Quantum Maximin and Causal Properties

Quantum maximin converts a covariant surface-selection problem into a two-stage optimization over achronal slices. Under explicit semiclassical, existence, stability, homology, and focusing hypotheses, a stable quantum-maximin surface is the minimal-generalized-entropy quantum extremal surface. The construction then supports entanglement-wedge nesting, strong subadditivity, and causal separation. These are conditional theorems: removing the hypotheses removes the proof.

Required background. Quantum Information supplies entropy inequalities, purification, and the algebraic meaning of nested subsystems.

Helpful background. Semiclassical Gravity supplies regulated generalized entropy and quantum focusing; Quantum Fields in Curved Spacetime supplies regional field algebras; Black-Hole Geometry and Mechanics supplies achronal slices, domains of dependence, and causal wedges.

Quantum maximin data

Fix a globally hyperbolic semiclassical spacetime, a nongravitating boundary region $A$, and its anchoring and homology class. Let $\mathfrak C_A$ be the admissible complete achronal slices ending on a boundary Cauchy slice containing $A$. On each $C\in\mathfrak C_A$, minimize the same renormalized functional over homologous surfaces $X\subset C$:

$$m(C,A) := \operatorname*{arg\,min}_{X\subset C,\,X\sim A} S_{\mathrm{gen}}[X;A].$$

Then maximize the attained minimum over slices,

$$M(A) := m(C_*,A), \qquad C_*\in \operatorname*{arg\,max}_{C\in\mathfrak C_A} S_{\mathrm{gen}}[m(C,A);A].$$

One assumes that these extrema exist, that the selected surface is stable under small changes of $C$, and that ties or null segments are treated by the stability prescription. The regulated bulk algebra and counterterms must be compared in one scheme. Generalized entropy and quantum expansion are defined in Akers, Engelhardt, Penington, and Usatyuk 2020, §2.1; quantum maximin, stability, and the required null-evolved representatives are defined in their §3.2. Existence is an assumption here, not a general existence theorem.

Equivalence to the minimal QES

Theorem, in its controlled scope. Assume the data above; assume perturbative semiclassical gravity, stable quantum maximin, and the quantum focusing or replacement conditions invoked by the comparison arguments. Then $M(A)$ is a quantum extremal surface and agrees with the minimal-generalized-entropy QES in its admissible class to the retained order. The proof is developed in Akers et al. 2020, §§3.1–3.4.

The proof architecture has two parts. Because $M(A)$ minimizes on $C_*$, its first variation tangent to $C_*$ vanishes. Because $C_*$ maximizes the minimized value, stability lets one vary the slice in the independent normal direction and obtain a second vanishing variation. Together the two variations give $\delta S_{\mathrm{gen}}/\delta X^a=0$ for both normals: $M(A)$ is a QES.

For minimality, let $X$ be a competing QES and let $\widetilde X$ be its null-evolved representative on $C_*$. Quantum focusing bounds the representative’s entropy from above by that of $X$, while on-slice minimization bounds it from below by that of $M(A)$:

$$S_{\mathrm{gen}}[M(A)] \le S_{\mathrm{gen}}[\widetilde X] \le S_{\mathrm{gen}}[X].$$

The first inequality uses minimization; the second uses focusing. This is the comparison in Akers et al. 2020, §3.4, Theorem 3 and Eq. (3.6). Reversing the second sign would destroy the proof. The same homology class, entropy side, and regulator must be used throughout.

This also shows why “maximin” is not merely a numerical recipe. Its slice variation supplies the missing stationarity direction, and its min–max ordering supplies the global comparison.

Nesting, entropy inequalities, and causality

Let $A\subset B$ be nested boundary regions. Under the quantum-maximin hypotheses, representatives of $M(A)$ and $M(B)$ can be compared on a common achronal slice. Their minimality and the cut-and-paste entropy inequality imply entanglement-wedge nesting,

$$\operatorname{EW}(A)\subseteq\operatorname{EW}(B).$$

The same common-slice construction combines geometric recombination with strong subadditivity of the regulated bulk entropy. It yields the corresponding strong-subadditivity inequality for the boundary generalized entropies. The theorem chain and its assumptions are stated in Akers et al. 2020, §4.

Causal conclusions require the focusing input. Under the quantum focusing conjecture, or the more limited entropy conditions used in a particular result, the QES lies outside the relevant causal influence of the boundary region’s domain of dependence; equivalently, the causal wedge is contained in the entanglement wedge. The quantum focusing conjecture itself is formulated in Bousso, Fisher, Leichenauer, and Wall 2016, §§2–4. It is an assumption here, not a theorem of arbitrary quantum gravity.

Hybrid systems with a nongravitating bath require a modified comparison in which the bath region is held fixed while the gravitating representative moves. The applicable extension is treated in Akers et al. 2020, §5; it should not be replaced by an unqualified statement about a fully gravitating tensor factor.

Independent checks and countercases

Classical limit. Setting the bulk-entropy shape variations to zero reduces quantum maximin to classical maximin. The resulting area theorem and strong-subadditivity argument agree with Wall 2014, §§2–4.

Nested-region check. If $A=B$, nesting becomes equality. If the state and geometry factor across two disconnected components, the common-slice cut-and-paste inequality reduces to ordinary bulk strong subadditivity plus additive area terms.

Failure when hypotheses are removed. If global hyperbolicity fails, an admissible family of complete achronal slices may not exist. If stability fails at a degenerate branch, varying the maximizing slice need not give a differentiable stationary surface. If quantum focusing is removed, the sign needed to compare null-evolved representatives is unavailable, so causal-wedge containment does not follow. These are failures of the argument, not demonstrations that the opposite conclusion always holds.

Topology change is another boundary. The perturbative theorem compares representatives in the declared class; it does not justify carrying an inequality through an uncontrolled change in spacetime topology or replica saddle.

Exercises

Consider a toy family of slices labeled by $u$, with on-slice surface coordinate $x$ and generalized entropy $S(u,x)=S_0+x^2-u^2$. Compute the maximin point. Then replace $x^2$ by $0$ and explain which theorem hypothesis becomes suspect.

Solution

For fixed $u$, the minimum is at $x=0$ with value $S_0-u^2$. Maximizing over $u$ gives $u=0$, hence $(u,x)=(0,0)$. Both first derivatives vanish, illustrating how the minimum and maximum supply independent stationarity directions. After replacing $x^2$ by zero, every $x$ minimizes on every slice: the on-slice Hessian has a zero mode and the selected surface is not unique. This alone does not prove instability. Indeed, every $(0,x)$ is still stationary in this exact toy function. What fails is the nondegenerate, unique-branch argument; the stability conditions for a degenerate maximin must instead be checked. The source explicitly treats stable degenerate surfaces in §3.4, Theorem 2.

References

• Akers et al. 2020, Chris Akers, Netta Engelhardt, Geoffrey Penington, and Mykhaylo Usatyuk, “Quantum Maximin Surfaces,” Journal of High Energy Physics 08 (2020) 140.
• Bousso et al. 2016, Raphael Bousso, Zachary Fisher, Stefan Leichenauer, and Aron C. Wall, “Quantum Focusing Conjecture,” Physical Review D 93 (2016) 064044.
• Wall 2014, Aron C. Wall, “Maximin Surfaces, and the Strong Subadditivity of the Covariant Holographic Entanglement Entropy,” Classical and Quantum Gravity 31 (2014) 225007.