Hill Number: Graph Theory, Biochemistry & Statistics

• Hill Number is a multi-context concept defining minimal graph crossings, binding cooperativity, and tail index estimation across different scientific disciplines.
• It utilizes rigorous combinatorial constructions and geodesic sphere methods in graph theory to achieve optimal drawing designs and validate minimal crossing numbers.
• Methodologies extend to biochemical models with deterministic and stochastic Hill functions and statistical techniques using generalized Hill estimators for heavy-tailed distributions.

The term "Hill number" appears in several distinct mathematical and scientific contexts, most notably in graph theory (as the conjectured minimal crossing number for complete graphs), biochemistry and statistical physics (as the Hill coefficient quantifying cooperative ligand binding), and in mathematical statistics (as an index or parameter in heavy-tail estimation). Each usage reflects different yet sometimes analogous mathematical phenomena. This article surveys the rigorous definitions, theoretical principles, historical roots, methodologies, and current research status surrounding prominent notions of the Hill number, emphasizing technical depth suitable for advanced researchers.

1. Hill Number in Crossing Number Theory

The Hill number, denoted $H(n)$, emerged from a conjecture in topological graph theory regarding the minimal number of crossings achievable in any drawing of the complete graph $K_n$ in the plane. Anthony Hill, working in the 1950s, proposed the following formula:

$$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$

Explicitly, for even $n=2k$: $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$ and for odd $n=2k+1$: $H(2k+1) = \frac{1}{4} k(k-1)(k-1)(k-2).$

This formula was motivated by explicit constructions producing drawings of $K_n$ with exactly $H(n)$ crossings. Subsequent work by Blažek and Koman (1963) rediscovered alternative planar drawings (notably 2-page and cylindrical layouts) achieving this count (Mohar, 2020).

The central conjecture—Hill's conjecture—asserts that $H(n)$ is the minimal crossing number for $K_n$0 for all $K_n$1.

2. Explicit Constructions and the “Book Proof”

A major breakthrough was the construction of a broad family of geodesic drawings on the unit sphere $K_n$2, all realizing exactly $K_n$3 crossings. For even $K_n$4:

• Select $K_n$5 points $K_n$6 in general position (no three colinear) on $K_n$7.
• Form the antipodal double $K_n$8.
• Draw geodesic arcs between all non-antipodal pairs in $K_n$9; do not connect antipodal pairs.

Counting crossings reduces to enumerating all disjoint pairs of basic 4-cycles formed along great circles, yielding exactly $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$0 crossings. Even after reinserting the perfect matching edges (the omitted antipodal connections), the final crossing count remains $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$1 due to symmetry and combinatorial cancelation. For odd $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$2, adding or removing a point from this construction preserves the formula, shifting it to $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$3 or $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$4 accordingly (Mohar, 2020).

3. General Drawing Schemes and Implications

The construction method generalizes the known families of cylindrical and 2-page planar drawings. For each even $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$5:

• The base configuration relies on antipodal symmetry and choice of geodesic representatives.
• The structure of arc intersections is governed by antipodal–general position, which precludes accidental confluence of crossings.
• For odd $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$6, adjustment is direct by vertex deletion or insertion.

Recent work by Ábrego et al. confirms that within both cylindrical and 2-page paradigms, no drawing of $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$7 can have fewer than $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$8 crossings, further reinforcing Hill's conjecture. The new spherical construction contains these earlier families as special cases and significantly broadens the class of optimal drawings.

4. Asymptotics: Moon’s Phenomenon

Moon (1968) established a striking asymptotic result: If $$H(n) = \frac{1}{4} \left\lfloor \frac{n}{2} \right\rfloor \left\lfloor \frac{n-1}{2} \right\rfloor \left\lfloor \frac{n-2}{2} \right\rfloor \left\lfloor \frac{n-3}{2} \right\rfloor.$$9 points are distributed independently and uniformly on $n=2k$0, and edges are drawn as shorter geodesics, the expected crossing number $n=2k$1 satisfies:

$n=2k$2

This arises because random point sets on the sphere are, with high probability, in antipodal–general position. The expected crossing number thus becomes sharply concentrated around $n=2k$3 up to lower-order fluctuations. This phenomenon has been generalised by Mohar and Wesolek to all antipodally symmetric distributions on $n=2k$4 with zero measure on any great circle (Mohar, 2020).

5. Hill Number and Hill Coefficient in Biochemistry and Stochastic Modeling

Outside graph theory, the term Hill number denotes the Hill coefficient, which parametrizes the nonlinearity (cooperativity) of ligand binding in biochemical and gene regulatory systems. The classical deterministic Hill function:

$n=2k$5

(with $n=2k$6 the ligand concentration, $n=2k$7 the Hill exponent) arises when $n=2k$8 ligands bind a single receptor in a one-step, mass–action equilibrium. Mechanistically, $n=2k$9 is required to be integer.

However, empirical fits are frequently parameterized by non-integer (decimal) Hill numbers. This effect is attributed either to heterogeneous intermediate steps with distinct dissociation constants or to mesoscopic (finite-size) stochastic fluctuations. The latter induces effective corrections so that, after linear-noise expansion, the effective Hill exponent becomes

$H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$0

which is generically non-integer for $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$1 with $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$2 the variance of concentration noise (Hernández-García et al., 2023).

Inverting an empirical decimal Hill coefficient allows recovery of underlying microscopic dissociation constants, both for sequential and independent binding models. The procedure extends unchanged in the presence of controlled fluctuations.

6. Hill Numbers in Statistical Tail-Index Estimation

In mathematical statistics, the “Hill number” also references the family of estimators for the extreme value (tail) index $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$3 of heavy-tailed distributions. The generalized Hill process is defined for i.i.d. $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$4 and order statistics $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$5 as

$H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$6

where $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$7 recovers the classical Hill estimator.

A key property is that for $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$8, $H(2k) = \frac{1}{4} k(k-1)(k-2)(k-3),$9 is asymptotically normal under mild regularity, with variance minimized at $n=2k+1$0. For $n=2k+1$1, the limiting law becomes non-Gaussian, expressible in terms of infinite series involving the Riemann zeta function. This transition reflects a deep connection between statistical estimation in extremes and analytic number theory (Lo et al., 2011).

7. Summary Table: Hill Number Across Disciplines

Context	Mathematical Definition	Core Application
Crossing number of $n=2k+1$2	$n=2k+1$3 from floor functions or binomial counts	Optimal graph drawing, combinatorics
Biochemical binding (Hill coefficient)	$n=2k+1$4 exponent in $n=2k+1$5	Cooperativity in enzymatic/gene binding
Generalized Hill estimator in statistics	$n=2k+1$6 parameter for tail index $n=2k+1$7	Heavy-tail/EV index estimation

Each use of the Hill number encapsulates a fundamental quantitative invariant: minimized topological complexity (graph crossings), degree of cooperativity and complexity in molecular biophysics, or the extremal nature of heavy-tailed distributions.

8. Current Research Directions and Open Problems

Hill's conjecture concerning crossing numbers remains open for general $n=2k+1$8 and motivates ongoing work in topological combinatorics and extremal graph theory. The geodesic sphere constructions, providing a “book proof” of the $n=2k+1$9 crossing count in a vast class of drawings, further strengthens the conjectural minimality, yet a general proof remains elusive (Mohar, 2020).

In systems biology and chemical physics, the rigorous derivation and mechanistic interpretation of decimal Hill coefficients continues to unify stochastic process modeling with observed "effective" non-integer cooperativities, advancing understanding of fluctuation-induced apparent nonlinearity (Hernández-García et al., 2023).

In extreme value theory, the generalized Hill process and its phase transition from Gaussian to non-Gaussian limit laws for small parameters inspires both statistical and analytic exploration, especially in the interplay between heavy-tail phenomena and deep functions from number theory (Lo et al., 2011).