Inverse Perron–Frobenius: Hilbert Cone Ergodicity

• Inverse Perron–Frobenius theorem is a spectral result stating that a bounded positive self-adjoint operator with a simple maximal eigenvalue admits a Hilbert cone ensuring ergodicity.
• It constructs an axis cone using the unique eigenvector, such that the operator becomes positivity improving through iterative applications.
• The theorem underpins rigorous applications in quantum physics and operator algebras by restoring ergodicity in situations where standard positivity is disrupted.

The inverse Perron–Frobenius theorem addresses the problem of determining, for a given bounded positive self-adjoint operator—particularly in infinite-dimensional real Hilbert spaces—whether it is possible to construct a Hilbert cone with respect to which the operator is ergodic. In contrast to the classical (direct) Perron–Frobenius theory, where a fixed cone is specified and ergodicity is analyzed, the inverse theorem investigates the reverse implication: does the spectral simplicity of the operator's maximal eigenvalue guarantee the existence of a cone rendering the operator ergodic? This construction is central to spectral theory, positivity in operator algebras, and the study of ground-state uniqueness in infinite-dimensional analysis (Tomioka, 2024). Related results in Banach lattice theory and finite-dimensional matrix theory provide foundational context and comparison (Gao, 2012).

1. Background: Perron–Frobenius Theory and Cones

Let $\mathcal H$ be a real Hilbert space with inner product $(\cdot,\cdot)$ and norm $\|u\| = (u,u)^{1/2}$. A Hilbert cone $K \subset \mathcal H$ is a nonempty, closed, convex cone that is also self-dual:

• $K + K \subset K$,
• $\alpha K \subset K$ for all $\alpha \ge 0$,
• $K \cap (-K) = \{0\}$,
• $(u, v) \ge 0$ for all $u, v \in K$,
• every $w \in \mathcal H$ can be written as $w = u - v$, $u, v \in K$, $(u, v) = 0$,
• $K = K^* = \{ x \in \mathcal H \mid (x, u) \ge 0$ for all $u \in K$ $\}$.

A bounded self-adjoint operator $A \in B(\mathcal H)$ is called positive if $A(K) \subset K$ and positivity improving if $A(K \setminus \{0\}) \subset K^\circ$, the interior of $K$. For such an operator, ergodicity with respect to $K$ means for any $u, v \in K \setminus \{0\}$, there exists $n \in \mathbb N$ such that $(u, A^n v) > 0$. The classical Perron–Frobenius theorem asserts that for a fixed cone $K$, $A$ is ergodic w.r.t. $K$ if and only if $\|A\|$ is a simple eigenvalue with strictly positive eigenvector in $K^\circ$ (Tomioka, 2024).

2. Statement of the Inverse Perron–Frobenius Theorem

The inverse problem is: given a bounded positive self-adjoint operator $A$ on $\mathcal H$ with simple maximal eigenvalue $\|A\|$, does there exist a cone $K$ such that $A$ is positivity improving (thus ergodic) with respect to $K$?

Theorem (Miyao–Tomioka (Tomioka, 2024), Theorem 3.4):

Let $\mathcal H$ be a real Hilbert space and $A$ a bounded positive self-adjoint operator. Suppose $\|A\|$ is a simple eigenvalue, $u_0 \in \mathcal H$ the unique (up to sign) unit eigenvector: $(A - \|A\| I)u_0=0$, $\|u_0\|=1$. Then there exists a Hilbert cone $K_A \subset \mathcal H$, depending on $u_0$, such that $A$ is positivity improving (hence ergodic) with respect to $K_A$.

This establishes the equivalence: simplicity of the top eigenvalue $\iff$ ergodicity with respect to some cone.

3. Explicit Construction of the Hilbert Cone

Given $u_0$, define the axis cone

$$P(u_0) = \left\{ u \in \mathcal H \;\middle|\; (u_0,u) \geq \|u - (u_0,u)u_0\| \right\}.$$

This is the set of $u$ such that the component along $u_0$ dominates the orthogonal component. $P(u_0)$ is closed, convex, self-dual ($P(u_0) = P(u_0)^*$), and its interior consists of those $u$ for which strict inequality holds. Vectors in $P(u_0) \setminus \{0\}$ with $(u_0,u) > \|u - (u_0,u)u_0\|$ are strictly positive with respect to the cone.

Key properties:

• $P(u_0)\cap(-P(u_0)) = \{0\}$,
• Orthogonality and domination along $u_0$,
• Variational characterization in terms of the axis $u_0$,
• Interior equals the set of strictly positive vectors,
• For every $w \in \mathcal H$, can orthogonally decompose $w = u-v$ with $u, v \in P(u_0)$, $(u,v)=0$.

4. Proof Outline and Core Mechanism

The proof proceeds as follows:

• By the spectral theorem, simplicity of $\|A\|$ with normalized eigenvector $u_0$ allows defining $P(u_0)$ as above.
• For $u \in P(u_0)\setminus\{0\}$, write the orthogonal decomposition $u = (u_0,u)u_0 + u_1$, $u_1 \perp u_0$.
• Compute the key inequality:

$$(u_0, Au) = (A u, u_0) = \|A\|(u_0, u) > \|A\|\|u_1\|,$$

with strictness due to the simplicity of $\|A\|$.

• This shows $Au$ lies in the interior $P(u_0)^\circ$, thus $A$ is positivity improving with respect to $P(u_0)$.
• Positivity improvement implies ergodicity: $A^n$ is positivity improving for all $n$, so for any $u,v \in P(u_0)\setminus\{0\}$ there exists $n$ with $(u, A^n v) > 0$.

The critical mechanism is the collapse of all directions onto the dominant eigenvector $u_0$ due to simplicity, enforcing strict positivity in iterations.

5. Application to the Magnetic Schrödinger Operator

A representative application is to the heat semigroup generated by the magnetic Schrödinger operator on $L^2(\mathbb R^d)$ with even real-valued potentials and vector potentials:

$H(e) = (-i\nabla + e\,a(x))^2 + V(x),$

where $H_0 = -\Delta + V$ has a simple ground-state eigenvalue with normalized eigenvector $y(x) > 0$. For small $|e|$, magnetic perturbations break standard pointwise positivity, but via the constructed axis cone $P(y) \subset L^2(\mathbb R^d)$, the perturbed heat semigroup $\{e^{-sH(e)}\}_{s>0}$ is ergodic with respect to $P(y)$ for $|e|<e_0$. Thus, strict positivity and ergodicity are recovered under algebraic perturbation by shifting the positivity structure to the axis cone determined by the ground state (Tomioka, 2024).

6. Comparison with Banach Lattice and Finite-Dimensional Results

In Banach lattice settings, inverse Perron–Frobenius/comparison theorems have been developed with related, but distinct, hypotheses. For positive bounded operators $T, S \geq 0$ on a Banach lattice $X$:

• If $0 \le S \le T$, $r(T) = r(S)$, and at least one of $T$ or $S$ is irreducible and power-compact (or its spectral radius is a resolvent pole), then $T = S$ (Gao, 2012).
• The underlying mechanism relies on the spectral projection at $r(T)$ being rank 1, strict positivity of eigenvectors/eigenfunctionals, and application of the Kreĭn–Rutman theorem.
• In finite dimensions, for nonnegative irreducible matrices $A$, the simplicity of the Perron root allows, after change of basis, recovery of strict positivity in the sense of the dominant eigenvector being the new positive axis.

This demonstrates the broader compatibility and analogies between Hilbert space, Banach lattice, and classical matrix frameworks, with the Miyao–Tomioka theorem (Tomioka, 2024) supplying the infinite-dimensional Hilbert space analogue.

7. Significance and Further Directions

The inverse Perron–Frobenius theorem provides a spectral criterion (simplicity of the top eigenvalue) for reconstructing a positivity structure tailored to the operator, ensuring ergodicity—this is especially impactful in infinite dimensions where positivity cones are not unique or a priori fixed. Applications span quantum statistical mechanics, perturbed diffusion semigroups, and spectral theory, wherever positivity is structurally disrupted but spectral simplicity remains. A plausible implication is that ergodicity and uniqueness of ground states can be systematically restored in systems lacking standard pointwise or algebraic positivity by axis-cone constructions. Potential research directions include the exploration of analogues in other operator-theoretic contexts and further extension to non-self-adjoint cases or non-Hilbert settings (Tomioka, 2024, Gao, 2012).