Arc spaces and the Rogers-Ramanujan identities (Q1937982)

Let \(f\) be a polynomial in \(n\) variables \(x_1\),\dots, \(x_n\) over a field \(k\) with \(f(0)=0\) and let \(k[[t]]\) be the ring of formal power series in one variable over \(k\). The arc space of the germ of algebraic variety \((X,0)\) defined by \(f\) (i.e. \(X=\{(x_1,\dots,x_n)\in k^n\;/\;f(x_1,\dots,x_n)=0\}\)) is \[ X_{\infty}:=\{(x_1(t),\dots,x_n(t))\in k[[t]]^n\;/ \;f(x_1(t),\dots,x_n(t))=0 \;\text{ and } x_i(0)=0, 1\leq i\leq n\}. \] By expanding \(f(x(t))\) as a power series in \(t\) gives \[ f(x(t))=F_1t+F_2t^2+F_3t^3+\cdots \] where the \(F_i\) are polynomials in the coefficients of \(t\) in \(x(t)\). Thus \(x(t)\in X_{\infty}\) if and only if the set of its coefficients is a solution of the equations \[ F_1=F_2=F_3=\cdots=0. \] Therefore \(X_{\infty}\) is a provariety, i.e., an algebraic variety defined by an infinite number of equations depending on an infinite number of variables. The coordinate algebra of \(X_{\infty}\) is the ring \[ J_{\infty}(X):=k[x_j^{(i)}; 1\leq j\leq n, i\in\mathbb{N}_{>0}]/(F_1,F_2,\cdots). \] This algebra has a natural grading given by assigning the weight \(i\) to the variables \(x_1^{i}\),\dots, \(x_n^{i}\) that makes the polynomials \(F_l\) homogeneous of weight \(l\). The authors of the paper under review introduce a new object, the Hilbert-Poincaré series of \((X,0)\), which is defined by \[ HP_{J_{\infty}(X)}(t)=\sum_{j=0}^{\infty}\dim_k\left(J_{\infty}(X)\right)_j.t^j, \] where \(\left(J_{\infty}(X)\right)_j\) denotes the \(j\)-th homogeneous component of \(J_{\infty}(X)\). This generating series is quite complicated to compute, even in basic cases. For instance, if \(X=\mathbb{A}^1\) is the affine line, then \[ HP_{J_{\infty}(\mathbb{A}^1)}(t)=\prod_{i\geq 1}\frac{1}{1-t^i}. \] In general the Hilbert-Poincaré series of \((X,0)\) coincides with the Hilbert-Poincaré series of the algebra \[ k[x_j^{(i)}; 1\leq j\leq n, i\in\mathbb{N}_{>0}]/L(I) \] where \(L(I)\) is the leading ideal of \(I=(F_1,F_2,\cdots)\) (with respect to a well chosen monomial ordering). The leading ideal is simpler than \(I\) since it is a monomial ideal. Thus for computing Hilbert-Poincaré series one need to compute \(L(I)\) and the Hilbert-Poincaré series of \(L(I)\). Computing \(L(I)\) is not easy in general since it is not a finitely generated ideal. Computing the Hilbert-Poincaré series of \(L(I)\) corresponds to counting partitions, leaving out those coming from weights of monomials in \(L(I)\). But even in basic cases, counting the corresponding partitions is not easy. For instance, in the case \(n=1\) and \(f=x_1^2\), the partitions appearing in this computation are the partitions without repeated or consecutive terms. One of the mains results obtained by the authors is the following (proved by using the Rogers-Ramanujan identity): \[ HP_{J_{\infty}(x_1^2=0)}=\prod_{ i\geq 1,\;i\equiv 1,4 \text{ mod }5}\frac{1}{1-t^i}. \] The rest of the paper is devoted to compute the Hilbert-Poincaré series of some particular germs of algebraic varieties.