What if the equations of black hole physics told you something that, by every known rule of quantum mechanics, should be impossible? Not just strange, not just counterintuitive, but mathematically illegal?

That is exactly what happened in 2022, when Juan Maldacena and colleagues at Princeton discovered a disturbing puzzle buried in the mathematics of two-sided black holes. The entanglement entropy — a quantity that in ordinary quantum systems must always be positive — could, under certain conditions, turn negative. It was as if a bank balance suddenly dipped below zero with no explanation, a violation of a rule so fundamental that most physicists had never thought to question it.

Now, a team led by Patrick Tran at the University of California, Berkeley — working with Stefano Antonini, Luca V. Iliesiu, and Pratik Rath — claims to have solved the mystery. Their preprint (arXiv:2509.15295) proposes that the apparent negativity was never a paradox at all, but a signal that something was missing from the calculation: contributions so tiny they are normally ignored, yet just large enough to rescue physics from its own contradiction.

The rule that seemed unbreakable

To understand why negative entropy caused a crisis, we first need to understand what entropy means in this context. The entropy we encounter in thermodynamics — the disorder of a gas, the spreading of heat — is just the beginning. Black hole physicists often work with something more refined called the Rényi entropy, a family of measures that probe how quantum information is shared between two halves of a system. The second Rényi entropy — one of several variants that can be defined — was among those that went negative. (Strictly speaking, what went negative is the so-called 'annealed' version, the simplest to compute; the physically required 'quenched' version was always positive.)

Think of it like this: imagine a sealed box divided by a wall. On one side, you have particles; on the other, nothing. The entanglement entropy measures how the particles' quantum states are entangled with the empty side — not something classical, but a purely quantum connection. In any normal system, this quantity can range from zero (no entanglement) to some positive number. It cannot be less than zero. That would be like saying two people are less than not connected at all.

The LMRS paradox — named after its discoverers Lin, Maldacena, Rozenberg, and Shan — showed that in a two-sided black hole containing a very large number of matter excitations behind the horizon, the second Rényi entropy dropped below zero. The usual mathematical machinery produced an answer that quantum mechanics says should not exist.

"The puzzle was originally presented in the context of BPS black holes in two-dimensional supergravity," the Berkeley team writes, "but the negativity persists for more general two-sided black holes." This was not an edge case that could be safely ignored. It was a sign that something fundamental was missing from the standard gravitational calculations.

The missing contributions

The standard approach to computing black hole entropies involves the gravitational path integral — a sum over all possible geometries that contribute to a given quantum process. In practice, physicists compute this sum perturbatively, adding contributions order by order, like calculating the area of a curved shape by adding smaller and smaller triangles.

The Berkeley team's insight was that the missing contributions were non-perturbative — contributions so small they cannot be captured by any finite number of perturbative terms. These are the gravitational equivalent of quantum tunneling: rare events that have no perturbative description at all.

By summing over all such non-perturbative contributions to the gravitational path integral, the team found that the entropy is rescued. The negativity disappears, replaced by a positive result consistent with quantum mechanics. This is not a patch or a fudge; it is a complete mathematical resolution within the same framework.

But the gravity calculation can only take us so far. The path integral itself is an approximation, and its regime of validity is limited. To go further, the team turned to a mathematical cousin: the matrix integral.

A dual picture emerges

Matrix integrals are powerful mathematical tools that describe the statistical behavior of large random matrices. They have become increasingly important in quantum gravity because they can capture non-perturbative effects that gravity alone cannot. The Berkeley team showed that the black hole entropy calculation has a precise dual description in a matrix integral — a map between two seemingly different problems that are secretly the same.

In this dual picture, the negative entropy problem reappears, but now the team could see exactly what was needed to fix it. The key turned out to be instanton saddles — special configurations in the matrix integral where one or two eigenvalues break away from the main distribution. These are the matrix model analogues of the non-perturbative gravitational contributions. One-eigenvalue instantons and two-eigenvalue instantons, as the team calls them, appear at precisely the point where the standard perturbative calculation would go negative, and their contributions restore positivity.

"Positivity is rescued by new saddles of the matrix integral," the authors explain. This is not a metaphysical claim about the nature of reality; it is a concrete mathematical result, verified by explicit calculation. The negative entropy was never a sign that quantum mechanics had broken down — it was a sign that we had not been summing all the relevant terms.

The meaning for black hole physics

What does this resolution tell us about black holes themselves? The implications are twofold.

First, it confirms that the standard rules of quantum mechanics — including the positivity of entropy — hold even in the extreme environment of a black hole interior. This may seem obvious, but it was not guaranteed. Black holes test the boundaries of our theories, and every time they pass a test, our confidence in the framework grows.

Second, it deepens our understanding of how the gravitational path integral works. The fact that non-perturbative contributions can resolve a paradox that perturbative calculations could not suggests that the full, non-perturbative path integral is more robust than its perturbative approximations might suggest. This is a technical insight, but it has philosophical weight: the universe, it seems, uses every tool available to maintain internal consistency.

The team also extended their analysis using random tensor network techniques — a third approach that provides yet another perspective on the same problem. Random tensor networks are computational models of entanglement that have become popular in the holographic community. By formulating the entropy puzzle within this framework, the Berkeley team showed that the same mechanism — non-perturbative contributions restoring positivity — operates there as well. Three different mathematical languages, one consistent answer.

Beyond the paradox

This is not a story about a bug in the code of physics that has now been patched. It is a story about how theoretical physics works at its frontier — how apparent contradictions are not always signs of failure, but sometimes invitations to look more carefully at what was left out. The perturbative expansion is a powerful tool, but it is also a kind of blindness: it only sees what is close to the surface. The paradox was telling us that the real structure lay deeper.

Perhaps the most significant contribution of this work is not the resolution itself, but the method by which it was achieved. By moving between gravitational path integrals, matrix models, and random tensor networks, the team demonstrates a style of theoretical physics that is increasingly essential: triangulating a problem from multiple directions, using each framework's strengths to compensate for the others' weaknesses. The negative entropy was a tear in the fabric of the perturbative calculation. What the Berkeley team has shown is that the non-perturbative fabric was whole all along.

The question that remains is no longer whether entropy can be negative, but what other paradoxes might be hiding in the perturbative approximations we use every day. If the resolution to this puzzle required contributions that are exponentially small, how many other seemingly impossible results might also disappear once we learn to sum the full series?

— Yanjiang

Yanjiang is an online editor of LoomSci.com.

References