---
title: Solomon-Terao Polynomial Explored
url: https://www.emergentmind.com/topics/solomon-terao-polynomial
type: topic
---

# Solomon-Terao Polynomial Explored

The Solomon-Terao polynomial is an arrangement-theoretic invariant constructed from Hilbert series of logarithmic modules. For a central hyperplane arrangement \(\mathcal{A}\subset V\) with coordinate ring \(S=\mathrm{Sym}^*(V^*)\), one forms generating functions from the graded logarithmic derivation modules \(D^p(\mathcal{A})\); in the contemporary literature, the term denotes either the bivariate polynomial \(\Psi(\mathcal{A};x,t)\) itself or a specialization such as \(ST(\mathcal{A};x)=\Psi(\mathcal{A};x,-1)\). Its central role is that it refines combinatorial information encoded by the characteristic polynomial, interacts with freeness and tameness, admits addition-deletion formalisms under explicit homological hypotheses, and extends beyond hyperplane arrangements to central equidimensional subspace arrangements through multi-logarithmic forms and multi-residues [2509.10047][1803.08672].

## 1. Definitions and terminological conventions

Let \(\mathcal{A}\) be a central arrangement in a vector space \(V=\mathbb{K}^\ell\), and let \(S=\mathbb{K}[x_1,\dots,x_\ell]\). For \(0\le p\le \ell\), the \(p\)-th logarithmic derivation module is
\[
D^p(\mathcal{A}) := \{ \theta \in \wedge^p\mathrm{Der}(S) \mid \theta(\alpha_H,f_2,\dots,f_p)\in S\alpha_H \ \forall H\in\mathcal{A},\ f_i\in S\},
\]
with \(D^1(\mathcal{A})=D(\mathcal{A})\). Using the Hilbert series \(\mathrm{Hilb}(D^p(\mathcal{A});x)\), one standard definition is
\[
\Psi(\mathcal{A};x,t):=\sum_{p=0}^{\ell}\mathrm{Hilb}(D^p(\mathcal{A});x)\,[t(x-1)-1]^p.
\]
A recurrent specialization is the Solomon-Terao polynomial
\[
ST(\mathcal{A};x):=\Psi(\mathcal{A};x,-1)=\sum_{p=0}^{\ell}\mathrm{Hilb}(D^p(\mathcal{A});x)\,(-x)^p\in\mathbb{Z}[x].
\]
The same framework extends to multiarrangements \((\mathcal{A},m)\) and to higher-order specializations \(ST_{d+1}(\mathcal{A},m;x)=\Psi(\mathcal{A},m;x,\frac{1-x^d}{x-1})\) [2509.10047][2305.10283].

The terminology is not completely uniform. In work emphasizing higher-order logarithmic modules and addition-deletion, \(\Psi(\mathcal{A};x,t)\) itself is called the Solomon-Terao polynomial. In work on the Solomon-Terao algebra, a different normalization of a bivariate polynomial \(\Psi(\mathcal{A};x,t)\) is used, namely
\[
\Psi(\mathcal{A}; x, t) := \sum_{p=0}^\ell t^{\ell-p}\cdot \mathrm{Hilb}(D^p(\mathcal{A}); x)\cdot (-1)^p.
\]
This notation shift does not alter the underlying theme: the invariant is assembled from graded logarithmic modules and is designed to interpolate between combinatorial, algebraic, and geometric data [1802.04056].

## 2. Relation with the characteristic polynomial

The characteristic polynomial of a hyperplane arrangement is defined from the intersection lattice \(L(\mathcal{A})\) and its Möbius function \(\mu\) by
\[
\chi(\mathcal{A};t)=\sum_{X\in L(\mathcal{A})}\mu(X)t^{\dim X}.
\]
Under the convention
\[
\Psi(\mathcal{A};x,t)=\sum_{p=0}^{\ell}\mathrm{Hilb}(D^p(\mathcal{A});x)\,[t(x-1)-1]^p,
\]
Solomon-Terao theory identifies the characteristic polynomial as the specialization at \(x=1\):
\[
(-1)^\ell \Psi(\mathcal{A};1,t)=\chi(\mathcal{A};t).
\]
This is the basic recovery statement: the bivariate Hilbert-series invariant degenerates to the classical combinatorial polynomial [2509.10047].

A geometric reformulation is available through the logarithmic ideal and the critical-point variety attached to the master function. For an arrangement of rank \(\ell\) with \(n=|\mathcal{A}|\), the variety \(X(\mathcal{A})\subset \mathbb{P}V\times \mathbb{P}^{n-1}\) satisfies
\[
[X(\mathcal{A})]=\chi(\mathcal{A};-h,k-h)\in A^\bullet(\mathbb{P}V\times \mathbb{P}^{n-1}),
\]
where \(\chi(\mathcal{A};s,t):=s^\ell\chi(\mathcal{A},t/s)\) is the homogenized characteristic polynomial. For tame arrangements, the logarithmic ideal also yields
\[
h(S/I;t,u)=\frac{P(\mathcal{A};t,-u)}{(1-u)^n},
\]
with \(P(\mathcal{A};x,y)=\sum_{p=0}^m h(D^p(\mathcal{A}),x)y^p\), and this identification is used to prove the Solomon-Terao formula under the tame hypothesis. The same work emphasizes that the highest-order poles of the Hilbert series are governed by the characteristic polynomial, whereas lower-order “tails” are not combinatorially determined even for arrangements with the same matroid [1110.2799].

## 3. Freeness, factorization, and algebraic realizations

For free arrangements, Solomon-Terao invariants admit explicit product decompositions. If \(\mathcal{A}\) is free with exponents \((d_1,\dots,d_\ell)\), then
\[
\Psi(\mathcal{A};x,t)=\prod_{i=1}^{\ell}\bigl(-t x^{d_i}+1+\cdots+x^{d_i-1}\bigr),
\]
and, for the specialization \(ST(\mathcal{A};x)=\Psi(\mathcal{A};x,-1)\),
\[
ST(\mathcal{A};x)=\prod_{i=1}^{\ell}(1+x+\cdots+x^{d_i}).
\]
These formulas show that in the free case the Solomon-Terao polynomial is entirely controlled by the exponent multiset [2305.10283][2509.10047].

An associated Artinian quotient, the Solomon-Terao algebra \(\mathrm{ST}(\mathcal{A},\eta)\), is defined from a homogeneous polynomial \(\eta\) through the Solomon-Terao ideal
\[
a(\mathcal{A},\eta)=\{\theta(\eta):\theta\in D(\mathcal{A})\},\qquad \mathrm{ST}(\mathcal{A},\eta)=S/a(\mathcal{A},\eta).
\]
For generic \(\eta\), this algebra is Artinian. In the framework of [1802.04056], \(\mathrm{Hilb}(\mathrm{ST}(\mathcal{A},\eta);x)=\Psi(\mathcal{A};x,1)\) for tame \(\mathcal{A}\). If \(\mathcal{A}\) is free and \(\eta\in U_d(\mathcal{A})\), then \(\mathrm{ST}(\mathcal{A},\eta)\) is a complete intersection and
\[
\mathrm{Hilb}(\mathrm{ST}(\mathcal{A},\eta);x)=\prod_{i=1}^{\ell}(1+x+\cdots+x^{d_i+d-2}).
\]
Conversely, \(\mathrm{ST}(\mathcal{A},\eta)\) is a complete intersection for generic \(\eta\) if and only if \(\mathcal{A}\) is free [1802.04056].

These algebraic realizations connect Solomon-Terao theory to geometric representation theory. For arrangements arising from a lower ideal \(I\) in a positive system, \(ST(\mathcal{A}_I;x)\) coincides with the topological Poincaré polynomial of the regular nilpotent Hessenberg variety \(\mathrm{Hess}(N,I)\). In the Weyl-arrangement case, \(\mathrm{ST}(\mathcal{A}_W,P_1)\cong \mathrm{Coinv}(W)\cong H^*(G/B,\mathbb{C})\), and for ideal arrangements,
\[
\mathrm{ST}(\mathcal{A}_I,P_1)\cong H^*(X(N,I)).
\]
This places Solomon-Terao polynomials and algebras at the intersection of arrangement theory, coinvariant-type algebras, and Hessenberg geometry [2509.10047][1802.04056].

## 4. Addition-deletion theory and \(B\)-sequences

Let \(H\in\mathcal{A}\), with deletion \(\mathcal{A}'=\mathcal{A}\setminus\{H\}\) and restriction \(\mathcal{A}^H\). A basic exact sequence considered in higher logarithmic degree is
\[
0\longrightarrow D^p(\mathcal{A}') \xrightarrow{\iota_H} D^p(\mathcal{A}) \xrightarrow{\pi_H} D^p(\mathcal{A}^H),
\]
where \(\iota_H\) is the natural inclusion and \(\pi_H\) is restriction modulo \(\alpha_H\). A central point is that this sequence is not right-exact in general. The addition-deletion theorems therefore require explicit surjectivity and projective-dimension hypotheses [2305.10283].

If \(\pi_H\) is surjective in codimension \(k+2\) along \(H\) and \(\mathrm{pdim}\,D^p(\mathcal{A}')=k<\ell-2\) for all \(p\), then
\[
\Psi(\mathcal{A};x,t)=x\Psi(\mathcal{A}';x,t)+\Psi(\mathcal{A}^H;x,t).
\]
If \(d:=|\mathcal{A}'|-|\mathcal{A}^H|\) and the \(B\)-sequence map is surjective in codimension \(k+2\) with \(\mathrm{pdim}\,D^p(\mathcal{A})=k<\ell-2\), then
\[
\Psi(\mathcal{A}';x,t)=\Psi(\mathcal{A};x,t)+x^d(t(x-1)-1)\Psi(\mathcal{A}^H;x,t),
\]
and at \(t=-1\),
\[
\Psi(\mathcal{A}';x,-1)=\Psi(\mathcal{A};x,-1)-x^{d+1}\Psi(\mathcal{A}^H;x,-1).
\]
When the arrangements involved are free, these hypotheses are automatically satisfied. Specializing \(x=1\) recovers the classical deletion-restriction formula for \(\chi(\mathcal{A};t)\), while specializing \(t=-1\) yields the corresponding recursion for Hessenberg Poincaré polynomials [2305.10283].

The homological mechanism is encoded by a generalized polynomial \(B\)-theory. The exact sequence
\[
0\to D^p(\mathcal{A}) \to D^p(\mathcal{A}') \xrightarrow{\delta} D^{p-1}(\mathcal{A}^H) B
\]
is called a \(B\)-sequence; for \(p=1\), it recovers Terao’s original polynomial \(B\)-theory. This extension clarifies that addition-deletion for Solomon-Terao polynomials is not purely combinatorial. Complementarily, work on Solomon-Terao algebras notes that deletion-restriction style formulas for \(\Psi(\mathcal{A};x,1)\) or the associated Hilbert series do not generally hold outside cases controlled by freeness [2305.10283][1802.04056].

## 5. Degree, regularity, and tame arrangements

A major recent problem has been the top degree of \(ST(\mathcal{A};x)\) for nonfree arrangements. For a central arrangement, tameness is defined by the condition that for all \(0\le p\le \ell\),
\[
\mathrm{pd}_S D^{\ell-p}(\mathcal{A})^* \le p.
\]
Under this hypothesis,
\[
ST(\mathcal{A};x)\text{ is a monic polynomial of degree }|\mathcal{A}|.
\]
This settles the top-degree problem for tame arrangements and confirms Conjecture 5.12 in [AMMN]. Since all \(3\)-arrangements are tame, it follows in particular that every \(3\)-arrangement satisfies
\[
\deg ST(\mathcal{A};x)=|\mathcal{A}|,
\]
with leading term \(x^{|\mathcal{A}|}\) [2509.10047].

The same work gives a more refined coefficient statement. For tame, irreducible arrangements,
\[
ST(\mathcal{A};x)=1+\ell x+\dots+(\ell+a)x^{n-1}+x^n,\qquad n:=|\mathcal{A}|,\ a\ge 0,
\]
where \(a\) is the number of relations of degree \(n\) among a minimal set of generators for \(D^{\ell-1}(\mathcal{A})\). The proof uses Castelnuovo-Mumford regularity of logarithmic derivation modules. For a possibly multi-arrangement \((\mathcal{A},m)\) with total multiplicity \(|m|\),
\[
\mathrm{reg}(D^p(\mathcal{A},m))\le |m|-\ell+p.
\]
This controls the degrees in minimal free resolutions and hence the highest nonvanishing degrees in the Solomon-Terao polynomial [2509.10047].

The multiarrangement extension is explicit. If
\[
ST_{d+1}(\mathcal{A},m;x)=\Psi(\mathcal{A},m;x,\frac{1-x^d}{x-1}),
\]
then for tame multiarrangements,
\[
\deg ST_{d+1}(\mathcal{A},m;x)=|m|+\ell(d-1).
\]
A plausible implication is that regularity bounds offer a systematic route to extracting precise degree data from logarithmic modules even when direct combinatorial control is unavailable [2509.10047].

## 6. Generalization to subspace arrangements

The theory extends from hyperplane arrangements to reduced equidimensional subspace arrangements \(\mathscr{X}\subseteq \mathbb{C}^\ell\) of codimension \(k\). Let \(S=\mathbb{C}[x_1,\dots,x_\ell]\), let \(I_{\mathscr{X}}\) be the vanishing ideal, and embed \(\mathscr{X}\) into a reduced homogeneous complete intersection subspace arrangement \(\mathscr{C}\) of the same codimension, defined by a regular sequence \(h_1,\dots,h_k\) with \(h=h_1\cdots h_k\). The multi-logarithmic forms are
\[
\Omega^q(\log \mathscr{X}/\mathscr{C})
=
\left\{
\omega\in \frac1h\Omega^q\ ;\
I_{\mathscr{X}}\omega\subseteq \frac1h I_{\mathscr{C}}\Omega^q,\
I_{\mathscr{X}}\wedge \omega\subseteq \frac1h I_{\mathscr{C}}\Omega^{q+1}
\right\},
\]
and the modules of multi-residues are
\[
\mathcal{R}_{\mathscr{X}}^q=\mathrm{res}\bigl(\Omega^{q+k}(\log \mathscr{X}/\mathscr{C})\bigr).
\]
They fit into the short exact sequence
\[
0\longrightarrow \frac1h I_{\mathscr{C}}\Omega^q
\longrightarrow
\Omega^q(\log \mathscr{X}/\mathscr{C})
\longrightarrow
\mathcal{R}_{\mathscr{X}}^{\,q-k}
\longrightarrow 0.
\]
A generalized \(\Psi\)-function is then defined for a graded family \(M^\bullet\) by
\[
\Psi(M^\bullet,x,t)=\sum_{q=0}^{n} Poin(M^q,x)[t(1-x)-1]^q,
\]
where \(Poin(M,x)\) is the Hilbert-Poincaré series [1803.08672].

For subspace arrangements, these generalized Solomon-Terao functions are polynomial. Moreover,
\[
\Psi(\Omega^\bullet(\log \mathscr{X}/\mathscr{C}),x,t)
=
\Psi\!\left(\tfrac1h I_{\mathscr{C}}\Omega^\bullet,x,t\right)
+
(t(1-x)-1)^k x^{-k}\Psi(\mathcal{R}_{\mathscr{X}}^\bullet,x,t).
\]
If for all \(Y\in L(\mathscr{X})\setminus\{V\}\) the residue condition
\[
\Psi(\mathcal{R}_{\mathscr{X}_Y}^\bullet,1,1)=1
\]
holds, then
\[
\chi(\mathscr{X}_Y,t)=t^\ell-\Psi(\mathcal{R}_{\mathscr{X}_Y}^\bullet,1,t),
\]
and if \(k\) is odd, then
\[
\chi(\mathscr{X}_Y,t)=\Psi(\Omega^\bullet(\log \mathscr{X}_Y/\mathscr{C}),1,t).
\]
For any line arrangement, the condition \(\Psi(\mathcal{R}_{\mathscr{X}}^\bullet,1,1)=1\) always holds, so
\[
\chi(\mathscr{X},t)=t^\ell-\Psi(\mathcal{R}_{\mathscr{X}}^\bullet,1,t).
\]
By contrast, for some equidimensional arrangements of dimension greater than one, the condition fails; the example with
\[
I_{\mathscr{X}}=(xy,xt,yz,zt)\subset \mathbb{C}^4
\]
has \(\Psi(\mathcal{R}_{\mathscr{X}}^\bullet,1,1)=2\), so the generalized formula does not recover the characteristic polynomial. This identifies a sharp boundary: the extension is universal for line arrangements of any codimension, but not for all higher-dimensional subspace arrangements [1803.08672].

Source: https://www.emergentmind.com/topics/solomon-terao-polynomial