How to tame your (black hole) saddles: Lessons from the Lorentzian Gravitational Path Integral

This paper resolves the divergence of the semiclassical partition function for spherically-symmetric AdS$_4$ Einstein-Maxwell black holes by employing a Lorentzian gravitational path integral and Picard-Lefshetz analysis to demonstrate that only a finite subset of complex saddles contributes at finite temperature, while also establishing convergence for the analogous uncharged BTZ black hole case.

The Gist

The Big Picture: The Black Hole Buffet

Imagine you are trying to calculate the total "flavor" of a cosmic buffet (the partition function) that contains every possible type of black hole. In physics, to do this, we usually look for the "best" dishes (called saddles or solutions) that represent the most likely states of the black hole.

For a long time, physicists tried to find these best dishes by looking at a "Euclidean" menu. Think of Euclidean space as a frozen, static version of reality where time is just another direction of space. It's like looking at a photograph of the buffet.

The Problem:
When you try to count the dishes on this frozen menu, you run into a disaster. Because electric charge and spin are "quantized" (they come in discrete packets, like steps on a ladder), there isn't just one version of a black hole for a given setting. There are infinitely many "twisted" versions of the same black hole, shifted by invisible quantum steps.

If you try to sum up the flavor of all these infinite twisted versions, the total flavor explodes to infinity. It's like trying to add up the cost of an infinite number of items in a store; the bill becomes meaningless. The math breaks down.

The Solution: Switching to "Real-Time" Cooking

The authors, Maciej Kolanowski and Donald Marolf, propose a different way to cook the meal. Instead of looking at the frozen photograph (Euclidean), they suggest looking at the Lorentzian version.

• Euclidean (Frozen): Time is imaginary. It's like a still image.
• Lorentzian (Real-Time): Time is real. It's like a live video feed.

They argue that to properly count the black holes, we must treat them as dynamic, real-time objects that can have "kinks" or "singularities" (like a sharp fold in a piece of paper) at their centers, but otherwise behave like normal, real-time black holes.

The Magic Trick: The "Steepest Descent" Hike

To figure out which of the infinite twisted black holes actually matter, they use a mathematical tool called Picard-Lefshetz theory.

The Analogy: The Mountain Hike
Imagine the "flavor" of the black hole is a landscape of mountains and valleys.

• The Saddles are the mountain peaks (or specific high points).
• The Path Integral is a hiker trying to cross this landscape.
• The Contour is the specific trail the hiker is allowed to walk on.

In the old "frozen" approach, the hiker was allowed to walk anywhere, including up infinite mountains, causing the total energy to blow up.

In the new "real-time" approach, the rules of the hike change:

1. The Terrain: The landscape is now defined by real-time physics.
2. The Rule: The hiker can only walk along "steepest descent" paths. Imagine water flowing down a mountain; it always takes the path of steepest descent.
3. The Result: When you apply this rule to the infinite twisted black holes, you discover that for any specific temperature, most of the infinite peaks are inaccessible. The "water" (the path) simply cannot flow up to them.

Only a finite number of these twisted black holes lie on the valid hiking path. The rest are "off-limits."

The Outcome: A Tamed Sum

By restricting the calculation to only the black holes that lie on these valid "real-time" hiking paths, the infinite sum suddenly becomes a finite sum.

• Before: Infinite black holes = Infinite chaos.
• After: Only a few specific black holes count = A convergent, sensible answer.

They found that at high temperatures, only a few specific "large" black holes contribute. At low temperatures, the answer comes mostly from the "edges" of the calculation (the endpoints) rather than the peaks.

The BTZ Black Hole (The 2D Cousin)

They also tested this on a simpler, 2-dimensional black hole (the BTZ black hole). In this case, the math was even nicer: all the twisted versions contributed, but because they decayed fast enough, the sum still converged. It was like a choir where everyone sings, but the volume drops off quickly enough that the song doesn't blow out the speakers.

Why the "KSW Condition" Didn't Work

There is a popular rule in physics called the KSW condition (Kontsevich-Segal-Witten). It's like a "health inspector" that checks if a complex black hole is "safe" to include in the sum.

The authors tried to use this health inspector. They found that it was too lenient. It would say "Yes, this weird, complex black hole is safe," even when their new "real-time hiking" rules said "No, that one is off the path."

The Analogy:
Imagine the KSW condition is a map that says, "You can drive anywhere as long as the road is paved." But the authors' method is like a GPS that says, "You can only drive where the traffic laws allow." The KSW map included roads that were technically paved but legally forbidden (or physically impossible to traverse in the right way). The authors showed that relying on the KSW map leads to the wrong destination.

Summary

1. The Puzzle: Counting all possible quantum black holes leads to an infinite, nonsensical result.
2. The Fix: Stop looking at the "frozen" (Euclidean) picture. Start looking at the "live" (Lorentzian) picture with specific rules about how time flows.
3. The Mechanism: Use a mathematical "hiking rule" (Picard-Lefshetz) to see which black holes are actually reachable.
4. The Result: The infinite list of black holes shrinks to a manageable, finite list. The math works, and the universe makes sense again.

In short, the paper teaches us that to understand the quantum nature of black holes, we must stop treating them like static statues and start treating them like dynamic, real-time travelers, carefully following the path of least resistance.

Technical Summary

1. Problem Statement

The paper addresses a fundamental puzzle in the semiclassical evaluation of gravitational partition functions for black holes with quantized charges (electric charge $Q$ and angular momentum $J$).

• The Periodicity Puzzle: In the grand canonical ensemble, the partition function $Z = \text{Tr} e^{-\beta(H - \mu Q - \Omega J)}$ must be periodic in the chemical potentials $\mu$ and $\Omega$ due to the quantization of charge and spin. Specifically, the trace is invariant under shifts $\mu \to \mu + \frac{2\pi i n}{q\beta}$ and $\Omega \to \Omega + \frac{2\pi i m}{s\beta}$ (where $n, m \in \mathbb{Z}$).
• The Divergence Issue: Standard semiclassical approximations suggest summing over all saddle points (black hole solutions) that satisfy these boundary conditions. This implies summing over an infinite set of complex saddles generated by these shifts. However, in the AdS$_4$ Einstein-Maxwell theory, summing over all such complex saddles leads to a divergent result at any finite inverse temperature $\beta$. The real part of the on-shell action for these saddles is not bounded from below, causing the sum to explode.
• The Contour Ambiguity: The root of the problem lies in the lack of a rigorous definition for the gravitational path integral contour. Traditional Euclidean path integrals suffer from the "conformal factor problem" (unbounded action) and often fail to specify which complex saddles are relevant when the boundary conditions involve complex potentials.

2. Methodology

The authors resolve the divergence by shifting the framework from a purely Euclidean approach to a Lorentzian gravitational path integral with specific boundary conditions.

• Lorentzian Contour with Singularities: Instead of integrating over real Euclidean metrics, the authors define the path integral over real Lorentz-signature metrics that are allowed to contain codimension-2 "conical" (or helical) singularities. These singularities arise naturally when constructing time-periodic spacetimes by cutting and gluing wedges of stationary black hole spacetimes (analogous to the construction of Euclidean black holes but in Lorentzian signature).
• Action Definition: The action for these singular spacetimes is defined following the prescription of [24], which includes a specific imaginary term proportional to the area of the singularity ($A_\gamma$). For a singularity with $N=0$ (the relevant case for black holes), the action acquires an imaginary contribution:
  $$S = S_{\text{EH}} + i \frac{A_\gamma}{4G_N}$$
• Constrained Path Integral: The partition function is formulated as an integral over the period $T$ of the time translation and the area $A_\gamma$ of the singularity. By fixing $A_\gamma$, the authors define a constrained path integral where the saddle points correspond to stationary black holes with specific horizon areas and charges.
• Picard-Lefschetz Analysis: The core of the methodology is a rigorous Picard-Lefschetz analysis. This mathematical tool determines which complex saddle points actually contribute to the integral by analyzing the "steepest descent" (Lefschetz thimble) contours. A saddle contributes if and only if its steepest ascent contour has a non-zero intersection number with the original integration contour.

3. Key Contributions and Results

A. Resolution of the Divergence in AdS$_4$ Einstein-Maxwell

The authors apply this framework to the spherically symmetric sector of AdS$_4$ Einstein-Maxwell theory (Reissner-Nordström black holes).

• Finite Subset of Saddles: The Picard-Lefschetz analysis reveals that at any finite $\beta$, only a finite subset of the infinite family of complex saddles (labeled by integers $n$) actually contribute to the partition function. The integration contour cannot be deformed to pass through the "bad" saddles that cause divergence.
• Convergence: Because only a finite number of saddles contribute, the sum over $n$ converges. The divergence found in previous works (e.g., [28]) was an artifact of summing over irrelevant saddles that do not lie on the correct Lefschetz thimbles.
• Temperature Dependence:
  • High Temperature (Small $\beta$): Only a finite number of saddles contribute.
  • Low Temperature (Large $\beta$):
    • If $|\mu_0| \geq 1$: The number of contributing saddles grows linearly with $\beta$.
    • If $|\mu_0| < 1$: No complex saddles contribute in the limit $\beta \to \infty$. The partition function is dominated entirely by endpoint contributions (from the boundary of the integration domain, $R=0$), which correspond to the "small black hole" or trivial solutions.
• Non-Monotonicity: The relevance of specific saddles is not monotonic with temperature. Some saddles contribute only within a finite window of $\beta$, turning on and off as the temperature changes (Stokes phenomena).

B. Analysis of AdS$_3$ BTZ Black Holes

The authors perform a parallel analysis for uncharged BTZ black holes in AdS$_3$ (pure Einstein-Hilbert gravity).

• All Saddles Contribute: In this case, the analysis shows that all saddles (labeled by $m$) contribute to the partition function.
• Convergence: Despite all saddles contributing, the sum converges. This is because the action grows sufficiently fast (quadratically in the shift parameter $m$) to ensure convergence, unlike the AdS$_4$ case where the sum of all saddles diverges.

C. Critique of the KSW Condition

The paper briefly addresses the Kontsevich-Segal-Witten (KSW) criterion, a condition often used to select valid complex metrics for path integrals.

• Insufficiency: The authors demonstrate that satisfying the KSW criterion is not sufficient to guarantee a saddle contributes. They show examples of complex saddles (e.g., Schwarzschild-AdS$_4$ at low temperatures) that satisfy the KSW condition along certain contours but are proven irrelevant by the Lorentzian Picard-Lefschetz analysis.
• Coordinate Dependence: They highlight that the KSW condition is not invariant under general complex coordinate transformations, making it ambiguous in the context of complex saddles without a fixed real slice.

4. Significance

• Rigorous Definition of Gravitational Path Integrals: The paper provides a concrete, non-perturbative definition of the gravitational path integral for black holes with quantized charges, resolving long-standing issues regarding the summation over complex saddles.
• Validation of Lorentzian Approach: It strongly supports the view that the fundamental definition of the gravitational path integral should be based on real Lorentzian metrics (with singularities) rather than complex Euclidean metrics. This approach naturally handles the periodicity of chemical potentials and avoids the conformal factor problem.
• Physical Predictions: The results clarify the thermodynamic behavior of charged black holes. Specifically, the transition from saddle-point dominance to endpoint dominance at low temperatures for $|\mu| < 1$ suggests a specific structure in the density of states that differs from naive semiclassical expectations.
• Future Applications: The framework is presented as a robust tool for analyzing more complex scenarios, including rotating black holes (AdS-Kerr-Newman) and supersymmetric indices (BPS states), where the selection of relevant saddles is critical for matching holographic duals (e.g., in AdS/CFT).

In summary, the paper "tames" the infinite set of black hole saddles by rigorously defining the integration contour in the Lorentzian sector, proving that only a physically relevant, finite subset contributes at finite temperature, thereby rendering the partition function well-defined and convergent.