Hilbert-Kunz multiplicity of binoids (Q2236780): Difference between revisions

In this paper, the authors develops a Hilbert-Kunz theory for binoids and solve several questions typically asociated to this functions in a combinatorial setting. Consider a local ring \((R,\mathfrak{m})\) containing a field \(K\) of characteristic \(p>0\). For an ideal \(I\) of \(R\) and \(q\) a power of \(p\), define \(I^{[q]}=\langle a^q\ |\ a\in I\rangle\). Take \(I\) an \(\mathfrak{m}\)-primary ideal and \(M\) a finite \(R\)-module and define the Hilbert-Kunz function of \(M\) with respect to \(I\) as \[HKF(I,M)(q)=\mathrm{length}(M/I^{[q]}M).\] The classical Hilbert-Kunz function was defined for \(I=\mathfrak{m}\) and was introduced by \textit{E. Kunz} [Am. J. Math. 91, 772--784 (1969; Zbl 0188.33702)] as a measure of the regularity of the ring: for example, Kunz proved that \(R\) is regular (this is, the number of generators of \(\mathfrak{m}\) is equal to \(\dim R\)) if and only if \(HKF(\mathfrak{m},R)(q)=q^{\dim R}\). \textit{P. Monsky} [Math. Ann. 263, 43--49 (1983; Zbl 0509.13023)] introduced the Hilbert-Kunz multiplicity as the limit \[e_{HK}(I,M):=\lim_{q\rightarrow\infty}\frac{HKF(I,M)(q)}{q^{\dim R}}.\] Then, \(e_{HK}(\mathfrak{m},R)=1\) if and only if \(R\) is regular. Despite the relation of the Hilbert-Kunz function with other Hilbert functions, its behaviour is far from well-understood: in fact, there is no general method to compute it and the attempt to do it has led or has implemented several \(p\)-methods for study singularities in local rings. Just to mention one of the several issues that this function has, the multiplicity (longly believed to be a rational) can be an irrational number and it is still not known the relation between those cases where \(e_{HK}(I,M)\) is rational (for example, regular local rings, complete local domains of dimension 1, normal affine semigroup rings, among others). For a brief history and methods involved in Hilbert-Kunz theory, see for example [\textit{C. Huneke}, in: Commutative algebra. Expository papers dedicated to David Eisenbud on the occasion of his 65th birthday. New York, NY: Springer. 485--525 (2013; Zbl 1275.13012)] or [\textit{C-Y. Jean Chan}, ``The shape of Hilbert-Kunz functions'', Preprint, \url{arXiv:2106.14053}] In this context, the authors offer a combinatorial setting to compute Hilbert-Kunz function and multiplicity of a family of algebras called binoid algebras. A binoid \((N,+,0,\infty)\) is a monoid \((N,+,0)\) with an absorbent element \(\infty\in N\) such that \(a+\infty=\infty\) for all \(a\in N\). We say that the binoid is integral if \(N\setminus\{\infty\}\) is a monoid. Despite \(\mathbb{N}_0\cup\{\infty\}\) being the simplest example of a binoid, many concepts from ring theory can be easily translate to a binoid by considering the ring with identity \((R,+,\cdot,0,1)\) as the binoid \((R,\cdot,1,0)\). In this way, for a binoid \(N\) we can define an ideal \(I\) as a subset of \(N\) such that for any \(a\in I\) and for any \(b\in N\), \(a+b\in I\). An ideal is prime if for any \(f+g\in I\) we have \(f\in I\) or \(g\in I\). The radical of \(I\) is the ideal \(\langle f\in N\ |\ f^n\in I\text{ for some } n\rangle\) and \(I\) is \(N_+\)-primary if its radical is \(N_+\) (the ideal of non-units of \(N\)). The (combinatorial) dimension of \(N\), denoted \(\dim N\), is the supremum of the length of a chain of prime ideals of \(N\). Analogous to modules, for binoids we have \(N\)-sets. Given a binoid \(N\), an \(N\)-set is a set \(S\) with a distinguished element \(p\in S\) and an operation \(+:N\times S\rightarrow S\) such that \begin{itemize} \item For all \(n,m\in N\) and \(s\in S\), \((n+m)+s=n+(m+s)\). \item For all \(s\in S\), \(0+s=s\). \item For all \(s\in S\), \(\infty+s=p\). \item For all \(n\in N\), \(a+p=p\). \end{itemize} Finally, the authors make a connection between \(N\)-set and algebras. Given a commutative ring \(K\) and a binoid \(N\), the binoid algebra associated to \(N\) is the quotient algebra \[K[N]:=K[X^n\ | \ n\in N]/\langle X^\infty\rangle.\] Therefore, two monomials \(aX^n\) and \(bX^m\) has as product \(abX^{n+m}\) if \(n+m\neq\infty\) and \(0\) otherwise. With all these, the authors define the Hilbert-Kunz function for binoids in the following way. Let \(N\) be finitely generated, semipositive (this is, the number of units of \(N\) is finite) binoid, \(T\) a finitely generated \(N\)-set and \(\mathfrak{n}\) an \(N_+\)-primary ideal of \(N\). Then the Hilbert-Kunz function of \(\mathfrak{n}\) on \(T\) at \(q\) is \[HKF^N(\mathfrak{n},T,q)=|T/([q]\mathfrak{n}+T)|-1,\] \noindent where for an ideal \(I\), \([q]I=\langle qf\ |\ f\in I\rangle\) for \(q>0\). The Hilbert-Kunz multiplicity of \(\mathfrak{n}\) on \(T\) is defined by \[e_{HK}(\mathfrak{n},T):=\lim_{q\rightarrow\infty}\frac{HKF^N(\mathfrak{n},T,q)}{q^{\dim N}}.\] These two functions are well-behaved with operations of binoids. More interesting, the computation of these functions can be reduced to one case (namely, to integral cancellative torsion-free binoids). Studying this case, the authors are able to prove their main theorem, that says that for a broad family of binoids \(N\) and for an \(N_+\)-primary ideal \(\mathfrak{n}\), \(e_{HK}(\mathfrak{n},N)\) always exists and it is a rational number. \noindent Theorem 7.6 Let \(N\) be a finitely generated, semipositive, cancallative, reduced binoid and \(\mathfrak{n}\) be an \(N_+\)-primary ideal of \(N\). Then \(e_{HK}(\mathfrak{n},N)\) exists and is rational. Finally, in order to relate the Hilbert-Kunz function for binoids and the Hilbert-Kunz function for rings, given that \(K[N]\) is not necessarily local, the authors give the definition of a Hilbert-Kunz function for semilocal (since we are in a commutative context, this means that the ring has finitely maximal ideals) Noetherian rings. The definition is the same as before but taking \(I\) as an \(\mathfrak{m}\)-primary ideal. Through this, it can be defined a proper Hilbert-Kunz function for binoid algebras \(K[N]\) of \(K[S]\) where \(S\) is an \(N\)-set with respect to \(K[\mathfrak{n}]\), where \(\mathfrak{n}\) is an \(N_+\)-primary ideal of \(N\), denoted by \(HKF^{K[N]}(K[\mathfrak{n}],K[S],q)\). This definition is purely algebraic, however we have that when \(N\) is a finitely generated, semipositive binoid and \(K\) is a field, \[HKF^{K[N]}(K[\mathfrak{n}],K[S],q)=HKF^N(\mathfrak{n},S,q)\] and the same happens with the respective multiplicities. This implies that the multiplicity for these binoid algebras always exists, it is independent of the characteristic of \(K\) and if \(N\) is one of the binoids used in Theorem 7.6, then the multiplicity is a rational number.