Phase-Space Nuclearity and Compactness Maps

Phase-space conditions test whether bounded local operations create a controllable family of states below a given energy scale. Haag–Swieca compactness asks for relative compactness of an energy-cutoff image; Buchholz–Wichmann nuclearity asks for a trace-class-like decomposition of an energy-damped localization map and, crucially, a quantitative norm bound. Nuclearity implies compactness, but neither locality nor positive energy implies nuclearity. The free-scalar estimate feeds the treatment of infinite-volume KMS states, passivity, and phase multiplicity.

Required background. Particles, mass-shell spectrum, and one-particle subspaces supplies the free-particle phase space; quasilocal C*-algebras and inductive limits supplies local operator balls; and representation types, factors, and local-algebra structure supplies the vacuum representation.

Helpful background. Trace ideals and Fredholm determinants supplies nuclear decompositions and determinant bounds, while nuclearity, phase-space bounds, and split distance develops their informational consequences.

Compactness and nuclearity are different conditions

Let $H\geq0$ be the vacuum Hamiltonian, $P_E=\mathbf1_{[0,E]}(H)$, and $\mathcal A(\mathcal O)_1$ the operator-norm unit ball. The Haag–Swieca map is

$$\Phi_{E,\mathcal O}: \mathcal A(\mathcal O)\longrightarrow\mathcal H, \qquad \Phi_{E,\mathcal O}(A)=P_EA\Omega.$$

Compactness means that $\Phi_{E,\mathcal O}(\mathcal A(\mathcal O)_1)$ is relatively compact. It excludes an infinite orthogonal family of uniformly localized, uniformly bounded-energy excitations. Haag and Swieca introduced this criterion and displayed generalized-free-field failures in Haag and Swieca 1965, pp. 308–320.

Energy nuclearity uses the smoother map

$$\Theta_{\beta,\mathcal O}(A) =e^{-\beta H}A\Omega, \qquad \beta>0.$$

This is a bounded linear map from the Banach space $\mathcal A(\mathcal O)$ to $\mathcal H$. It is nuclear if

$$\Theta_{\beta,\mathcal O}(A) =\sum_{n=1}^{\infty}\varphi_n(A)\xi_n, \qquad \sum_n\|\varphi_n\|\,\|\xi_n\|<\infty,$$

and its nuclear norm is the infimum of the displayed sum. Every nuclear map is compact. The converse fails for Banach-space maps: singular values can tend to zero without being summable. Moreover, the mere finiteness of $\|\Theta_{\beta,\mathcal O}\|_1$ at isolated values does not give the uniform small-$\beta$ or geometric control used in thermodynamic and split theorems.

What the norm measures

The factor $e^{-\beta H}$ suppresses high energy while $A\in\mathcal A(\mathcal O)_1$ restricts preparation to $\mathcal O$. Thus $\beta$ is a resolution scale, not automatically the inverse temperature of an existing equilibrium state. A typical local phase-space bound has the form

$$\log\|\Theta_{\beta,\mathcal O_R}\|_1 \leq C\left(1+\frac{R}{\beta}\right)^d o(\beta^{-d})$$

for a region of spatial size $R$ in $d$ spatial dimensions, with the precise constants, exponent, and range stated by the model theorem. Such a bound expresses local volume growth rather than a finite-dimensional local Hilbert space.

For a free bosonic field, localization and energy damping first define trace-class or sufficiently summable one-particle operators. Second quantization turns their singular values $s_j$ into a Fock-space bound controlled schematically by

$$\prod_j(1-s_j)^{-1}, \qquad \log\prod_j(1-s_j)^{-1} =\sum_j-\log(1-s_j).$$

This determinant mechanism is why summability, not merely $s_j\to0$, matters. It also makes the dependence on the number of species explicit. Buchholz and Wichmann formulate the energy nuclearity condition, derive its causal-independence consequences, and verify it for free fields in Buchholz and Wichmann 1986, pp. 321–344.

Massive free-scalar estimate

For the massive scalar in $3+1$ dimensions, the one-particle Hamiltonian is $\omega(\mathbf p)=\sqrt{\mathbf p^2+m^2}$. A double-cone localization estimate combined with $e^{-\beta\omega}$ produces summable one-particle localization-damping operators. Lifting the estimate to symmetric Fock space gives, for fixed $m>0$ and a double cone of radius $R$, a coarse bound of the form

$$\log\|\Theta_{\beta,\mathcal O_R}\|_1 \leq C_m\left(1+\frac{R}{\beta}\right)^3 C_m'$$

over the stated small-$\beta$ regime. At large $\beta$, the mass suppresses nonvacuum contributions. The constants are not universal observables: they depend on the localization estimate and on how the enclosing region is chosen. The robust conclusion is nuclearity with local three-dimensional high-temperature growth.

An independent dimensional check uses the one-particle integral

$$R^3\int\frac{\mathrm d^3\mathbf p}{(2\pi)^3} e^{-\beta\sqrt{\mathbf p^2+m^2}}.$$

For $\beta m\ll1$, rescaling $\mathbf q=\beta\mathbf p$ gives order $R^3\beta^{-3}$. For $\beta m\gg1$, the integral is exponentially suppressed by $e^{-\beta m}$. This reproduces the two qualitative regimes without claiming the exact nuclear norm.

Adversarial test: proliferating species

Take a tensor product of mutually independent free scalar species with masses $m_j$ and degeneracies $g_j$. Locality and positive energy hold species by species. The logarithm of the Fock nuclearity bound adds species contributions, whose low-temperature behavior contains a sum comparable to

$$\sum_j g_j e^{-\beta m_j}.$$

Choose $g_j$ to grow so rapidly that this sum diverges for some or every $\beta>0$. Then the local energy-damped map is not nuclear, and with still faster proliferation even compactness can fail. The spectrum condition did not control species density.

The strongest surviving statements are locality and positive energy. Nuclearity also does not imply a mass gap: massless theories can satisfy suitably formulated nuclearity bounds. Conversely, a mass gap alone does not bound the number of species.

Exercises

Let a compact operator have singular values $s_n=1/n$. Why does this model the distinction between compactness and nuclearity?

Solution

Since $s_n\to0$, the operator is compact. Its trace norm would be $\sum_ns_n=\sum_n1/n$, which diverges, so it is not trace class. Nuclear maps between Hilbert spaces coincide with trace-class operators. The QFT maps have a Banach-space domain, but the same lesson survives: relative compactness does not supply a summable nuclear decomposition or its quantitative norm.

References

• Buchholz, Detlev, and Eyvind H. Wichmann. 1986. “Causal Independence and the Energy-Level Density of States in Local Quantum Field Theory.” Communications in Mathematical Physics 106: 321–344. DOI.
• Haag, Rudolf, and Jorge A. Swieca. 1965. “When Does a Quantum Field Theory Describe Particles?” Communications in Mathematical Physics 1: 308–320. DOI.