Stability Gap in Quantum Many-Body Theory

• Stability Gap is a property ensuring that the many-body spectral gap in lattice fermion systems remains open under weak, local perturbations.
• It is demonstrated using methods such as Majorana doubling, frustration-free representations, and quasi-adiabatic continuation to obtain explicit gap estimates.
• These rigorous results underpin the robustness of insulating, superconducting, and topological phases, even in the presence of disorder and aperiodicity.

The "stability gap" is a technical concept with distinct but analogous roles across mathematics, physics, and machine learning. In quantum many-body theory, particularly for lattice fermion systems, the stability gap denotes the persistence and quantitative lower bound of the many-body spectral gap above the Fermi-sea ground state under weak, local perturbations. Its rigorous proof requires new techniques based on Majorana fermion representations, frustration-free forms, and quasi-adiabatic continuation, ensuring that metallic or insulating states remain robust against generic, short-range interactions. The concept also interfaces with the general theory of gapped phases and the resilience of topological quantum order.

1. Problem Setting: Lattice Fermions and the Spectral Gap

Consider a finite $d$-dimensional lattice $A\subset\mathbb Z^d$ with an underlying single-particle Schrödinger operator

$(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$

where $t_{x,y}=t_{y,x}$ are exponentially decaying hopping amplitudes. The system is filled up to a Fermi energy $E_F$ lying in a spectral gap: $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$, uniform in $|A|$.

The many-body "free" Hamiltonian is

$$H_0 = \sum_{x,y\in A} a_x^*(t_{x,y} - E_F\delta_{x,y}) a_y,$$

where $a_x, a_x^*$ are fermionic creation/annihilation operators, and $|\Omega\rangle$ is the filled Fermi-sea. The many-body spectral gap $A\subset\mathbb Z^d$0 above $A\subset\mathbb Z^d$1 matches the one-body spectral gap and does not vanish as $A\subset\mathbb Z^d$2.

When a generic even, short-range perturbation $A\subset\mathbb Z^d$3 is added, with each $A\subset\mathbb Z^d$4 an even polynomial of the fermionic operators localized on $A\subset\mathbb Z^d$5 and decaying exponentially, one asks: does the gapped phase survive? Is there a nontrivial lower bound $A\subset\mathbb Z^d$6 for sufficiently small coupling $A\subset\mathbb Z^d$7? This property—the nonvanishing of the many-body excitation gap under weak interaction—is termed the "stability of the spectral gap" (Koma, 2020).

2. Stability Theorem and Main Estimates

The main result—Theorem 2.4 in (Koma, 2020)—states:

Spectral Gap Stability Theorem:

Under the assumptions of a finite, uniform one-body gap ($A\subset\mathbb Z^d$8) and exponentially decaying, even interaction terms $A\subset\mathbb Z^d$9 satisfying $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$0, there exists $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$1 such that for $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$2,

• $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$3 has a unique ground state $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$4,
• The gap above the ground state satisfies

$(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$5

with $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$6 system-size independent.

This shows any gapped many-body Fermi sea is robust, in the sense that arbitrarily weak interactions cannot close the gap provided their range is sufficiently short and their decay is sufficiently fast (Koma, 2020).

3. Proof Architecture: Majorana Formulation and Quasi-Adiabatic Continuation

The technical approach consists of five key steps:

1. Frustration-free representation via Majorana doubling: The complex fermion operators $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$7 are decomposed into Majorana operators $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$8, $(h\psi)(x) = \sum_{y\in A} t_{x,y} \psi(y),$9, which are self-adjoint, obey canonical anticommutation relations, and allow the free Hamiltonian $t_{x,y}=t_{y,x}$0 to be embedded as a sum of positive-semidefinite, local terms in an enlarged (doubled) Fock space [(Koma, 2020), Sec. 5.1]. This recasts $t_{x,y}=t_{y,x}$1 as explicitly frustration-free: each local term annihilates the ground state.
2. Quasi-adiabatic continuation: Let the perturbed Hamiltonian $t_{x,y}=t_{y,x}$2 be assumed gapped along $t_{x,y}=t_{y,x}$3. Kato’s and Hastings' quasi-adiabatic construction is used to build a differentiable family of unitaries $t_{x,y}=t_{y,x}$4 moving the ground-state projector $t_{x,y}=t_{y,x}$5 back to $t_{x,y}=t_{y,x}$6 (for the unperturbed model): $t_{x,y}=t_{y,x}$7. The conjugated Hamiltonian $t_{x,y}=t_{y,x}$8 is of the form $t_{x,y}=t_{y,x}$9, with $E_F$0 annihilating the unperturbed ground state: $E_F$1.
3. Lieb–Robinson bounds and locality: The generator of the quasi-adiabatic continuity (and hence of $E_F$2) is shown to be local with sub-exponentially decaying tails using Lieb–Robinson bounds for the evolution of local fermionic observables. Local decomposability is maintained [(Koma, 2020), Sec. 5.3].
4. Relative boundedness and explicit gap estimate: If the perturbation $E_F$3 satisfies $E_F$4 and $E_F$5 with $E_F$6, then $E_F$7 enjoys a gap bounded by $E_F$8 (Proposition 5.1 in (Koma, 2020)). The argument applies since $E_F$9 inherits locality and annihilates the ground state.
5. Bootstrap, universality, and independence of details: The procedure is self-consistent: the magnitude of $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$0 can be made arbitrarily small by taking $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$1 small enough, closing the argument by continuity. No translation invariance or periodicity is required, so the result generalizes to disordered, aperiodic, or multi-band systems.

4. Key Technical Ingredients and Lemmas

• Spectral gap for a many-body Hamiltonian: For a self-adjoint Hamiltonian $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$2 with unique ground-state projector $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$3, the spectral gap above the ground state is

$\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$4

• Relative bound and gap reduction: If $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$5 and $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$6, then

$\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$7

This is the precise mechanism by which a weak, local interaction cannot close the gap unless it becomes too large in operator norm.

• Local decomposition of evolved terms: Each $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$8 in the generalized perturbation can be decomposed into local finite-range pieces $\operatorname{dist} (\sigma(h),E_F)\geq \Delta_0>0$9 supported on a ball of radius $|A|$0 about $|A|$1, with $|A|$2 where $|A|$3 is sub-exponential.

5. Broader Context and Implications

The persistence of the many-body spectral gap under weak interactions is foundational for understanding the robustness of insulating, superconducting, and topological phases in lattice systems. Key implications include:

• Universality: The approach does not depend on translation invariance or absence of disorder; the result remains valid for irregular graphs, multi-band models, and even systems with aperiodic structure (Koma, 2020).
• No fine-tuning required: The only requirements are a positive free gap and weak, finite-range (or exponentially decaying) interactions.
• Extension to topological phases: The gap stability framework is essential for showing the robustness of topologically nontrivial ground states (e.g., quantum Hall states, symmetry-protected topological phases) under generic physical perturbations, provided the system remains gapped.

6. Relation to Related Theories and Methods

• Frustration-free formalism and Majorana doubling enable techniques analogous to those used in quantum spin systems to be ported to the CAR (canonical anticommutation relation) algebra of fermions.
• Quasi-adiabatic continuation is the main technical tool in constructing the appropriate conjugating unitaries and can be related to the general theory of spectral flow and automorphic equivalence of ground-state projectors, as developed in advanced studies of quantum phases.
• Lieb–Robinson bounds provide explicit control on the speed of propagation of information and are crucial for localizing the perturbation in the energy-space mapping (Koma, 2020).
• Relative boundedness lemmas are central in estimating how much of the original spectral gap can be "lost" to the interaction term while still guaranteeing positivity of the new gap.

7. Quantitative Summary

Property	Unperturbed Model	With Local Interaction	Reference
Ground state	Filled Fermi sea $|A|$4	$|A|$5 (unique)	(Koma, 2020)
Spectral gap	$|A|$6 (system-size indep.)	$|A|$7	(Koma, 2020)
Perturbation locality	Range $|A|$8 or $|A|$9	Maintained by Lieb–Robinson	(Koma, 2020)
Gap closure threshold	$$H_0 = \sum_{x,y\in A} a_x^*(t_{x,y} - E_F\delta_{x,y}) a_y,$$0 exists; for $$H_0 = \sum_{x,y\in A} a_x^*(t_{x,y} - E_F\delta_{x,y}) a_y,$$1 the gap remains open		(Koma, 2020)

This establishes a non-perturbative, explicit lower bound on the stability of the many-body spectral gap against generic, short-range perturbations and rigorously justifies the physical intuition that gapped Fermi seas are stable phases of matter under weak interactions. For broader quantum statistical contexts and operator-algebraic generalizations of gap stability, see also (Nachtergaele et al., 2017).