Lagrange Subgroup: Interpolation and Modular Uses

• Lagrange subgroup is a context-dependent term that can denote either a 1-dimensional Lie subgroup of U(n) obtained via Lagrange interpolation or the modular subgroup Γ³ in PSL₂(ℤ).
• It bridges discrete reversible computation with continuous quantum gate evolution by extending finite cyclic groups to a smooth, unitary family.
• The term is distinct from related concepts like Lagrange subsets, Hom-subgroups, and Lagrangian subgroups, underscoring the need for precise contextual definitions.

The expression Lagrange subgroup does not have a single standardized meaning across the mathematical literature. In one usage, it denotes the 1-dimensional Lie subgroup of $U(n)$ obtained by applying Lagrange interpolation to the powers of a finite-order permutation or unitary matrix, thereby embedding a finite cyclic subgroup into a continuous unitary family (Vos et al., 2015). In another, it refers to the modular subgroup $\Gamma^3$, the subgroup generated by the cubes of $\mathrm{PSL}_2(\mathbb Z)$, which is the natural group-theoretic setting for the study of real numbers of classical Lagrange value $3$ (Schmidt et al., 2012). Related literatures use nearby but distinct notions—most notably Lagrange subsets, Hom-subgroups in a Lagrange-type divisibility theorem, and Lagrangian subgroups in finite group cohomology—so precision of context is essential.

1. Terminological scope and competing usages

The term is not formalized uniformly. In "From reversible computation to quantum computation by Lagrange interpolation" the authors do not introduce “Lagrange subgroup” as a separate formal definition, but they characterize an interpolating family $m(\theta)$ obtained from a finite cyclic subgroup $\{q^j\}$ as a construction that “steps from a finite cyclic group of order $p$ to a 1-dimensional Lie group, subgroup of the unitary group $U(n)$” (Vos et al., 2015). In this setting, “Lagrange subgroup” refers to a subgroup produced by interpolation.

In "Hausdorff dimension of the set of real numbers of Lagrange value three", by contrast, the subgroup $\Gamma^3$ is explicitly the subgroup generated by the cubes of $\mathrm{PSL}_2(\mathbb Z)$, and it is the subgroup through which the classical Lagrange value $\Gamma^3$0 is recast as a geometric problem on an orbifold and its Teichmüller space (Schmidt et al., 2012). Here the phrase is attached not to interpolation, but to a specific modular subgroup relevant to the Lagrange spectrum.

Several adjacent terms should not be conflated with either usage. A Lagrange subset of a finite group is merely a subset $\Gamma^3$1 with $\Gamma^3$2, and the classification of groups in which every such subset is a factor is a separate problem (Hooshmand et al., 15 Apr 2025). A Hom-subgroup belongs to the nonassociative theory of Hom-groups, where a Lagrange theorem of divisibility holds for finite Hom-groups (Hassanzadeh, 2018). A Lagrangian in finite group cohomology is a maximal isotropic subgroup of size $\Gamma^3$3 for a symplectic cohomology class, again a distinct notion (David et al., 2013).

Usage	Object	Source
Interpolated Lagrange subgroup	1-dimensional Lie subgroup of $\Gamma^3$4 obtained from a finite cyclic subgroup by Lagrange interpolation	(Vos et al., 2015)
Modular Lagrange subgroup	$\Gamma^3$5, generated by cubes of $\Gamma^3$6	(Schmidt et al., 2012)
Related but different terminology	Lagrange subsets, Hom-subgroups, Lagrangian subgroups	(Hooshmand et al., 15 Apr 2025, Hassanzadeh, 2018, David et al., 2013)

2. Interpolation from finite cyclic subgroups to continuous unitary subgroups

In the computational usage, the starting point is the matrix model of reversible computation. A classical reversible circuit on $\Gamma^3$7 bits is represented by a permutation matrix of size $\Gamma^3$8, and all such matrices form the group

$\Gamma^3$9

where $\mathrm{PSL}_2(\mathbb Z)$0 is the symmetric group on $\mathrm{PSL}_2(\mathbb Z)$1 objects (Vos et al., 2015). Since quantum circuits on $\mathrm{PSL}_2(\mathbb Z)$2 qubits are represented by unitary matrices of the same size, permutation matrices sit inside $\mathrm{PSL}_2(\mathbb Z)$3 as a discrete subclass.

If a permutation matrix $\mathrm{PSL}_2(\mathbb Z)$4 has finite order $\mathrm{PSL}_2(\mathbb Z)$5, then

$\mathrm{PSL}_2(\mathbb Z)$6

with $\mathrm{PSL}_2(\mathbb Z)$7 the identity matrix. Its powers

$\mathrm{PSL}_2(\mathbb Z)$8

form a finite cyclic subgroup isomorphic to $\mathrm{PSL}_2(\mathbb Z)$9 (Vos et al., 2015). The possible orders range from $3$0 up to the Landau function $3$1, the maximum order of a permutation in $3$2.

The key construction uses ordinary Lagrange interpolation, with interpolation nodes chosen as the $3$3-th roots of unity. Writing

$3$4

the paper defines a matrix-valued interpolant $3$5 satisfying

$3$6

Its main formula is

$3$7

with equivalent representations including

$3$8

The coefficients are the Lagrange fundamental polynomials

$3$9

which satisfy the Kronecker-delta property at the interpolation points. Consequently the family passes exactly through each element of the discrete cycle (Vos et al., 2015).

This usage motivates the expression Lagrange subgroup as an interpolative enlargement of a finite cyclic subgroup. A precise formulation consistent with the paper is: start with a finite-order unitary matrix $m(\theta)$0, interpolate the cycle $m(\theta)$1 across the $m(\theta)$2-th roots of unity, and obtain a continuous one-parameter subgroup $m(\theta)$3 (Vos et al., 2015).

3. Subgroup property, unitarity, and examples in reversible and quantum computation

The subgroup character of the interpolated family is not merely heuristic. A central result is

$m(\theta)$4

which is the defining property of a one-parameter group (Vos et al., 2015). It follows that

$m(\theta)$5

and the interpolated family is closed under multiplication and inversion. Because the family is continuous, the resulting object is a 1-dimensional Lie group.

The same paper proves that if $m(\theta)$6 is unitary, then $m(\theta)$7 is unitary, with

$m(\theta)$8

Hence

$m(\theta)$9

so the family lies inside $\{q^j\}$0 and forms a subgroup there (Vos et al., 2015). When the initial matrix belongs to the subgroup $\{q^j\}$1, defined as the unitary matrices whose row sums and column sums are all $\{q^j\}$2, the interpolation stays in $\{q^j\}$3 because the coefficients satisfy

$\{q^j\}$4

The computational significance is that a discrete reversible gate cycle is promoted to a continuous family of quantum gates. For $\{q^j\}$5, the 2-cycle generated by NOT yields

$\{q^j\}$6

the NEGATOR gate, with $\{q^j\}$7 and $\{q^j\}$8 (Vos et al., 2015). For $\{q^j\}$9, the paper shows that a 4-cycle can be interpolated so that the midpoint $p$0 is a square root of NOT, i.e. a $p$1 gate. Another choice of 4-cycle gives a different interpolation $p$2, demonstrating that the interpolation is not unique: different cycles can connect the same endpoints by different continuous paths.

This suggests that, in this literature, a Lagrange subgroup is best understood as an interpolative Lie-theoretic completion of a discrete cyclic subgroup inside the full unitary group. The emphasis is constructive rather than classification-theoretic.

4. The modular subgroup $p$3 and the Lagrange value $p$4

A second usage appears in the study of the classical Lagrange spectrum. For an irrational $p$5, the classical Lagrange value is

$p$6

The paper "Hausdorff dimension of the set of real numbers of Lagrange value three" studies the special value $p$7 not directly through $p$8, but through the subgroup $p$9, defined as the subgroup generated by the cubes of $U(n)$0 (Schmidt et al., 2012).

For a zonal Fuchsian group $U(n)$1, the analogous quantity is

$U(n)$2

When $U(n)$3, this recovers the classical Lagrange value. The paper states that the classical Lagrange value of a real number is the same as its $U(n)$4-Lagrange value, because left cosets of this subgroup have representatives given by translations, so the relevant approximation behavior is unchanged (Schmidt et al., 2012).

The main theorem proves that the set of real numbers of Lagrange value exactly $U(n)$5 has Hausdorff dimension $U(n)$6. More generally, the abstract states the corresponding result for each element of the Teichmüller space of the commutator subgroup of the classical modular group, while the body formulates it for the Teichmüller space of hyperbolic genus zero orbifolds with one cusp and three elliptic fixed points of order two (Schmidt et al., 2012).

The geometric mechanism is organized by Fricke triples $U(n)$7 satisfying

$U(n)$8

with modular case $U(n)$9. In that case the adjusted Fricke equation becomes Markoff’s equation. The proof constructs a tree of triples using the moves

$\Gamma^3$0

and then uses horocycle-shadow excision intervals whose overlapping structure produces a Cantor set of Hausdorff dimension zero (Schmidt et al., 2012).

In this literature, then, the “Lagrange subgroup” is a specific modular subgroup governing a geometric reformulation of a Diophantine approximation problem. The subgroup is not produced by interpolation; it is chosen because it preserves the relevant Lagrange-value structure at the threshold $\Gamma^3$1.

5. Neighboring notions often confused with Lagrange subgroups

A substantial source of ambiguity is that several active research areas use “Lagrange” language for objects that are not subgroups in either of the senses above.

In finite group factorization theory, a Lagrange subset is a subset $\Gamma^3$2 such that $\Gamma^3$3. The product $\Gamma^3$4 is called direct, written $\Gamma^3$5, if every element of $\Gamma^3$6 has a unique representation $\Gamma^3$7, and a subset is a factor if it admits such a complement. The classification theorem shows that a finite group has the property that every Lagrange subset is a factor if and only if it is one of

$\Gamma^3$8

and no nonabelian finite group satisfies this strong CFS property (Hooshmand et al., 15 Apr 2025). A later note gives a comparatively direct proof and states the same nontrivial list as

$\Gamma^3$9

(Kabenyuk, 28 Dec 2025). These papers concern subset factorization, not a notion of Lagrange subgroup.

In the nonassociative setting of Hom-groups, a Hom-group $\mathrm{PSL}_2(\mathbb Z)$0 is a set with twisted associativity,

$\mathrm{PSL}_2(\mathbb Z)$1

multiplicativity of the twist, twisted unitality,

$\mathrm{PSL}_2(\mathbb Z)$2

and inverses. A subset $\mathrm{PSL}_2(\mathbb Z)$3 is a Hom-subgroup if $\mathrm{PSL}_2(\mathbb Z)$4 is itself a Hom-group. For finite Hom-groups, cosets still partition the ambient set, and the paper proves the Lagrange-type theorem

$\mathrm{PSL}_2(\mathbb Z)$5

This is a generalization of Lagrange’s theorem, but not a distinct object called a Lagrange subgroup (Hassanzadeh, 2018).

A further neighboring notion is the Lagrangian of a symplectic $\mathrm{PSL}_2(\mathbb Z)$6-form in group cohomology. If $\mathrm{PSL}_2(\mathbb Z)$7 is symplectic and $\mathrm{PSL}_2(\mathbb Z)$8 is isotropic with

$\mathrm{PSL}_2(\mathbb Z)$9

then $\Gamma^3$00 is called a Lagrangian (David et al., 2013). The paper proves that symplectic forms over finite nilpotent groups always admit Lagrangians, exhibits examples where none is normal, and shows that normal Lagrangians always exist when all $\Gamma^3$01-Sylow subgroups have order less than $\Gamma^3$02. This is conceptually close to classical symplectic geometry, but terminologically different from “Lagrange subgroup.”

The distinction matters because the same root word labels structurally different constructions: interpolation in $\Gamma^3$03, modular subgroups for the Lagrange spectrum, subset factorizations, twisted subgroup divisibility, and maximal isotropic subgroups.

6. Conceptual synthesis

Across the literature, Lagrange subgroup is best treated as a context-dependent expression rather than a universal technical term. The strongest explicit meanings are the interpolated one-parameter subgroup $\Gamma^3$04 arising from Lagrange interpolation of a finite cyclic subgroup (Vos et al., 2015), and the subgroup $\Gamma^3$05 generated by cubes, which organizes the geometry of the Lagrange value $\Gamma^3$06 problem (Schmidt et al., 2012).

These two usages are mathematically unrelated in construction but share a common structural role. In the first, a subgroup mediates the passage from classical reversible computation to quantum computation by embedding a discrete cycle into a continuous unitary family (Vos et al., 2015). In the second, a subgroup mediates the passage from classical Diophantine approximation to a geometric-dynamical formulation on orbifolds and their Teichmüller spaces (Schmidt et al., 2012). This suggests a common editorial principle: the phrase tends to designate the subgroup that carries the specifically “Lagrange” structure of the problem at hand.

At the same time, nearby literatures show that the term should not be generalized loosely. A Lagrange subset need not be a subgroup (Hooshmand et al., 15 Apr 2025); a Hom-subgroup is defined independently of any special Lagrange nomenclature (Hassanzadeh, 2018); and a Lagrangian subgroup belongs to the theory of symplectic group cohomology rather than to either of the two principal uses above (David et al., 2013). For technical writing, therefore, the expression Lagrange subgroup should always be accompanied by its ambient framework—unitary interpolation, modular group theory, subset factorization, Hom-group theory, or symplectic cohomology—to avoid categorical ambiguity.