--- title: Fixed-Generators Hilbert Schemes url: https://www.emergentmind.com/topics/fixed-generators-hilbert-schemes type: topic --- # Fixed-Generators Hilbert Schemes Searching arXiv for the primary paper and closely related work on fixed-generator loci in Hilbert schemes. arXiv search: "Motivic classes of fixed-generators Hilbert schemes of unibranch curve singularities and Igusa zeta functions" Fixed-generators Hilbert schemes are loci inside Hilbert schemes in which the defining ideals are constrained to have a prescribed minimal number of generators. In the most developed form of the subject, this means the closed subschemes \[ Hilb_{0,k}^n(C)= \{ I \in Hilb^n_0(C) \, | \, \dim_\mathbb{K} I/m_0I = k \}, \] for an irreducible plane curve singularity \((C,0)\), where \(Hilb^n_0(C)\) is the punctual Hilbert scheme of colength-\(n\) ideals in the completed local ring \(\widehat{\mathcal O}_{C,0}\), and \(k\) is the exact number of minimal generators by Nakayama’s lemma. Recent work has made these loci accessible through semigroup combinatorics, syzygies, motivic classes, and, in the one-generator case, motivic integration and Igusa zeta functions [2507.17642]. Related developments treat exact-generator loci on punctual Hilbert schemes of singular curves, Brill–Noether loci on Hilbert schemes of points on smooth surfaces, and monogenic-generator fibers of Hilbert-type moduli spaces [2509.06761]. ## 1. Definition and local algebraic framework Let \((C,0)\) be an irreducible plane curve singularity over an algebraically closed field \(\mathbb K\) of characteristic \(0\), with completed local ring \[ \widehat{\mathcal O}_{C,0}\cong \mathbb K[[x,y]]/(f)\cong \mathbb K[[p_C,q_C]]\subseteq \mathbb K[[t]]. \] Its semigroup is \[ \Gamma_C=\{val_t(g)\mid g\in \widehat{\mathcal O}_{C,0}\}, \] with hole set \[ H=\mathbb N\setminus \Gamma, \] and \(\delta=|H|\). For a semigroup ideal \(\Delta\subseteq \Gamma\), the conductor is \[ c(\Delta)=\min\{j\in \Delta\mid j+\mathbb N\subseteq \Delta\}. \] The punctual Hilbert scheme at the singular point is \[ Hilb^n_0(C)=\{ I \subseteq \widehat{\mathcal O}_{C,0} \text{ ideal} \, | \dim_\mathbb K \widehat{\mathcal O}_{C,0} / I = n\}, \] and the fixed-generators locus is the closed subscheme \[ Hilb_{0, k}^n(C)= \{ I \in Hilb^n_0(C) \, | \, \dim_\mathbb{K} I/m_0I = k \}. \] For \(k=1\), this is the principal Hilbert scheme, or one-generator locus [2507.17642]. The key local invariant is the valuation set \(\Delta=val_t(I)\subseteq \Gamma\), which is a semigroup ideal. If \(g_1,\dots,g_k\) is a minimal generating set, one writes \(n_i=val_t(g_i)\). Since different minimal generating sets can have different leading valuations because of cancellation, the relevant notion is not an arbitrary valuation tuple but an independent one. For \[ \Delta_{g_i}=val_t(g_1,\dots,g_{i-1},g_{i+1},\dots,g_k), \] the values \((n_1,\dots,n_k)\) are independent if \[ n_i\not\in \Delta_{g_i} \] for all \(i\). The foundational theorem states that every \(I\in Hilb_{0,k}^n(C)\) admits a minimal generating set whose valuations form an independent \(k\)-tuple, and that this independent tuple is unique [2507.17642]. This uniqueness turns the number-of-generators condition into a rigid combinatorial datum. It also isolates the fixed-generators Hilbert scheme as a refinement of the punctual Hilbert scheme by exact generator number, rather than merely by colength. ## 2. Stratification by valuations, semigroup ideals, and syzygies The uniqueness of the independent valuation tuple yields a first decomposition. For an independent tuple \(\underline n=(n_1,\dots,n_k)\), \[ Hilb_{0,k}^{n,\underline n}(C)= \{ I \in Hilb_{0,k}^{n}(C) \, | \, I = (g_1, \dots , g_k) , val_t(g_i) = n_i\}, \] and \[ Hilb_{0,k}^{n}(C)=\coprod_{\underline n} Hilb_{0,k}^{n,\underline n}(C). \] However, the tuple \(\underline n\) does not determine the valuation semigroup \(val_t(I)\) in general, since one only has \[ \coprod_i \{ n_i + \Gamma \} \subseteq \Delta, \] possibly strictly. One therefore refines further to \[ Hilb_{0,k}^{n,\underline n,\Delta}(C) = \{ I \in Hilb_{0,k}^{n,\underline n}(C) \, | \, val_t(I)=\Delta \}, \] with \[ Hilb_{0,k}^{n}(C)=\coprod_{\underline n,\Delta} Hilb_{0,k}^{n,\underline n,\Delta}(C). \] The one-generator case is completely explicit. If \(I=(f_n)\) is principal with valuation \(n\), then \[ f_n = \tau_n + \sum_{h \in H \, | \, n+h \in \Gamma} \lambda_h^n \tau_{n+h}, \] so \[ Hilb_{0,1}^n(C)= \{ (f_n) \, | \, \lambda_h^n \in \mathbb K \} \] is isomorphic to affine space of dimension \(|\{n+H\}\cap \Gamma|\). This makes the one-generator locus a building block for higher-generator loci [2507.17642]. A second structural statement relates \(k\)-generator strata to products of principal loci. For an independent tuple \(\underline n=(n_1,\dots,n_k)\), define \[ \Psi_{\underline{n}}: \prod_{i=1}^k Hilb_{0,1}^{n_i}(C) \to \coprod_nHilb_{0,k}^{n,\underline{n}}(C), \qquad ((f_1), \dots , (f_k)) \mapsto (f_1, \dots , f_k). \] Over a fixed valuation-semigroup stratum \(\Delta\), this is a trivial affine fibration: \[ \Psi_{\underline{n}}^{-1}(Hilb_{0,k}^{n,\underline{n},\Delta}(C)) \cong Hilb_{0,k}^{n,\underline{n},\Delta}(C) \times \mathbb{A}^{\sum_i |S_i|}, \] where \[ S_i = \{ n_i + H \} \cap \Delta. \] Consequently, \[ [Hilb_{0,k}^{n,\underline{n},\Delta}(C)] = [\Psi_{\underline{n}}^{-1}(Hilb_{0,k}^{n,\underline{n},\Delta}(C))]\mathbb{L}^{ - \sum_{i}|\{n_i+H \} \cap \Delta|}. \] The remaining input is the Oblomkov–Rasmussen–Shende/Piontkowski description of ideals with fixed valuation semigroup via generator and syzygy spaces. For a semigroup ideal \(\Delta\subseteq \Gamma\), with syzygy parameter space \(Syz\), one stratifies by rank: \[ S_a = \{ p \in Syz \, | \, rk(M_\Delta (p))=a \}. \] The fixed-generators geometry is then encoded by affine parameter counts for generators, overlap corrections coming from \(\Delta\)-normal forms, and determinantal rank strata coming from syzygies [2507.17642]. ## 3. Motivic classes and the \((p,q)\)-curve simplification The motivic classes live in the Grothendieck ring \(K_0(\mathrm{Var}_\mathbb K)\), with \(\mathbb L=[\mathbb A^1]\). For arbitrary unibranch plane branches, the main formula expresses the class of the fixed-\(k\)-generator Hilbert scheme as \[ [Hilb_{0,k}^{n}(C)] = \sum_{\Delta} \mathbb L^{\sum_i |\Gamma_{> \nu_i} \setminus \Delta |} \sum_{\underline{n}} \mathbb L^{- \sum_i |\{n_i+H \} \cap \Delta|} \sum_{a = 1}^{m_\Delta} [S_a]\mathbb L^{ - a } . \] Here \(\Delta\) runs over semigroup ideals, \(\nu_i\) are generators of \(\Delta\) as semigroup ideal, \(m_\Delta\) is the number of syzygies in a free resolution of the monomial ideal \(I_\Delta\subseteq \mathbb K[\Gamma]\), and \(S_a\subseteq Syz\) is the rank-\(a\) stratum. The formula refines the ORS framework by organizing it according to the exact number of minimal generators [2507.17642]. For \((p,q)\)-curves, \[ (p_C,q_C)=(t^p,t^q), \qquad \gcd(p,q)=1, \] the situation becomes more rigid. If \(I\in Hilb_{0,k}^{\star,\underline n}(C)\), then \[ val_t(I) = \coprod_i \{ n_i + \Gamma \}. \] There are no hidden valuation generators arising from nontrivial cancellations, so the semigroup ideal is determined purely by the independent tuple \(\underline n\). Writing \(M_n\) for the set of independent tuples of colength \(n\), one obtains the explicit formula \[ [Hilb_{0,k}^{n}(C)] = \sum_{\underline{n} \in M_n} \left( \prod_i [Hilb_{0,1}^{n_i}(C)] \right) \mathbb L^{- \sum_i |\{ n_i + H \} \cap \Gamma_{\underline{n}}|}. \] Since each principal locus has class \[ [Hilb_{0,1}^{n_i}(C)] = \mathbb L^{|\{n_i+H\}\cap\Gamma|}, \] the whole class is a finite sum of monomials in \(\mathbb L\). Two consequences are emphasized: positivity, meaning \([Hilb_{0,k}^n(C)]\) is a polynomial in \(\mathbb L\) with positive coefficients, and topological invariance, meaning that for \((p,q)\)-curves this polynomial depends only on \(\Gamma_C\), hence only on the topological type [2507.17642]. The non-toric case remains subtler. For \[ \widehat{\mathcal O}_{C,0}= \mathbb K[[t^4, t^6+t^7]], \] with \[ \Gamma = \{ 0, 4, 6, 8, 10, 12, 13, 14, 16, \dots \}, \qquad H = \{1,2,3,5,7,9,11,15\}, \qquad \delta=8, \] the colength-\(4\) fixed-generator pieces are \[ [Hilb_{0,1}^4(C)]=\mathbb L^3,\qquad [Hilb^4_{0,2}(C)] = \mathbb L^3 + \mathbb L,\qquad [Hilb^4_{0,3}(C)] = \mathbb L^2 +1. \] This shows both the effectiveness of the independent-tuple stratification and the necessity, outside the \((p,q)\)-case, of excluding semigroup-ideal data that do not arise from actual ideals. ## 4. The one-generator locus, arc spaces, and motivic Igusa theory A distinct but connected part of the theory concerns a special open component of the principal Hilbert scheme. Let \[ X^n_C = ord_C^{-1}(n)\subseteq (\mathbb A^2)_\infty \] be the \(n\)-th contact locus of the curve. The paper constructs a morphism \[ \Pi: \pi_{n+1}(X^n_C) \to \operatorname{Im}(\Pi)\subseteq Hilb_{0,1}^n(C), \] sending a truncated arc to the scheme-theoretic intersection of its image with \(C\). Its image is the “unibranch” open component \[ UHilb^n_{0,1}(C):=\operatorname{Im}(\Pi)\subseteq Hilb_{0,1}^n(C), \] consisting of principal ideals that admit an irreducible representative upstairs in \(\mathbb K[[x,y]]\). This need not equal all of \(Hilb_{0,1}^n(C)\) for small \(n\); for the cusp, \(X^5=\emptyset\) while \(Hilb_{0,1}^5(C)\cong \mathbb P^1\) [2507.17642]. Passing to the quotient by smooth reparametrizations gives a branch-space morphism whose fibers are powers of \(\mathbb L\) times a truncation correction \(F^{n,k}\). The resulting formula is \[ [UHilb^n_{0,1}(C)] = \frac{\mathbb L}{\mathbb L-1}\sum_k \mathbb L^{k} \frac{ [\pi_{n+1}(X^{n,k}_C)]}{[F^{n,k}]}. \] This identifies the motivic class of the open one-generator component directly with pieces of the contact locus. The motivic Igusa zeta function \[ Z_C(s) = \sum_{n=1}^\infty \mu(X^n_C) \mathbb L^{-ns} \] can then be rewritten in terms of these branch/Hilbert-scheme images: \[ Z_C(s)= (\mathbb L-1) \mathbb L^{-2}\sum_{n \geq 1}\mathbb L^{-n(s+1)}\sum_k \mathbb L^{-k}[\operatorname{Im}(\Pi_b^k)] . \] The coefficients of the motivic zeta function are thus encoded by open principal Hilbert strata. The embedded-resolution formula of Denef–Loeser converts this into a resolution-theoretic description. If \(\rho:(Y,E)\to (\mathbb A^2,C)\) is an embedded resolution with exceptional divisors \(E_i\), multiplicities \(N_i\), discrepancies \(\nu_i\), and strata \(E_J^o\), the class \([UHilb^n_{0,1}(C)]\) can be written explicitly in terms of the \(E_J^o\), the \(N_i\), the \(\nu_i\), and the branch-space corrections \(F^{n,k}\). Since for plane curves each \(E_J^o\) is either a point or a \(\mathbb P^1\) minus finitely many points, this proves polynomiality in \(\mathbb L\) and topological invariance. The same analysis yields the nonemptiness threshold \[ UHilb^n_{0,1}(C)\neq \emptyset \qquad \text{for all } n\ge N, \] where \(N=\max_i N_i\), and motivates the conjecture that for \(n\gg 0\), \[ UHilb^n_{0,1}(C)=Hilb_{0,1}^n(C). \] ## 5. Other fixed-generator theories on curves and surfaces On singular curve germs, a parallel theory uses semimodule trees and Hilbert–Samuel strata. For a unibranch singularity \((C,O)\) with local ring \(A\subset \mathbf C[[t]]\), semigroup \(\Gamma\), and punctual Hilbert scheme \[ C^{[\ell]}:=\{I\subset A \mid \dim_k(A/I)=\ell\}, \] one has the semigroup stratification \[ C^{[\ell]}=\bigsqcup_{\Delta\in \mathscr D_\ell} C^{[\Delta]}. \] The fixed-generator loci are \[ C^{[\ell],m}=\{I\in C^{[\ell]}\mid m(I)=m\}, \qquad C^{[\Delta],m}=\{I\in C^{[\Delta]}\mid m(I)=m\}, \] together with the upper-bound loci \(C^{[\Delta],\le m}\). For plane branches \(\Gamma=\langle p,q\rangle\), the basic pieces controlling \(C^{[\Delta],\le m}\) are \[ Y_{\underline{i}_{\underline{j}}} \cong (\mathbf C^{*})^{n-m}\times \mathbb A^{N(\Delta)-n+m}, \] and intersections of such pieces have the same torus-times-affine form. A key consequence is that if \(m