Sigma-Irregularity Index

• Sigma-Irregularity Index is a graph invariant that quantifies degree heterogeneity through quadratic penalties on differences in vertex degrees.
• It features local edge-based and global vertex pair-based formulations, serving as a critical tool in chemical graph theory and extremal combinatorics.
• Analytical bounds, explicit formulas for trees and graphs, and algorithmic approaches underpin its applications in network analysis and QSPR/QSAR studies.

The Sigma-Irregularity Index ($\sigma$-index, also total $\sigma$-irregularity in its global form) is a graph-theoretic invariant designed to quantify the degree heterogeneity of a graph through quadratic penalties on degree differences. Originally motivated by chemical graph theory and extremal combinatorics, the index provides a rigorous measure for the structural irregularity of a graph’s degree sequence. Both local (edge-based) and global (vertex pair-based) formulations are studied, and a wide array of extremal, structural, and analytic results are available for graphs, trees, bipartite constructions, and graphs with prescribed degree constraints or sequences.

1. Formal Definitions and Structural Variants

Let $G=(V,E)$ be a simple undirected graph with $n=|V|$ and $m=|E|$. For each vertex $v\in V$, denote its degree by $d_G(v)$. The main forms of the $\sigma$-irregularity index are:

• Edge-based Sigma-Irregularity (Standard $\sigma$-index):

$\sigma(G) = \sum_{uv\in E} (d_G(u) - d_G(v))^2$

This is also referred to as the quadratic Albertson index and measures local (edgewise) degree imbalance (Lin et al., 2021, Hamoud et al., 9 Jun 2025).

• Total Sigma-Irregularity ($\sigma_t$-index):

$$\sigma_t(G) = \sum_{\{u,v\} \subseteq V} (d_G(u) - d_G(v))^2$$

Here the sum is over all unordered pairs, yielding a global measure of the spread of the degree sequence. It satisfies

$\sigma_t(G) = n^2 \operatorname{Var}(G)$

where $\operatorname{Var}(G)$ is the variance of the degree sequence. This form is more sensitive to the global degree distribution and is used in applications where such structure is critical (Filipovski et al., 2024).

• Generalized and Modified Sigma-Irregularity:

Extensions include the $\ell_p$-type irregularity

$$A_p(G) \equiv \left( \sum_{uv\in E} |d_G(u) - d_G(v)|^p \right)^{1/p}$$

with $p=2$ recovering the $\sigma$-index, and modified exponents in global indices to study extremal graphs under alternative weighting of degree differences (Knor et al., 2024, Lin et al., 2021).

2. Analytical Bounds and Closed-Form Expressions

2.1. Extremal Graphs and Trees

For connected graphs, the extremal values of the $\sigma$-index have been rigorously characterized:

• Maximum $\sigma$ for Trees:

$\sigma(S_n) = (n-1)(n-2)^2$ achieved uniquely by the star $S_n$ (Hamoud et al., 9 Jun 2025, Lin et al., 2021).

• Minimum $\sigma$ for Trees:

$\sigma(P_n) = 2$ achieved by the path $P_n$.

For total $\sigma$-irregularity, the graphs attaining the maximum are split graphs with very few distinct degree values (specifically, $S_{\lceil n/4\rceil,\lfloor 3n/4\rfloor}$ and $S_{\lfloor n/4\rfloor, \lceil 3n/4\rceil}$) (Filipovski et al., 2024, Hamoud et al., 13 Feb 2025). For trees under maximum degree constraint $\Delta$, the maximal $\sigma$-index and the structural nature of extremal realizations become increasingly intricate with rising $\Delta$ (Bašić, 1 Feb 2026).

2.2. Explicit Formulae for Trees

A general closed-form expression for trees with degree sequence $(d_1, d_2, \ldots, d_n)$ is given by (Hamoud et al., 22 Oct 2025, Hamoud et al., 9 Jun 2025): $$\sigma(T) = \sum_{i\in\{1,n\}}(d_i+1)(d_i-1)^2 + \sum_{i=2}^{n-1}(d_i+2)(d_i-1)^2 + \sum_{i=2}^{n-1}(d_i-d_{i+1})^2 + 2n-2$$

In standard families:

• For star $S_n$: $\sigma(S_n) = (n-1)(n-2)^2$
• For path $P_n$: $\sigma(P_n) = 2$
• For caterpillar $\mathscr C(n,m)$ with $m$ leaves per backbone vertex:

$$\sigma(\mathscr C(n, m)) = 2m^3 + (n-2)m(m+1)^2 + 2$$

(Hamoud et al., 22 Oct 2025, Hamoud et al., 13 Feb 2025)

3. Inequalities, Extremal Bounds, and Degree Parameters

3.1. Degree-based General Bounds

For a tree $T$ with minimum degree $\delta$ and maximum degree $\Delta$ (Hamoud et al., 9 Jun 2025, Hamoud et al., 19 Jul 2025): $$\frac{(\operatorname{irr}(T))^2}{n-1} \leq \sigma(T) \leq (n-1)(\Delta-\delta)^2$$ where $\operatorname{irr}(T)$ is the linear Albertson index.

General lower bounds for connected graphs with minimum degree $\delta$ and maximum degree $\Delta$: $$\sigma(G) > \frac{\delta}{\Delta+1} (\Delta-\delta)^3 n$$ and

$\sigma(G) > \frac{1}{\Delta+1} (\Delta-1)^3 n$

(Hamoud et al., 19 Jul 2025)

3.2. Relationship with Zagreb Indices

The $\sigma$-index admits an expression in terms of Zagreb-type indices (Hamoud et al., 22 Oct 2025, Hamoud et al., 8 Oct 2025): $$\sigma(G) = \sum_{v\in V} d(v)^3 - 2\sum_{uv\in E} d(u) d(v)$$

3.3. Hierarchies, Inequalities, and the Minkowski Norm Perspective

The $\sigma$-index fits into a larger hierarchy of $\ell_p$-norm-based irregularity measures (Lin et al., 2021):

• $A_1(G) =$ Albertson index (linear, $\ell_1$)
• $A_2(G) = \sqrt{\sigma(G)}$ (quadratic, Euclidean norm)
• $A_\infty(G) = \max_{uv\in E} |d(u) - d(v)|$ (supremal difference)

Power mean inequalities yield, for $p > q$: $A_p(G) \geq m^{1/p-1/q} A_q(G)$ and, for $p \geq 2$, $A_p(G) \leq \sqrt{\sigma(G)}$, with equality only for extremely degenerate degree distributions (Lin et al., 2021).

4. Extremal Problems and Constructions Under Degree Constraints

4.1. Trees with Prescribed Maximum Degree

For trees of fixed order $n$ and bounded maximum degree $\Delta$:

• The unique minimizer among all such trees is the "broom" (a star with a path attached to one leaf), while maximiziers exhibit only degrees in $\{1,2,\Delta\}$, with detailed cases classified according to $n\pmod{\Delta}$ (Bašić, 1 Feb 2026).
• For $\Delta = 6$, possible penalty values for extremal trees are $\{0, 10, 20, 22, 30, 40\}$, with explicit degree and edge multiplicities in each residue class.

4.2. Trees with Prescribed Degree Sequences

For any graphic sequence $d=(d_1, \ldots, d_n)$, the maximal $\sigma$ for trees is realized by a greedy attachment procedure, always pairing the largest possible degree gap, while the minimum is obtained by balancing degree differences as evenly as possible along a path (Hamoud et al., 2024, Hamoud et al., 22 Oct 2025).

4.3. Bipartite and Chemical Graphs (Degree Bounds)

Among chemical trees (with $\Delta\leq 4$), extremal analysis for the total $\sigma$-irregularity reveals unique maximizers that split vertex degrees among $\{1,2,3,4\}$ as evenly as possible, subject to realizability (Knor et al., 2024).

5. Global, Modified, and Limit Behavior

5.1. Total and Modified Sigma-Irregularity

The total $\sigma$-irregularity ($\sigma_t$) is exactly $n^2 \cdot \operatorname{Var}(G)$, thus maximizing for graphs with bidegreed split structures (Filipovski et al., 2024). Modifications such as applying non-quadratic exponents ($|d(u)-d(v)|^{f(n)}$ for $f(n)>0$) yield threshold phenomena where the antiregular graph, uniquely defined by the degree multiset $\{1,2,\dots,n-1\}$ with one repeated value, becomes the (unique) maximizer if $f(n)\leq \log_{n-2}((n^2-n-2)/(n^2-n-4))$ (Knor et al., 2024).

5.2. Inequality Relationships with the Albertson Index

Cauchy-Schwarz and related inequalities anchor the quadratic $\sigma$-index between linear and higher-norm irregularity measures (Hamoud et al., 9 Jun 2025, Hamoud et al., 22 Oct 2025): $$\sqrt{\sigma(G)} \leq \operatorname{irr}(G) \leq \sqrt{2m \sigma(G)}$$ These relationships facilitate translation between different irregularity metrics and offer a continuum of irregularity measures sensitive to degree outliers.

6. Applications and Structural Implications

6.1. Chemical and Network Graph Theory

• $\sigma(G)$ and $\sigma_t(G)$ have proven effective in QSPR/QSAR studies as molecular structural descriptors, correlating with chemical properties linked to bonding heterogeneity (Hamoud et al., 22 Oct 2025, Hamoud et al., 23 Sep 2025).
• In network theory, $\sigma(G)$ is a heterogeneity measure, with sharp lower and upper bounds providing thresholds for robustness, vulnerability, or transmission properties.

6.2. Algorithmic and Structural Utility

Closed-form expressions and greedy attachments yield efficient algorithms for computing extremal $\sigma$-values from degree sequences alone, without need for explicit graph construction (Hamoud et al., 2024, Hamoud et al., 22 Oct 2025).

6.3. Laplacian Spectral Relationships

By linking to Laplacian eigenvalues,

$\sigma(G) \leq \frac{1}{n}\lambda_n \sigma_t(G)$

where $\lambda_n$ is the largest Laplacian eigenvalue, drawing a connection between $\sigma$-irregularity and spectral graph theory (Filipovski et al., 2024).

7. Open Problems and Research Directions

Several open questions remain:

• For which (slowly vanishing) exponent functions $f(n)$ is the antiregular graph the unique maximizer of the modified global sigma-index?
• For trees of fixed order $n$ and prescribed maximal degree, full characterization of minimal and maximal $\sigma$-trees in all parameter ranges is open for $\Delta>6$ (Bašić, 1 Feb 2026).
• Explicit connections between $\sigma$-irregularity and other Zagreb indices for wider graph families are still being explored.
• Behavioral analysis of $\sigma$ under graph operations (join, product, subdivision) and extensions to directed, weighted, or multigraphs remain active topics.

---

References:

• “Extremizing antiregular graphs by modifying total σ-irregularity” (Knor et al., 2024)
• “Some results on σₜ‐irregularity” (Filipovski et al., 2024)
• “Degree Sequence of Albertson and $σ$-Indices on Trees of Order $n\geqslant 3$” (Hamoud et al., 9 Jun 2025)
• “Closed-Form Analysis and Extremal Bounds of Albertson and Sigma Indices in Trees with Prescribed Degree Sequences” (Hamoud et al., 22 Oct 2025)
• “Trees with maximum $σ$-irregularity under a prescribed maximum degree 6” (Bašić, 1 Feb 2026)
• “Extremal Bounds on the Properties of Sigma and Albertson Indices for Non-Decreasing Degree Sequences” (Hamoud et al., 23 Sep 2025)
• “Behavior of The Extremal Bounds on the $σ$-Irregularity” (Hamoud et al., 8 Oct 2025)
• “Sigma index in Trees with Given Degree Sequences” (Hamoud et al., 2024)
• “The general Albertson irregularity index of graphs” (Lin et al., 2021)