_-)^s\).
Lemma 3.1
For \(a,b\in I\) with \(a\prec b\), set
Then \(F_a^b(s)\) can be continued to an entire function on \(\mathbb {C}\).
Proof
Let \(\zeta ^{}_{H}(s, a)\) denote the Hurwitz zeta function defined by
Note that it satisfies \(\zeta _H(s,a)=\zeta (s;1-a)\), where the right hand side is defined by (3.1) with \(r=1\). It is well-known that \(\zeta ^{}_{H}(s, a)\) is meromorphically continued to the whole complex s-plane, and holomorphic except the pole at \(s=1\) with the principal part \(1/(s-1)\) (see e.g. [1, Theorem 12.4]).
If \(a,b\in I_+\) or \(a,b\in I_-\), then \(F_a^b(s)\) is a finite sum and the statement is obvious. On the other hand, if \(a=n_a+t_+\in I_+\) and \(b=-n_b+t_-\in I_-\), then we have
Hence this can be meromorphically continued to \(\mathbb {C}\), and the only possible pole at \(s=1\) is removed, since the residue is \(1+(-1)^1=0\). \(\square \)
By the definition of \(F^{b}_{a}(s)\), for \(s_1, \dots , s_r \in \mathbb {C}\) satisfying \(\Re (s_1),\ldots ,\Re (s_r)>1\), we see that
Lemma 3.2
Let C be a positive real number. The series expression (3.3) holds for \(s_1, \dots , s_r \in \mathbb {C}\) satisfying \(\Re (s_i)>-C\) and \(\Re (s_j)>C+2\) \((\forall j \ne i)\). In particular, if \(k \in \mathbb {Z}_{\ge 0}\) and \(s_j \in \mathbb {C}\) \((j \ne i)\) satisfying \(\Re (s_j)>k+2\), we have
Proof
Let us write \(\mathfrak {D}_C\) for the domain of \((s_1,\ldots ,s_r)\in \mathbb {C}^r\) defined by \(\Re (s_i)>-C\) and \(\Re (s_j)>C+2\) \((\forall j \ne i)\). For the first assertion, it suffices to show that the expression in (3.3) gives a holomorphic function on \(\mathfrak {D}_C\) (recall that \(\zeta ^{}_{\widehat{\mathcal {U}}}\) is known to be holomorphic on \(\mathbb {C}^r\)).
We divide the series in the right hand side of (3.3) as
where \((*)\) stands for the condition \(t_+\prec a_1\prec \cdots \prec a_{i-1}\prec a_{i+1}\prec \cdots \prec a_r\prec t_-\). Then the first sum is equal to
and all series \(\zeta (s_1,\dots ,s_j;-t_+)\) and \(\zeta (s_r,\dots ,s_{j+1}; t_-)\) appearing here converge uniformly on compact sets of \(\mathfrak {D}_C\). Hence the first sum gives a holomorphic function on \(\mathfrak {D}_C\). The same argument works for the second sum as well. For the third sum, note that
holds for \(a_{i-1}=n_{i-1}+t_+\in I_+\) and \(a_{i+1}=-n_{i+1}+t_-\in I_-\). Hence the third sum is equal to
Two nested series converge absolutely and uniformly on compact sets in \(\mathfrak {D}_C\); this can be seen, for example, by observing that
Therefore, the third sum also gives a holomorphic function on \(\mathfrak {D}_C\), and the proof of our first assertion is complete. The second assertion is deduced from the first, applied to all \(C>k\). \(\square \)
Lemma 3.3
For a non-negative integer k and \(a, b \in I\) with \(a\prec b\), we have
Proof
For \(a,b \in I_+\) or \(a,b \in I_-\), the statement follows immediately from Faulhaber’s formula. If \(a\in I_+\) and \(b\in I_-\), we have
Let \(B_n(x)\in \mathbb {Q}[x]\) be the n-th Bernoulli polynomial defined by \(ze^{xz}/(e^z-1)=\sum _{n=0}B_n(x)z^n/n!\) (for the properties of Bernoulli polynomials, we refer to [2, §4.3]). Since \(\zeta (s;a)=\zeta _H(s,1-a)\) and
we have
Thus we obtain
\(\square \)
Proof of Theorem 1.4
By using Lemmas 3.2 and 3.3, this is proved in the same way as Theorem 1.3. \(\square \)
Proof of Theorem 1.2
We proceed by induction on the depth r of \({\varvec{k}}=(k_1,\ldots ,k_r)\). The statement is trivial for \(r=0\), since we have \(\zeta ^{}_{\widehat{\mathcal {A}}}(\varnothing )=1\) and \(\zeta ^{}_{\widehat{\mathcal {S}}}(\varnothing )=1\) by convention. Let us consider the case \(r=1\). If \({\varvec{k}}=(k)\) with \(k>0\), there is nothing to prove. On the other hand, if \({\varvec{k}}=(-k)\) with \(k\ge 0\), we have
and
hence the statement is true.
Let \({\varvec{k}}=(k_1,\ldots ,k_r)\in \mathbb {Z}^r\) with \(r\ge 2\). Again, there is nothing to do if \({\varvec{k}}\in \mathbb {Z}_{>0}^r\). Otherwise, we can apply Theorem 1.3 and Theorem 1.4 to see that \(\zeta ^{}_{\widehat{\mathcal {A}}}({\varvec{k}})\) and \(\zeta ^{}_{\widehat{\mathcal {S}}}({\varvec{k}})\) admit a common expression as a \(\mathbb {Q}\)-linear combination of the values of depth \(r-1\). Thus, by the induction hypothesis, the statement is also true for \({\varvec{k}}\). This completes the proof of Theorem 1.2. \(\square \)
4 Kaneko–Zagier conjecture
In this section, we present the precise statement of Theorem 1.1 related with the Kaneko–Zagier conjecture and its refinement.
First let us recall the original conjecture on \(\mathcal {A}\)-MZVs and \(\mathcal {S}\)-MZVs proposed by Kaneko and Zagier [5]. For a general index \({\varvec{k}}\in \mathbb {Z}^{r}\) (\(r\ge 0\)), let
be the \(\mathcal {A}\)-MZV and let \(\mathcal {Z}_{\mathcal {A}}\) denote the \(\mathbb {Q}\)-subalgebra of \(\mathcal {A}\) generated by \(\mathcal {A}\)-MZVs \(\zeta ^{}_{\mathcal {A}}({\varvec{k}})\) for all positive indices \({\varvec{k}}\in \mathbb {Z}^{r}_{>0}\). By construction, we have the identity \(\zeta ^{}_\mathcal {A}({\varvec{k}})=\zeta ^{}_{\widehat{\mathcal {A}}}({\varvec{k}})\bmod {\varvec{p}}\) under the natural isomorphism \(\mathcal {A}\cong {\widehat{\mathcal {A}}}/{\varvec{p}}{\widehat{\mathcal {A}}}\).
Following Komori [6], we define the \(\mathcal {S}\)-MZV (resp. \({\widehat{\mathcal {S}}}\)-MZV) for a general index \({\varvec{k}}\in \mathbb {Z}^r\) by
By Corollary 1.5, they belong to the rings \(\mathcal {Z}[\pi i]/\pi i\mathcal {Z}[\pi i]\cong \overline{\mathcal {Z}}\) and \(\overline{\mathcal {Z}}\llbracket t \rrbracket \), respectively.
Conjecture 4.1
(Kaneko–Zagier conjecture) There exists a (necessarily unique) \(\mathbb {Q}\)-algebra isomorphism \(\mathcal {Z}_{\mathcal {A}} \rightarrow \overline{\mathcal {Z}}\) which sends \(\zeta ^{}_{\mathcal {A}}({\varvec{k}})\) to \(\zeta ^{}_{\mathcal {S}}({\varvec{k}})\) for any positive index \({\varvec{k}}\). In particular, \(\mathcal {A}\)-MZVs and \(\mathcal {S}\)-MZVs of positive indices satisfy the same \(\mathbb {Q}\)-linear relations, i.e., a \(\mathbb {Q}\)-linear relation \(\sum _{\varvec{k}}a_{\varvec{k}}\zeta ^{}_\mathcal {A}({\varvec{k}})=0\) among \(\mathcal {A}\)-MZVs of positive indices holds if and only if the corresponding relation \(\sum _{\varvec{k}}a_{\varvec{k}}\zeta ^{}_\mathcal {S}({\varvec{k}})=0\) among \(\mathcal {S}\)-MZVs holds.
To state the corresponding conjecture for \({\widehat{\mathcal {A}}}\)- and \({\widehat{\mathcal {S}}}\)-MZVs, we equip the \(\mathbb {Q}\)-algebras \({\widehat{\mathcal {A}}}\) and \(\overline{\mathcal {Z}}\llbracket t \rrbracket \) with the \({\varvec{p}}\)-adic and t-adic topology, respectively. Then let \(\mathcal {Z}_{\widehat{\mathcal {A}}}\) denote the closed \(\mathbb {Q}\)-subalgebra of \({\widehat{\mathcal {A}}}\) generated by \({\varvec{p}}\) and \({\widehat{\mathcal {A}}}\)-MZVs \(\zeta ^{}_{\widehat{\mathcal {A}}}({\varvec{k}})\) of positive indices \({\varvec{k}}\in \mathbb {Z}^r\). Explicitly, \(\mathcal {Z}_{\widehat{\mathcal {A}}}\) consists of elements of the form \(\sum _{i=1}^\infty a_i\zeta ^{}_{\widehat{\mathcal {A}}}({\varvec{k}}_i){\varvec{p}}^{n_i}\in {\widehat{\mathcal {A}}}\) with \(a_i\in \mathbb {Q}\), positive indices \({\varvec{k}}_i\), and integers \(n_i\ge 0\) satisfying \(n_i\rightarrow \infty \) (\(i\rightarrow \infty \)).
Conjecture 4.2
(Refined Kaneko–Zagier conjecture, [8]) There exists a (necessarily unique) topological \(\mathbb {Q}\)-algebra isomorphism \(\mathcal {Z}_{{\widehat{\mathcal {A}}}} \rightarrow \overline{\mathcal {Z}}\llbracket t \rrbracket \) which sends \({\varvec{p}}\) to t, and \(\zeta ^{}_{{\widehat{\mathcal {A}}}}({\varvec{k}})\) to \(\zeta ^{}_{{\widehat{\mathcal {S}}}}({\varvec{k}})\) for any positive index \({\varvec{k}}\). In particular, \({\widehat{\mathcal {A}}}\)-MZVs and \({\widehat{\mathcal {S}}}\)-MZVs satisfy the same \(\mathbb {Q}\)-linear relations.
Remark 4.3
At present, it has not been proven that the refined Kaneko–Zagier conjecture (Conjecture 4.2) implies the original Kaneko–Zagier conjecture (Conjecture 4.1). In fact, assuming the refined conjecture, the original one is equivalent to the equality
of ideals in \(\mathcal {Z}_{\widehat{\mathcal {A}}}\). This equality is, however, a spacial case of Rosen’s “asymptotic extension conjecture” [9, Conjecture A], which is unsolved yet. We note that the implication of Conjecture 4.1 from Conjecture 4.2, assuming Rosen’s conjecture, is also proved in [12, Proposition 3.6].
As an obvious consequence of Theorem 1.2, we obtain the following result. The part (ii) is the precise version of Theorem 1.1.
Theorem 4.4
-
(i)
Suppose that Conjecture 4.1 is true. Then the isomorphism \(\mathcal {Z}_{\mathcal {A}} \rightarrow \overline{\mathcal {Z}}\) sends \(\zeta ^{}_{\mathcal {A}}({\varvec{k}})\) to \(\zeta ^{}_{\mathcal {S}}({\varvec{k}})\) for any general index \({\varvec{k}}\). In particular, \(\mathcal {A}\)-MZVs and \(\mathcal {S}\)-MZVs of general indices satisfy the same \(\mathbb {Q}\)-linear relations.
-
(ii)
Suppose that Conjecture 4.2 is true. Then the isomorphism \(\mathcal {Z}_{{\widehat{\mathcal {A}}}} \rightarrow \overline{\mathcal {Z}}\llbracket t \rrbracket \) sends \(\zeta ^{}_{{\widehat{\mathcal {A}}}}({\varvec{k}})\) to \(\zeta ^{}_{{\widehat{\mathcal {S}}}}({\varvec{k}})\) for any general index \({\varvec{k}}\in \mathbb {Z}^r\). In particular, \({\widehat{\mathcal {A}}}\)-MZVs and \({\widehat{\mathcal {S}}}\)-MZVs of general indices satisfy the same \(\mathbb {Q}\)-linear relations.
5 Values at non-positive integer points
As another application of Theorem 1.4, we show a formula for the values of \(\zeta ^{}_{{\widehat{\mathcal {U}}}}({\varvec{s}}; t_\pm )\) at non-positive integer points. This is a generalization of Komori’s formula [6, Theorem 1.17] on \(\zeta ^{}_{\widehat{\mathcal {U}}}(-k_1,\ldots ,-k_r;t)=\zeta ^{}_{\widehat{\mathcal {U}}}(-k_1,\ldots ,-k_r;0,t)\). Our proof looks quite different from Komori’s argument.
Theorem 5.1
For any non-negative integers \(k_1,\ldots ,k_r\), the value \(\zeta ^{}_{\widehat{\mathcal {U}}}(-k_1,\ldots ,-k_r;t_\pm )\) is a polynomial function of rational coefficients in \(t_+\) and \(t_-\). Moreover, the generating function
of these values can be expressed as
Proof
First we treat the case of \(r=1\). Since
this is indeed a rational polynomial. Moreover, we have
This shows the claim for \(r=1\).
For \(r\ge 2\), we proceed by induction on r. By using Theorem 1.4, we can deduce that \(\zeta ^{}_{\widehat{\mathcal {U}}}(-k_1,\ldots ,-k_r;t_\pm )\) is a rational polynomial in \(t_\pm \) from the induction hypothesis. Moreover, by Theorem 1.4 (iii), we have
Therefore, we see that
On the other hand, it is elementary to show that the right hand side of (5.1) satisfies the same recurrence relation. Thus, by induction on r, we obtain the identity (5.1). \(\square \)
Remark 5.2
Theorem 5.1 may be rephrased as follows: the function
has an expansion
with rational polynomial coefficients \(P_r(k_1,\ldots ,k_r;y_\pm )\in \mathbb {Q}[y_\pm ]\), and the identity
holds for any \(k_1,\ldots ,k_r\ge 0\). This was the logical order that Komori pursued in the case of \((y_+,y_-)=(0,y)\) (cf. [6, Theorems 1.1 and 1.17]). In our proof of Theorem 5.1, we have avoided to show the existence of the expansion (5.2) directly from the definition of \(F_r(z_1,\ldots ,z_r;y_\pm )\).
Remark 5.3
We can prove the dihedral symmetries that generalize [Komori, Theorem 1.19]: For a positive integer r, we define two operators A and B on functions f in \(z_1, \ldots , z_r, t_{+}, t_{-}\) by
which extend to an action of the dihedral group \(D_{r+1} = \langle A, B \mid A^{r+1}=B^2=(AB)^2=1 \rangle \) of order \((2r+2)\). Then, by using the expression (5.1), it is proved that \(G_r(z_1, \ldots , z_r; t_{\pm })\) is invariant under this action:
These symmetries provide certain relations among the values \(\zeta _{{\widehat{\mathcal {U}}}}(-k_1,\ldots ,-k_r;t_\pm )\).
Remark 5.4
In the notation of Remark 2.2, we also have
The proof is the same as that of Theorem 5.1.
Data availability
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
References
Apostol, T.M.: Introduction to Analytic Number Theory, Undergraduate Texts in Mathematics. Springer, Berlin (1976)
Arakawa, T., Ibukiyama, T., Kaneko, M.: Bernoulli Numbers and Zeta Functions (with an Appendix by Don Zagier). Springer, Tokyo (2014)
Ihara, K., Kaneko, M., Zagier, D.: Derivation and double shuffle relations for multiple zeta values. Compos. Math. 142, 307–338 (2006)
Jarossay, D.: Adjoint cyclotomic multiple zeta values and cyclotomic multiple harmonic values, preprint (2019). arXiv:1412.5099v5
Kaneko, M., Zagier, D.: Finite multiple zeta values (in preparation)
Komori, Y.: Finite multiple zeta values, symmetric multiple zeta values and unified multiple zeta functions. Tohoku Math. J. 73, 221–255 (2021)
Matsumoto, K.: On analytic continuation of various multiple zeta-functions. In: Bennett, M.A., et al. (eds.) Number Theory for the Millennium, II, pp. 417–440. A K Peters, Natick (2002)
Ono, M., Seki, S., Yamamoto, S.: Truncated \(t\)-adic symmetric multiple zeta values and double shuffle relations. Res. Number Theory 7, Article number 15 (2021)
Rosen, J.: Asymptotic relations for truncated multiple zeta values. J. Lond. Math. Soc. (2) 91(2), 554–572 (2015)
Rosen, J.: The completed finite period map and Galois theory of supercongruences. Int. Math. Res. Not. IMRN 23, 7379–7405 (2019)
Seki, S.: Finite multiple polylogarithms. Doctoral dissertation in Osaka University (2017)
Takeyama, Y., Tasaka, K.: Supercongruences of multiple harmonic \(q\)-sums and generalized finite/symmetric multiple zeta values. Kyushu J. Math. 77(1), 75–120 (2023)
Acknowledgements
The authors would like to thank Professor Koji Tasaka who kindly informed us that Rosen’s conjecture is needed to deduce Conjecture 4.1 from Conjecture 4.2. The authors also deeply appreciate Professor Yasushi Komori’s helpful comments, especially on the content of Sect. 5. Finally, the authors also would like to express their gratitude to the referee for careful reading and many helpful comments.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Masataka Ono is supported by Waseda University Grants for Special Research Projects (Grant no. 2021C-673). Shuji Yamamoto is supported by JSPS KAKENHI JP18H05233, JP18K03221 and JP21K03185.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Ono, M., Yamamoto, S. On the refined Kaneko–Zagier conjecture for general integer indices. Math. Z. 305, 1 (2023). https://doi.org/10.1007/s00209-023-03318-2
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00209-023-03318-2