--- title: ADE Classification of Hypersurface Singularities url: https://www.emergentmind.com/topics/ade-type-classification-of-hypersurfaces type: topic --- # ADE Classification of Hypersurface Singularities Searching arXiv for the cited papers to ground the article in current arXiv records. ADE-type classification of hypersurfaces refers to classification schemes in which hypersurface singularities are organized by the simple series \(A\), \(D\), and \(E\), and, in some recent formulations, by module-theoretic or arithmetic invariants that recover the same list. In the materials considered here, ADE language appears in several closely related settings: analytic normal forms for isolated singularities, hypersurfaces over complete regular local rings classified by Cohen–Macaulay type, projective plane curves whose simple singularities force the gradient ideal to be of linear type, surface rings described by invariant theory and matrix factorizations, and Lie-theoretic constructions in which the exceptional curves of a minimal resolution form an ADE Dynkin diagram [2509.15396]. ## 1. Local normal forms and the meaning of ADE for hypersurfaces For reduced projective plane curves, “simple singularities” means ADE singularities. The local analytic models used are \[ A_k:\ y^2-x^{k+1}=0, \] \[ D_{k+3}:\ y^2x-x^{k+2}=0\qquad (k\ge 1), \] \[ E_6:\ x^3+y^4=0, \] \[ E_7:\ y^3-xy^3 \quad \text{(as stated in the paper’s normal form)}, \] \[ E_8:\ x^3-y^5=0. \] The same source explicitly notes that “the standard ADE classification is being invoked” and that the paper uses analytic normal forms adapted to local computations [1901.03833]. For projective hypersurfaces in \(\mathbf P^3\), the local hypersurface equation is written in three variables after formal analytic change of coordinates. The ADE normal forms are \[ A_n:\ w^2+x^2+y^{n+1}=0, \] \[ D_n:\ w^2+y(x^2+y^{n-2})=0, \] \[ E_6:\ w^2+x^3+y^4=0, \] \[ E_7:\ w^2+x(x^2+y^3)=0, \] \[ E_8:\ w^2+x^3+y^5=0. \] These are treated as isolated singularities on projective hypersurfaces in \(\mathbf P^3\) and are used to construct differential operators adapted to the local Jacobian ideal [2111.10740]. For surface rings of ADE type, the hypersurfaces are presented as explicit quotient rings \[ A_n := k[X,Y,Z]/(X^{n+1}-YZ), \] \[ D_n := k[X,Y,Z]/(X^2+Y^{n-1}+YZ^2)\qquad (n\ge 4), \] \[ E_6 := k[X,Y,Z]/(X^2+Y^3+Z^4),\qquad E_7 := k[X,Y,Z]/(X^2+Y^3+YZ^3),\qquad E_8 := k[X,Y,Z]/(X^2+Y^3+Z^5). \] These are described as the isolated rational double points or simple surface singularities in the usual ADE sense [1604.08435]. Taken together, these formulations show that ADE classification is not tied to a single presentation of the equation. The series \(A\), \(D\), and \(E\) persists across plane-curve singularities, projective hypersurfaces in \(\mathbf P^3\), and two-dimensional hypersurface rings, while the normal form is adjusted to the ambient category and the local computational problem. ## 2. Cohen–Macaulay-type classification over complete regular local rings A recent local-ring formulation classifies hypersurface singularities by the size of the category of maximal indecomposable Cohen–Macaulay modules. For a local ring \((R,\mathfrak m,\kappa)\), \[ \mathcal{MCM}(R)=\{[M] : M \text{ is a maximal indecomposable Cohen–Macaulay } R\text{-module}\}, \] and if \(|\mathcal{MCM}(R)|=\lambda\), the ring is said to have \(\lambda\)-Cohen–Macaulay type. The new notion is **sparse Cohen–Macaulay type**: \[ |\mathcal{MCM}(R)|<|\kappa|. \] The same work proves that for a finite-dimensional Noetherian local ring with infinite residue field, \[ |\mathcal{MCM}(R)| \le |R| \le 2^{|\kappa|}, \] so non-sparse behavior lies between \(|\kappa|\) and \(2^{|\kappa|}\) [2509.15396]. The structural input is a sparse version of the Huneke–Leuschke–Takahashi theorem: if \((R,\mathfrak m,\kappa)\) is a Cohen–Macaulay local ring with sparse Cohen–Macaulay type, then \[ \dim(R/\mathfrak p)\le 1 \quad \text{for every } \mathfrak p\in \operatorname{Sing}(R). \] Equivalently, the singular locus has dimension at most one. The proof uses a cardinal prime avoidance lemma together with specialization-closedness of \(\operatorname{Sing}(R)\) [2509.15396]. The hypersurface case is treated for rings of the form \(R/\langle f\rangle\) with \(R\) regular local and \(f\in\mathfrak m\). A key reduction is the Buchweitz–Greuel–Schreyer surjection \[ \mathcal{MCM}\left(\frac{R}{\langle f\rangle}\right)\twoheadrightarrow \mathcal{IS}(R,f):=\{I\subset R : f\in I^2\}. \] Hence, if there are \(|\kappa|\) many ideals \(I\) with \(f\in I^2\), then \(R/\langle f\rangle\) cannot have sparse CM type. To organize normal-form reduction over arbitrary complete regular local rings, the paper introduces a relation \(f\looparrowright g\), meaning “\(f\) can be written as \(g\),” as a substitute for classical contact equivalence in mixed characteristic [2509.15396]. Sparse CM type imposes strong order constraints. If \(f\in\mathfrak m\) has sparse CM type, then: 1. \(\operatorname{ord}(f)\le 3\). 2. If \(\dim R>1\) and \(\kappa\) is algebraically closed, then \(\operatorname{ord}(f)\le 2\). 3. For every \(g\in\mathfrak m\), \(f\notin \langle g^3\rangle\). 4. For every \(\alpha,\beta\in\mathfrak m\), \(f\notin \langle \alpha,\beta^2\rangle^3\). For elements of order \(2\), the splitting lemma states that after choosing generators \(a_1,\dots,a_n\) of \(\mathfrak m\), \[ f=u_1a_1^2+\cdots+u_ra_r^2 + g, \qquad g\in\mathfrak m^3, \] for some units \(u_i\). The integer \(r\) is the **rank** \(\rk(f)\). If \(\dim R>2\) and \(f\) has sparse CM type over an algebraically closed \(\kappa\), then \[ \rk(f)\ge \dim(R)-2. \] This reduces the classification to full rank, rank \(n-1\), and rank \(n-2\) [2509.15396]. From here the classification reproduces the \(A\)-, \(D\)-, and \(E\)-series. If \(f\) has order \(2\) and \(\rk(f)=n\), then \[ f\looparrowright A_1 = x_1^2+\cdots+x_n^2. \] If \(f\) has order \(2\) and \(\rk(f)=n-1\), then either \[ f\looparrowright A_k \quad \text{for some } k\in\{2,3,\dots,\infty\}, \] or, after completion, \(f\looparrowright A_\infty\), where \[ A_k: x_1^{k+1}+x_2^2+\cdots+x_n^2 \qquad (k\ge1), \] \[ A_\infty: x_1^2+x_2^2+\cdots+x_{n-1}^2. \] In the remaining rank \(n-2\) case, the analysis reduces to the 3-jet in dimension \(2\). For order \(3\), the cubic leading term is equivalent to one of \[ xy(x+y), \qquad x^2y, \qquad x^3. \] The first two produce the \(D\)-series: \[ D_k: x_1(x_2^2+x_1^{k-2}) + x_3^2+\cdots+x_n^2 \qquad (k\ge4), \] and \[ D_\infty: x_1x_2^2+x_3^2+\cdots+x_{n-1}^2. \] When the 3-jet is \(x^3\), the classification becomes characteristic-dependent and yields \[ E_6: x_1^3+x_2^4+x_3^2+\cdots+x_n^2, \] \[ E_7: x_1(x_1^2+x_2^3)+x_3^2+\cdots+x_n^2, \] \[ E_8: x_1^3+x_2^5+x_3^2+\cdots+x_n^2, \] together with the extra forms \(E_6^1\), \(E_7^1\), \(E_8^1\), \(E_8^2\) in characteristics \(3\) and \(5\) [2509.15396]. The principal theorem states that if \((R,\mathfrak m,\kappa)\) is a complete regular local ring of dimension \(n\ge1\), with \(\kappa\) infinite and \(\operatorname{char}(\kappa)\ne 2\), and if \(f\in\mathfrak m^2\) has sparse Cohen–Macaulay type, then \(f\) can be written as one of the ADE hypersurface singularities \(A_k\), \(D_k\), \(E_6\), \(E_7\), \(E_8\), with the additional forms \(E_6^1\), \(E_7^1\), \(E_8^1\), \(E_8^2\) in small positive characteristics. If \(\kappa\) is uncountable, the theorem also allows \(A_\infty\) and \(D_\infty\) [2509.15396]. ## 3. Jacobian and gradient linear type for projective hypersurfaces with ADE singularities A different ADE-type rigidity phenomenon concerns Jacobian and gradient ideals. For a reduced polynomial \(f\), the gradient ideal is \[ J(f) = \left(\frac{\partial f}{\partial x_0},\dots,\frac{\partial f}{\partial x_n}\right), \] and the affine Jacobian ideal is \[ I(f) = (f, J(f)). \] An ideal \(I\) is **of linear type** if the natural surjection \[ \operatorname{Sym}_A(I)\to \mathcal R_A(I) \] is an isomorphism. In the affine case this gives **Jacobian linear type** for \(I(f)\); in the projective case it gives **gradient linear type** for \(J(f)\) [1901.03833]. For a reduced affine hypersurface \(X=V(f)\subset \mathbb A^n\) with only isolated singularities, the main criterion is an equivalence among local singularity-theoretic conditions. Writing \[ M(f)=R/J(f), \qquad T(f)=R/I(f), \] and denoting by \(\mu_m(f)\) and \(\tau_m(f)\) the localized Milnor and Tjurina numbers, the hypersurface is **locally Eulerian** if \[ f\in J(f)_m \quad \text{for every singular point } m, \] equivalently if \[ \mu_m(f)=\tau_m(f)\quad \text{for all singular points }m. \] Under the isolated singularity hypothesis, the following are equivalent: \(X\) is of Jacobian linear type; \(I(f)\) is locally a complete intersection at each singular point; \(X\) is locally Eulerian; the Tjurina algebra \(T(f)\) is Artinian Gorenstein; and the module \[ \operatorname{Hom}_{M(f)}(T(f),M(f))\cong (J(f):f)/J(f) \] is cyclic [1901.03833]. For projective hypersurfaces \(X=V(f)\subset \mathbb P^n\) with only isolated singularities, the analogous criterion states that \(X\) is of gradient linear type if and only if the gradient ideal \(J(f)\) is locally a complete intersection at each singular prime. If \[ R^s \xrightarrow{\varphi} R^{n+1} \to J(f)\to 0 \] is a minimal presentation, this is also equivalent to the condition that the ideal generated by the entries of \(\varphi\) has codimension \(n+1\) [1901.03833]. The ADE content is concentrated in the theorem that **any reduced projective plane curve with simple singularities is of gradient linear type**. The proof is explicit and local: at each singular point one passes to an affine chart, writes the local equation \(F(x,y)\), and shows \[ F \in J(F)_q. \] This establishes local Eulerianity for each ADE type and hence gradient linear type. The same paper also proves that **any reduced singular quartic plane curve is of gradient linear type**, using the genus formula and an exhaustive analysis of singularity configurations [1901.03833]. A common misconception is that ADE labeling is purely taxonomic. In this setting the ADE list has direct blowup-algebra consequences: simple singularities force the Jacobian or gradient ideal to satisfy the linear-type property. ## 4. Arithmetic and algorithmic consequences in projective dimension three For hypersurfaces in \(\mathbf P^3\) over finite fields, ADE classification becomes a computational tool in the study of zeta functions. For \(X/\mathbf F_q\), \[ Z(X,t)=\exp\!\left(\sum_{r\ge 1}\frac{N_r t^r}{r}\right), \qquad N_r = |X(\mathbf F_{q^r})|. \] For projective hypersurfaces in \(\mathbf P^3\), the cohomological expression used is \[ Z(X,t) = (1-t)(1-qt)(1-q^2t)\,P(t), \] where \[ P(t)=\det\!\left(1-t\,q^3\,\mathrm{Frob}_{q}^{-1}\mid H^2_{\mathrm{rig}}(U)\right), \qquad U=\mathbf P^3\setminus X. \] The computational problem is therefore the Frobenius action on rigid or de Rham cohomology of the complement [2111.10740]. The setup requires an equisingular lift \[ f\in \mathbf F_p[w,x,y,z] \quad\leadsto\quad \widetilde f\in \mathbf Z_p[w,x,y,z], \] with the condition that \[ \mathbf Z_p[w,x,y,z]/(\widetilde f_w,\widetilde f_x,\widetilde f_y,\widetilde f_z) \] has no \(p\)-torsion. The paper gives the example of \(A_4\) at \(p=5\), where one derivative becomes \(5t^4\), vanishes mod \(5\), and causes torsion problems [2111.10740]. The algorithm is based on the Koszul/de Rham complex with modified differential \[ d_f(w)=f\,dw - df\wedge w. \] If \(p(X)\) is the global Milnor number, then in degrees \[ m\ge 3(N-2) \] the relevant Koszul cohomology has dimension \(p(X)\); this is the **stable range**. For weighted homogeneous isolated singularities, the spectral sequence degenerates at \(E_2\) [2111.10740]. The main conceptual result is Theorem 3.5: for each ADE type, in the stable range, \[ h \in J(f) \quad\Longleftrightarrow\quad D_i(h)=0 \text{ for all ADE operators } D_i. \] For \(A_n\), the annihilating operators are evaluation at the origin together with \[ \partial_t|_{0},\ \partial_t^2|_{0},\ \dots,\ \partial_t^{n-1}|_{0}. \] For \(D_n\), \(E_6\), \(E_7\), and \(E_8\), the paper gives analogous families built from evaluation and higher-order derivatives adapted to the normal forms. Proposition 3.6 shows that under a formal analytic change of coordinates, these differential operators transform into linear combinations of operators of the same or lower order, making the criterion coordinate-independent in practice [2111.10740]. Theorem 3.10 states that for projective hypersurfaces in \(\mathbf P^3\) with only ADE singularities, the subdiagonal on the \(E_2\)-page vanishes. The computational workflow is then: compute a basis on the \(E_2\)-page, propagate basis elements by Theorem 3.7, compute ADE annihilating operators, apply inverse Frobenius, reduce cohomology using the operator criterion in place of Gröbner membership tests, and form the Frobenius matrix and its characteristic polynomial. The paper emphasizes that the ADE method extends the ordinary-double-point method of Stetson and Baranovsky from the \(A_1\) case to the full ADE family [2111.10740]. This suggests that ADE classification is not only a local analytic description of singularities but also a mechanism for explicit arithmetic computation. ## 5. Surface rings, invariant theory, and Lie-theoretic geometry In two dimensions, ADE hypersurfaces admit a representation-theoretic and invariant-theoretic description. The surface rings \[ A_n,\ D_n,\ E_6,\ E_7,\ E_8 \] are treated both as explicit hypersurfaces and as invariant rings \[ R \cong k[x,y]^G \subset k[x,y], \qquad G\subset \mathrm{SL}_2(k), \] when the group order is invertible in \(k\). The corresponding finite groups are: cyclic of order \(n+1\) for \(A_n\), binary dihedral of order \(4n-8\) for \(D_n\), binary tetrahedral of order \(24\) for \(E_6\), binary octahedral of order \(48\) for \(E_7\), and binary icosahedral of order \(120\) for \(E_8\) [1604.08435]. These rings have finite Cohen–Macaulay type, and the classification of indecomposable maximal Cohen–Macaulay modules is made explicit via matrix factorizations \[ \phi\psi=\psi\phi=f\cdot \mathrm{Id}. \] The strategy is to classify the indecomposable MCM modules via known lists of reduced indecomposable matrix factorizations and then identify each module as a first syzygy module \[ \operatorname{coker}(\phi)\cong \operatorname{Syz}_R(F_1,\dots,F_{m+1}). \] For \(A_n\), for example, the indecomposable non-free MCM modules are represented by \[ M_m = \operatorname{Syz}_R(X^m,Z), \qquad m=1,\dots,n, \] with \[ M_m^\vee \cong M_{n+1-m}. \] The same source gives complete explicit module lists for \(D_n\), \(E_6\), \(E_7\), and \(E_8\), and uses them to compute Hilbert–Kunz functions [1604.08435]. The Hilbert–Kunz multiplicity of an ADE surface ring satisfies \[ \HKM(R)=2-\frac{1}{|G|}, \] assuming the group order is invertible in \(k\) and the ring is \(F\)-rational. Explicit formulas are also given for the Hilbert–Kunz functions of \(D_{n+2}\), \(E_6\), \(E_7\), and \(E_8\) as functions of \(q=p^e\) [1604.08435]. The Lie-theoretic side begins with a singular surface \[ X \cong \mathbb C^2/\Gamma, \] where \(\Gamma\subset SL(2,\mathbb C)\) is finite. If \(Y\to X\) is the minimal resolution, the exceptional locus is a union of smooth rational curves \(C_i\) with \[ C_i^2=-2, \] and the dual graph is an ADE Dynkin diagram. Inside \[ \mathbb I=\left\{\sum_i a_i[C_i]\mid a_i\in\mathbb Z\right\}, \] the set \[ \Phi:=\{a\in \mathbb I\mid a^2=-2\} \] is a simply-laced root system of a simple Lie algebra \(\mathfrak g\). The classes \([C_i]\) form a base of \(\Phi\) [1811.02777]. The Brieskorn–Slodowy–Grothendieck diagram connects the ADE singular surface, the nilpotent cone, the flag variety \(G/B\), and the Springer resolution \[ G\times^B \mathfrak n \cong T^*(G/B). \] On \(G/B\) and on \(T^*(G/B)\), the paper constructs holomorphic \(\mathfrak g\)-bundles with bracket-preserving holomorphic structure \[ \bar\partial_{\mathfrak g} = \bar\partial_0 + \sum_{\alpha\in\Phi^-} \operatorname{ad}(\varphi_\alpha), \] where \(\varphi_\alpha\) is a \((0,1)\)-form with values in the negative-root line bundle \(L_\alpha\). On the minimal resolution \(\widetilde S\), the line bundles corresponding to roots satisfy \[ L_a|_{\widetilde S}\cong \mathcal O_{\widetilde S}\Bigl(-\sum_i n_i C_i\Bigr) \] for \[ a=\sum_i n_i\alpha_i. \] Thus the ADE root data, the exceptional curves, the flag variety, and the cotangent bundle are different manifestations of the same classification structure [1811.02777]. ## 6. Relation to weighted-homogeneous classifications and limits of ADE language ADE-type classification is not identical with every classification of hypersurfaces. One related but distinct framework classifies two-dimensional graded normal hypersurfaces \[ R=k[x,y,z]/(f) \] by their weighted type \[ (a,b,c;h)=\deg(x,y,z;f), \] with \(a\)-invariant \[ a(R)=h-(a+b+c). \] For fixed \(a(R)>0\), the number of possible types is finite. The explicit classification is carried out for \[ 0