Arithmetic Root Systems

• Arithmetic root systems are structures that embed arithmetic data—such as number fields, lattices, and congruence conditions—into classical root definitions.
• They refine and extend standard root systems by incorporating constraints from Weyl groups, Nichols algebras, and cyclotomic integrality, leading to context-dependent classifications.
• Their applications span arithmetic Tutte theory, toric arrangements, and complex reflection groups, providing concrete tools for analyzing algebraic and combinatorial invariants.

Arithmetic root systems are root-theoretic structures in which the ambient arithmetic datum—such as a number field, a lattice, a Dedekind domain, a bicharacter, or a congruence condition on heights—enters essentially into the definition of roots, reflections, or multiplicities. The phrase does not designate a single uniform object across the literature. It is used for classical finite root systems realized inside rings of integers of number fields (Popov et al., 2018), for the finite root systems attached to Cartan graphs and Weyl groupoids of Nichols algebras of diagonal type (Yuan et al., 2024), and for the $\mathbb Z_k$-root systems introduced for finite complex reflection groups, for which the term “cyclotomic root systems” is proposed (Broué et al., 2017). Closely related arithmetic variants include arithmetic Tutte theory for pairs $(\Phi,\Lambda)$ (Ardila et al., 2013), congruence-defined subsystems $R(m)$ cut out by root heights (Polo, 12 Apr 2025), and Lorentzian hyperbolic systems of arithmetic type in the sense of Gritsenko–Nikulin (Allcock, 2012).

1. Terminological scope and conceptual distinctions

In the number-field setting, an arithmetic root system is an ordinary finite root system placed inside $\mathscr O_K$, the ring of integers of a number field $K$, with the Weyl action constrained to come from field automorphisms and multiplication operators (Popov et al., 2018). In arithmetic Tutte theory, by contrast, the arithmetic datum is not a new class of roots but the choice of ambient lattice $\Lambda$, encoded by multiplicities $m(B)$ in the arithmetic Tutte polynomial $M_A(x,y)$ (Ardila et al., 2013). In the Nichols algebra literature, “finite arithmetic root systems” arise from diagonal braidings, reflections, Cartan graphs, and Weyl groupoids; there the root system is attached to a tuple of one-dimensional Yetter–Drinfeld modules and its finiteness is equivalent to finiteness of the associated Weyl groupoid (Yuan et al., 2024). In the theory of complex reflection groups, roots become rank-one projective $\mathcal O_k$-modules on reflecting lines, with dual data and cyclotomic integrality conditions, yielding $\mathbb Z_k$-root systems or “cyclotomic root systems” (Broué et al., 2017).

These usages are not interchangeable. The number-field paper explicitly distinguishes its notion from the Nichols algebra and Weyl groupoid usage (Popov et al., 2018), while the cyclotomic theory is designed as an arithmetic extension of Bourbaki root systems for complex reflection groups rather than a Nichols algebra construction (Broué et al., 2017). Related frameworks such as quotient root systems and generalized root systems extend positive-root combinatorics or restriction theory, but they are not arithmetic root systems in the Nichols algebra sense (Dimitrov et al., 2023, Cuntz et al., 2024). This suggests that “arithmetic root systems” functions as a context-dependent umbrella term rather than a single canonical definition.

2. Root systems over number fields

For a number field $(\Phi,\Lambda)$0 of degree $(\Phi,\Lambda)$1, viewed as an $(\Phi,\Lambda)$2-dimensional $(\Phi,\Lambda)$3-vector space, the ambient arithmetic symmetry group is

$(\Phi,\Lambda)$4

with semidirect product decomposition

$(\Phi,\Lambda)$5

A type $(\Phi,\Lambda)$6 admits a realization in $(\Phi,\Lambda)$7 if $(\Phi,\Lambda)$8, there exists a subset $(\Phi,\Lambda)$9 of that rank forming a root system of type $R(m)$0, and its Weyl group satisfies $R(m)$1 (Popov et al., 2018).

The classification is extremely restrictive. For reduced root systems of rank $R(m)$2, the Weyl group can be isomorphic to a subgroup of $R(m)$3 for some number field $R(m)$4 of degree $R(m)$5 only for

$R(m)$6

However, actual realizability inside $R(m)$7 is stricter: a root-system type, reduced or not, admits such an arithmetic realization if and only if its rank is $R(m)$8 or $R(m)$9 (Popov et al., 2018). In particular, the two reducible rank-$\mathscr O_K$0 types allowed abstractly at the Weyl-group level do not occur as root systems in rings of integers.

The proof is driven by arithmetic constraints on finite subgroups $\mathscr O_K$1. If $\mathscr O_K$2 is finite, there is an exact sequence

$\mathscr O_K$3

with $\mathscr O_K$4 cyclic, implying

$\mathscr O_K$5

Additional bounds,

$\mathscr O_K$6

combined with Weyl-group order data and the lower estimate

$\mathscr O_K$7

force $\mathscr O_K$8, and the rank-$\mathscr O_K$9 realizations are then excluded by further arguments (Popov et al., 2018).

All realizable cases are constructed explicitly. In rank $K$0, $K$1 and the nonreduced $K$2 occur in $K$3. In rank $K$4, $K$5 realizes $K$6 and $K$7, while $K$8 realizes $K$9, $\Lambda$0, $\Lambda$1, $\Lambda$2, and $\Lambda$3 (Popov et al., 2018). The construction uses the reflection formula

$\Lambda$4

in the Gaussian and Eisenstein examples.

3. Arithmetic lattices, toric arrangements, and arithmetic Tutte theory

A second major arithmetic use of root systems treats them as lattice vector configurations. For a configuration $\Lambda$5, the arithmetic Tutte polynomial is

$\Lambda$6

where

$\Lambda$7

This refines the ordinary Tutte polynomial by recording the arithmetic saturation of subsets inside the ambient lattice (Ardila et al., 2013).

For the classical root systems $\Lambda$8, the arithmetic data depend on whether the roots are placed in the integer lattice, root lattice, or weight lattice. The paper computes the corresponding arithmetic Tutte generating functions in all three cases (Ardila et al., 2013). The distinction is substantive: the same root configuration can have different arithmetic Tutte polynomials in $\Lambda$9, $m(B)$0, and $m(B)$1. Type $m(B)$2 is unimodular in the integer and root lattices, so arithmetic and ordinary Tutte polynomials coincide there, whereas the weight-lattice case is subtler because $m(B)$3. Types $m(B)$4 already exhibit nontrivial multiplicities over $m(B)$5.

The arithmetic Tutte polynomial governs several invariants of toric and hypertoric arrangements. For the toric arrangement $m(B)$6 in $m(B)$7, it controls region counts, Poincaré polynomials, zonotope volumes, Ehrhart data, and the dimensions of Dahmen–Micchelli and De Concini–Procesi–Vergne spaces: $m(B)$8

$m(B)$9

A finite-field method gives

$M_A(x,y)$0

provided $M_A(x,y)$1 for all $M_A(x,y)$2 (Ardila et al., 2013).

The computational apparatus is also root-system specific. For types $M_A(x,y)$3, subsets of roots are encoded by signed graphs with six parameters $M_A(x,y)$4, and the arithmetic multiplicities become graph-theoretic: $M_A(x,y)$5 In this line of work, arithmetic root theory is best understood as the study of the pair $M_A(x,y)$6, rather than of a new intrinsic root-system class (Ardila et al., 2013).

4. Finite arithmetic root systems in Nichols algebras

In the Nichols algebra framework, arithmetic root systems arise from braided vector spaces of diagonal type over a field $M_A(x,y)$7 of characteristic $M_A(x,y)$8. With basis $M_A(x,y)$9, the braiding is

$\mathcal O_k$0

The generalized Dynkin diagram records the vertex labels $\mathcal O_k$1 and the symmetric edge labels $\mathcal O_k$2 (Yuan et al., 2024).

Reflections are defined through braided adjoint nilpotency. If $\mathcal O_k$3 is $\mathcal O_k$4-finite, then

$\mathcal O_k$5

and the reflected tuple $\mathcal O_k$6 is built from the vectors $\mathcal O_k$7 (Yuan et al., 2024). These data generate a semi-Cartan graph $\mathcal O_k$8, and the corresponding Weyl groupoid acts on $\mathcal O_k$9 by

$\mathbb Z_k$0

In this setting, the arithmetic root system attached to $\mathbb Z_k$1 is the root system $\mathbb Z_k$2 of the Cartan graph $\mathbb Z_k$3. Its root sets are extracted from the PBW decomposition of the Nichols algebra, and finiteness is decisive. The central theorem states that, assuming $\mathbb Z_k$4 admits all reflections, the following are equivalent: $\mathbb Z_k$5 is finite, $\mathbb Z_k$6 is a finite Cartan graph, $\mathbb Z_k$7 is finite, and $\mathbb Z_k$8 is finite. In that case $\mathbb Z_k$9 is the unique root system of type $(\Phi,\Lambda)$00 (Yuan et al., 2024).

Positive characteristic enters through the quantum-integer criterion

$(\Phi,\Lambda)$01

with $(\Phi,\Lambda)$02. Because $(\Phi,\Lambda)$03 may occur when $(\Phi,\Lambda)$04, the set of admissible Cartan integers and hence the possible finite arithmetic root systems depend on $(\Phi,\Lambda)$05 (Yuan et al., 2024). The paper classifies all finite-dimensional Nichols algebras of diagonal type of ranks $(\Phi,\Lambda)$06 over fields of positive characteristic, using “good $(\Phi,\Lambda)$07,” “good $(\Phi,\Lambda)$08,” and “good $(\Phi,\Lambda)$09” neighborhoods as local models for finite connected indecomposable Cartan graphs.

5. Cyclotomic root systems for complex reflection groups

For finite complex reflection groups, arithmetic root theory takes a Dedekind-domain form. Let $(\Phi,\Lambda)$10 be a number field stable under complex conjugation, with ring of integers $(\Phi,\Lambda)$11, and let $(\Phi,\Lambda)$12 be dual $(\Phi,\Lambda)$13-vector spaces with a nondegenerate Hermitian pairing. A $(\Phi,\Lambda)$14-root is a triple

$(\Phi,\Lambda)$15

where $(\Phi,\Lambda)$16 and $(\Phi,\Lambda)$17 are rank-one finitely generated $(\Phi,\Lambda)$18-modules, $(\Phi,\Lambda)$19 is a nontrivial root of unity, and

$(\Phi,\Lambda)$20

The associated reflection is

$(\Phi,\Lambda)$21

for $(\Phi,\Lambda)$22, $(\Phi,\Lambda)$23 with $(\Phi,\Lambda)$24 (Broué et al., 2017).

A $(\Phi,\Lambda)$25-root system is a finite set of such roots satisfying three axioms: the $(\Phi,\Lambda)$26-modules generate $(\Phi,\Lambda)$27, the system is stable under the attached reflections, and the Cartan pairings

$(\Phi,\Lambda)$28

lie in $(\Phi,\Lambda)$29 (Broué et al., 2017). This arithmetic integrality condition replaces the usual crystallographic condition. Because $(\Phi,\Lambda)$30 and $(\Phi,\Lambda)$31 may be nonfree rank-one projective modules, class-group effects and ideal factorizations become intrinsic to the theory.

The resulting theory extends Bourbaki root systems to complex reflection groups. It supports root and coroot lattices,

$(\Phi,\Lambda)$32

weight and coweight lattices, Cartan matrices in the principal case, parabolic restriction, and connection indices (Broué et al., 2017). For irreducible reflection groups over their field of definition, the paper classifies genera of root systems in the imprimitive families $(\Phi,\Lambda)$33 and for the primitive exceptional groups. It also generalizes the notion of bad primes: for an irreducible well-generated complex reflection group $(\Phi,\Lambda)$34 of rank $(\Phi,\Lambda)$35, the bad prime ideals are those dividing

$(\Phi,\Lambda)$36

where $(\Phi,\Lambda)$37 is the connection index (Broué et al., 2017).

The authors explicitly propose the term “cyclotomic root systems” for this framework. That terminology is important because it distinguishes this module-theoretic arithmetic extension of classical root data from the Weyl-groupoid-based arithmetic root systems of Nichols algebra theory (Broué et al., 2017).

6. Congruence constructions, Lorentzian arithmeticity, and related boundaries

A different arithmetic construction imposes a congruence on heights. For a reduced irreducible root system $(\Phi,\Lambda)$38 with fixed positive system and Coxeter number $(\Phi,\Lambda)$39, and for $(\Phi,\Lambda)$40,

$(\Phi,\Lambda)$41

Let $(\Phi,\Lambda)$42 be the positive roots of height exactly $(\Phi,\Lambda)$43, and $(\Phi,\Lambda)$44 the simple roots of $(\Phi,\Lambda)$45. The complete classification shows that $(\Phi,\Lambda)$46 always, and $(\Phi,\Lambda)$47 except for an explicit finite list of cases in types $(\Phi,\Lambda)$48, where one additional simple root $(\Phi,\Lambda)$49 of height $(\Phi,\Lambda)$50 must be added (Polo, 12 Apr 2025). In the non-exceptional cases $(\Phi,\Lambda)$51 is of Levi type. In the exceptional cases the constant $(\Phi,\Lambda)$52 is the dimension of a minuscule representation of the dual group, and the possible values include natural-representation dimensions and spin dimensions such as $(\Phi,\Lambda)$53 in the $(\Phi,\Lambda)$54 and $(\Phi,\Lambda)$55 even-$(\Phi,\Lambda)$56 families (Polo, 12 Apr 2025).

Arithmeticity also appears in Lorentzian Kac–Moody theory. For rank-$(\Phi,\Lambda)$57 hyperbolic simple-root systems in $(\Phi,\Lambda)$58, Allcock classifies those satisfying the Gritsenko–Nikulin conditions: the Tits cone interior equals the future cone, the Weyl-group normalizer has finite index in $(\Phi,\Lambda)$59, and there exists a Weyl vector $(\Phi,\Lambda)$60 with

$(\Phi,\Lambda)$61

In the timelike case, equivalent to having finitely many simple roots, there are exactly $(\Phi,\Lambda)$62 such systems, with as many as $(\Phi,\Lambda)$63 simple roots (Allcock, 2012). Here “arithmetic type” refers to arithmetic reflection groups and automorphic correction rather than to lattices in the arithmetic Tutte or Nichols algebra senses.

Related generalizations sharpen the terminological boundary. Quotient root systems extend inversion-set combinatorics beyond ordinary Weyl groups, but the framework remains Euclidean and finite-type (Dimitrov et al., 2023). Generalized root systems in the sense of Dimitrov–Fioresi are finite subsets of Euclidean spaces satisfying closure rules based on signs of inner products; every irreducible example of rank at least $(\Phi,\Lambda)$64 is equivalent to a quotient of a classical finite Weyl root system, not an arithmetic root system in the Nichols algebra sense (Cuntz et al., 2024). A plausible implication is that the arithmetic qualifier is most informative only when the governing datum—number field, lattice index, projective $(\Phi,\Lambda)$65-module, bicharacter, or congruence condition—is made explicit.