p-Laplacian Δₚ: Nonlinear Analysis & Applications

• p-Laplacian Δₚ is a nonlinear, degenerate elliptic operator defined via a divergence structure that generalizes the classical Laplacian.
• It exhibits distinct variational, spectral, and geometric properties with singular behavior for 1<p<2 and degeneracy for p>2.
• Applications span nonlinear PDEs, discrete graph analysis, and image processing, supported by advanced numerical and variational methods.

The $p$-Laplacian $Δ_p$ is a nonlinear, degenerate elliptic operator defined for $1 < p < ∞$ via the divergence structure $Δ_p u := ∇·(|∇u|^{p-2} ∇u)$, with $u$ typically a function on a domain $\Omega \subset \mathbb{R}^n$ or on discrete/combinatorial structures such as graphs and hypergraphs. It generalizes the classical Laplacian ($p=2$), and exhibits distinct analytic, geometric, and variational properties governed by the parameter $p$. The $p$-Laplacian is central in nonlinear potential theory, nonlinear PDE, calculus of variations, geometric analysis, discrete mathematics, and spectral theory.

1. Classical Definition, Variational Structure, and Generalizations

For $u: \Omega \rightarrow \mathbb{R}$, the classical quasilinear divergence form is given by

$Δ_p u = ∇ · (|\nabla u|^{p-2} \nabla u)$

with $p>1$. The $p$-Laplacian is the Euler–Lagrange operator of the $p$-Dirichlet energy

$J_p[u] = \int_\Omega \frac{1}{p} |\nabla u|^p\,dx$

A function $u$ is $p$-harmonic (solution to $Δ_p u = 0$) if it is a critical point of $J_p$ under Dirichlet boundary conditions. Formal limits as $p\to2$ recovers the linear Laplacian, and as $p\to\infty$, one formally arrives at the $\infty$-Laplacian

$$Δ_\infty u = \sum_{i,j=1}^n u_{x_i x_j} u_{x_i} u_{x_j}$$

Normalized variants, such as the game-theoretic $p$-Laplacian

$Δ_p^N u = \frac{1}{p} |\nabla u|^{2-p} Δ_p u$

are uniformly elliptic for $p \in (1, ∞)$ but are non-divergence type (Kawohl et al., 2016).

2. Degeneracy, Ellipticity, and Viscosity Theory

The $p$-Laplacian is neither uniformly elliptic nor comparable to the Laplacian unless $p=2$. For $1 < p < 2$, the operator is singular at $|\nabla u|=0$ (coefficient blows up), while for $p>2$ it becomes degenerate (coefficient vanishes) at critical points. As a consequence, classical $C^2$ solutions may not exist; one works with weak solutions in $W^{1,p}$ or viscosity solutions. The viscosity definition for $p$-superharmonicity is as follows: $u$ is a viscosity supersolution of $Δ_p u=0$ if for every test function $\phi$ touching $u$ from below at $x_0$, $Δ_p \phi(x_0) \leq 0$ (Brustad, 2017, Teso et al., 2020).

3. Nonlinear Spectral Theory and Eigenvalue Problems

The Dirichlet eigenvalue problem is

$$-Δ_p u = λ |u|^{p-2} u \quad \text{in} \; \Omega,\; u=0 \;\text{on}\; \partial\Omega$$

with the first eigenvalue characterized variationally by

$$λ_{1,p}^p = \min_{u \in W_0^{1,p}(\Omega) \setminus \{0\}} \frac{\int_\Omega |\nabla u|^p}{\int_\Omega |u|^p}$$

There is uniqueness and positivity of eigenfunctions for $1 < p < ∞$ (Kawohl et al., 2016). As $p \to \infty$, $λ_{1,p} \to 1/R(\Omega)$, where $R(\Omega)$ is the inradius. Extensions to Finsler manifolds yield upper and lower bounds in terms of Busemann–Hausdorff volume and reversibility constant (Combete et al., 2017). Cheeger-type inequalities of the form $λ_{1,p} \geq (h/p)^p$ remain valid, where $h$ is an appropriate Cheeger constant. Nonlinear spectral theory, including decomposition and filtering, has been developed for $p$-Laplacian with $1Cohen et al., 2019).

4. Discrete, Graph, and Hypergraph $p$-Laplacians

On graphs and hypergraphs, the $p$-Laplacian is defined for $u: V \to \mathbb{R}$ by

$$Δ_p u(x) = \frac{1}{μ(x)} \sum_{y \sim x} w(x,y) |u(y) - u(x)|^{p-2} (u(y) - u(x))$$

where $w$ is the edge weight and $μ$ the vertex measure (Hu, 22 Jan 2026). The discrete $p$-Laplacian is the variational derivative of the energy $$E_p(u) = \frac{1}{p} \sum_{x,y} w(x,y) |u(y)-u(x)|^p$$ (Zhao, 2 Dec 2025). Hypergraph versions include vertex and hyperedge $p$-Laplacians, with spectral theory based on Rayleigh quotients and min–max principles (Jost et al., 2020). The Nehari-manifold method can be employed in these discrete settings to construct ground states and prove multiplicity under growth and monotonicity conditions (Zhao, 2 Dec 2025).

5. Mean Value Formulas, Superposition Principles, and the Dominative $p$-Laplacian

The $p$-Laplacian admits asymptotic mean value formulas, valid in the viscosity sense:

$I_r[u](x) = \Delta_p u(x) + o(1)$

with explicit constants, and for certain $p$ in the plane, pointwise identities hold (Teso et al., 2020). The Dominative $p$-Laplacian operator $D_p$ provides a linear upper envelope to $Δ_p$:

$Δ_p u \leq |\nabla u|^{p-2} D_p u$

$D_p$ is sublinear: sums of dominative $p$-superharmonic functions remain dominative $p$-superharmonic, explaining a superposition principle for $p$-Laplace fundamental solutions and their combinations (Brustad, 2017).

6. Extensions: $p$-Laplacian with Measures, Fractional Operators, and Curvature Notions

For positive finite Borel measures $\mu$ satisfying an Adams-type embedding, the $p$-Laplacian is defined via duality:

$$\int_\Omega |\nabla u|^{p-2} \nabla u \cdot \nabla \phi \, dx = \int_\Omega \phi f \, d\mu$$

Weak solutions exist uniquely via monotone operator theory, and eigenvalue problems admit existence/minimization principles (Tuhola-Kujanpää et al., 2011). Fractal and singular measures are admissible if suitable ball growth holds.

Fractional $p$-Laplacians on hyperbolic spaces have been defined via three equivalent formulations: nonlinear Bochner semigroup, singular integral kernel (involving hyperbolic heat kernel and modified Bessel functions), and nonlinear extension problems. Explicit normalizing constants ensure correct convergence to $Δ_p$ as $s \to 1^-$, with geometric implications for heat kernels and manifold curvature (Kim et al., 2022).

For graphs, the $CD_p(m,K)$ curvature-dimension condition involves the iterated $\Gamma_{2,p}$ calculus and is a nonlinear Bakry–Émery-type extension. The property of curvature being preserved under Cartesian product fails for $p>2$, in contrast to the linear ($p=2$) theory (Hu, 22 Jan 2026).

7. Computational Methods and Applications

Numerical solution of $Δ_p$-type equations is challenging due to nonlinearity and degeneracy. Barrier–Newton interior-point algorithms enable uniform polynomial-time solution of discretized $p$-Laplacian problems for all $p \in [1,\infty]$. Complexity is $O(n \log n)$ Newton iterations, with robust performance even for degenerate limits $p=1$ (total variation/minimal surface) and $p=\infty$ (least gradient) (Loisel, 2020). Applications encompass nonlinear Darcy flow, sandpile models, minimal-surface computations, and image-processing diffusion.

---

The $p$-Laplacian $Δ_p$ encompasses a family of nonlinear, degenerate elliptic operators with rich analytic, spectral, and geometric behaviors across continuous, discrete, geometric, and probabilistic frameworks. Its nonlinear superposition, spectral theory, curvature-dimension relations, measure-theoretic and fractional extensions, and robust numerical solution methods situate it as a central object in modern nonlinear analysis, geometric PDE, discrete mathematics, and applied computational domains.