Second quantization and compactness of induced Gaussian quantum Markov semigroups

Technical Summary: Second Quantization and Compactness of Induced Gaussian Quantum Markov Semigroups

Problem Statement
The paper investigates the compactness properties of Gaussian Quantum Markov Semigroups (QMSs) acting on the Hilbert–Schmidt space $B_2(\Gamma(\mathbb{C}^d))$. Specifically, it addresses the behavior of "induced" semigroups $(T^{(s)}_t)_{t \ge 0}$ generated by a Gaussian QMS $(T_t)_{t \ge 0}$ with a faithful normal invariant state $\rho$. These induced semigroups arise from $s$-embeddings ($s \in [0,1]$), which generalize the GNS ($s=0$), KMS ($s=1/2$), and anti-GNS ($s=1$) representations. The central problem is to characterize when these induced operators are compact, possess a compact resolvent, or belong to the trace class, and to determine the relationship between these properties, the stability of the underlying drift matrix, and the existence of coercive spectral gaps.

Methodology
The primary methodological tool is a unitary second-quantization representation for every $s$-embedding. The author constructs a unitary operator $U_s: \Gamma(\mathbb{C}^{2d}) \to B_2(\Gamma(\mathbb{C}^d))$ that maps the induced semigroup to a second-quantized contraction semigroup on an auxiliary Fock space:
$$U_s^* T^{(s)}_t U_s = \Gamma(e^{tZ_s})$$
where $\Gamma$ denotes the symmetric second quantization, and $Z_s$ is a transformed drift matrix defined by $Z_s = S_s^{1/2} Z S_s^{-1/2}$. Here, $Z$ is the real drift matrix of the original QMS, and $S_s$ is the $s$-weighted covariance matrix associated with the invariant state $\rho$, constructed via the weighted geometric mean of the covariance matrix $S$ and the symplectic form $J$.

Key steps in the methodology include:

1. Reduction to Finite Dimensions: By establishing the unitary equivalence, the infinite-dimensional compactness problem on the Hilbert–Schmidt space is reduced to the spectral properties of the finite-dimensional drift matrix $Z$ and its transformed version $Z_s$.
2. Handling Non-Gaussian States: For cases where the invariant state $\rho$ is not Gaussian, the paper utilizes an invariant-state decomposition (referencing [GP26]) to identify a faithful Gaussian reference state $\sigma$ that induces the same semigroup on the Hilbert–Schmidt space, allowing the second-quantization framework to apply generally.
3. Spectral Analysis: The generator $L^{(s)}$ of the induced semigroup is identified with the differential second quantization $\lambda(Z_s)$. The spectrum and algebraic multiplicities are derived directly from the eigenvalues of the one-particle matrix $Z_s$ acting on symmetric tensor powers.

Key Contributions and Results

1. Equivalence of Compactness Conditions:
   The main theorem (Theorem 4.1) establishes that for a finite-mode Gaussian QMS with a faithful normal invariant state, the following conditions are equivalent:

   • The drift matrix $Z$ is stable (all eigenvalues have strictly negative real parts).
   • The QMS admits a unique normal invariant state.
   • For some (and thus every) $s \in [0,1]$, the induced generator $L^{(s)}$ has a compact resolvent.
   • For some (and thus every) $s \in [0,1]$ and $t > 0$, the operator $T^{(s)}_t$ is compact.
   • For every $s$ and $t > 0$, $T^{(s)}_t$ belongs to the trace class $S_1(H)$.

   A crucial finding is that compactness for a single embedding at a single positive time implies immediate trace-class membership for all embeddings.

2. Exact Norm and Trace Formulas:
   Under the stability condition, the paper derives exact formulas for the trace norm and Hilbert–Schmidt norm of the induced semigroups:

   • $\|T^{(s)}_t\|_{S_1} = \det(1_{2d} - |e^{tZ_s}|)^{-1}$
   • $\|T^{(s)}_t\|_{S_2}^2 = \det(1_{2d} - (e^{tZ_s})^* e^{tZ_s})^{-1}$
   • $\text{tr}_H(T^{(s)}_t) = \det(1_{2d} - e^{tZ})^{-1}$
     Notably, the trace identity is independent of the embedding parameter $s$.

3. Spectral Characterization:
   The paper provides a complete description of the spectrum $\sigma(L^{(s)})$ and its algebraic multiplicities. The spectrum consists of all linear combinations $\sum_{j=1}^{2d} \alpha_j \lambda_j$ where $\lambda_j$ are the eigenvalues of the drift matrix $Z$ and $\alpha \in \mathbb{N}^{2d}$. The algebraic multiplicity of a spectral value is the number of such combinations yielding that value. This result holds even if $Z$ is not diagonalizable.

4. Decoupling Compactness from Spectral Gaps:
   The paper distinguishes between compactness and the existence of a coercive spectral gap (defined via the numerical range on the orthogonal complement of the kernel).

   • Compactness $\nRightarrow$ Spectral Gap: Example 5.3 demonstrates a system with a stable drift (hence compact resolvent and trace-class membership) where the coercive spectral gap vanishes for all embeddings due to the presence of a non-trivial kernel in the symmetric part of the drift.
   • Spectral Gap $\nRightarrow$ Compactness: Example 5.4 shows that if the invariant state is not unique (leading to an infinite-dimensional kernel), a positive spectral gap on the orthogonal complement of the kernel does not imply a compact resolvent.

Significance and Claims
The paper claims to provide a unified representation for the full family of $s$-embeddings, extending previous work that focused on specific cases (like GNS or KMS) or specific dynamics. By reducing the compactness of the induced semigroups to the stability of the finite-dimensional drift, the work offers a clear, algebraic criterion that does not require assumptions on the diagonalizability of the drift or the strict positivity of its dissipative part.

The author emphasizes that their results clarify the relationship between compactness and spectral gaps, showing that immediate compactness can occur even when coercive spectral gaps vanish, and that spectral gaps do not guarantee compactness in the absence of a unique invariant state. The work complements existing analyses (such as those in [FL25]) by providing a structural proof for the spectral formula applicable to any finite number of modes, independent of the specific embedding parameter.

The paper does not propose new experimental applications or future physical implications beyond the mathematical characterization of these semigroups. Its scope is strictly confined to the functional analytic properties of Gaussian QMSs on Hilbert–Schmidt spaces.