Statistical Physics of the Polarised IKKT Matrix Model

Sean A. Hartnoll and Jun Liu · 2025

Read Statistical Physics of the Polarised IKKT Matrix Model by Sean A. Hartnoll and Jun Liu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-06481.txt.

Introduction

A candidate theory of everything, the IKKT matrix model, can be read as a statistical system whose temperature-like dial is set by a background flux. Sean Hartnoll and Jun Liu study a polarized version of the model in this 2025 paper, treating its matrix integral as a partition function.

At strong flux, one neat spherical setup dominates the model. As the dial crosses a sharp threshold, the system jumps in a first-order phase change into a spray of separate instantons, the remains of a rival setup. Mixing numerics with analysis, the authors map the model's regimes, offering a rare tractable window into holographic physics.

Catalog record

CREATORSean A. Hartnoll and Jun Liu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-06481

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Statistical Physics of the Polarised IKKT Matrix Model Sean A. Hartnoll and Jun Liu Abstract The polarised IKKT matrix model is the worldpoint theory of N N D-instantons in a background three-form flux of magnitude Ω \Omega , and promises to be a highly tractable model of holography. The matrix integral can be viewed as a statistical physics partition function with inverse temperature Ω 4 \Omega^{4} . At large Ω \Omega the model is dominated by a matrix configuration corresponding to a ‘polarised’ spherical D1-brane. We show that at a critical value of Ω 2 ​ N \Omega^{2}N the model undergoes a first order phase transition, corresponding to tunneling into a collection of well-separated D-instantons. These instantons are the remnant of a competing saddle in the high Ω \Omega phase corresponding to spherical ( p , q ) (p,q) fivebranes. We use a combination of numerical and analytical arguments to capture the different regimes of the model. 1 Introduction Recent works [ 1 , 2 , 3 ] have revived a supersymmetric mass deformation [ 4 ] of the IKKT matrix model [ 5 ] , and demonstrated that it constitutes a viable model of holographic duality and emergent spacetime. The IKKT model is the worldpoint theory of N D-instantons in type IIB string theory, and the mass deformation captures the polarisation of the D-instantons into higher dimensional Euclidean branes by background fluxes, in the spirit of the Myers effect [ 6 ] . The model has the form of a statistical physics partition function, with neither time nor space. The absence of a quantum mechanical time is the source of some conceptual challenges, but means that the model is highly tractable. In particular, supersymmetric localisation can be used to express the partition function exactly as a sum over ‘fuzzy sphere’ matrix configurations, labeled by N dimensional representations \mathcal{R} of \mathfrak{su}(2) , such that [ 3 ] Z=\sum_{\mathcal{R}}Z_{\mathcal{R}}\,,\qquad Z_{\mathcal{R}}=C_{\mathcal{R}}e^{-S^{0}_{\mathcal{R}}}\int d\vec{m}_{\mathcal{R}}\,e^{-S^{\text{eff}}_{\mathcal{R}}(\vec{m}_{\mathcal{R}})}\,. (1) We will shortly give formulae for all quantities appearing here. For each representation there are integrals over moduli \vec{m}_{\mathcal{R}} , one modulus for each irreducible representation appearing in \mathcal{R} . In a localisation calculation, fluctuation effects about the supersymmetric fuzzy sphere minima are packaged exactly into the normalisation factors C_{\mathcal{R}} and moduli space measure e^{-S^{\text{eff}}_{\mathcal{R}}} . The expression ( 1 ) is the starting point of this paper. We will not write down the polarised IKKT matrix action explicitly, see [ 1 , 2 , 3 ] . Our objective is to understand the statistical physics of ( 1 ), and its connection to emergent spacetime physics. The ‘energetics’ of the partition function is determined by the action S^{0}_{\mathcal{R}} of the fuzzy sphere configurations S^{0}_{\mathcal{R}}=-\frac{9\Omega^{4}}{2^{15}}\sum_{M}n_{M}M(M^{2}-1)\,. (2) Here \Omega is the mass deformation parameter. The partition function Z is a function of \Omega and N only. We may think of the polarised IKKT model as a canonical partition function with \Omega^{4} being the inverse temperature. In ( 2 ) the irreducible representation of dimension M appears with degeneracy n_{M} in the decomposition of \mathcal{R} . That is, the sum over representations in ( 1 ) is equivalent to a sum over partitions \{n_{M}\} such that \sum_{M}n_{M}M=N\,. (3) In ( 2 ) we see that the supersymmetric fuzzy spheres do not all have the same action. This is distinct from, for example, the mass-deformed BMN matrix quantum mechanics in which all representations have the same energy [ 7 ] . In particular, in the large \Omega\to\infty limit the partition function is dominated by the maximally irreducible fuzzy sphere. This has n_{N}=1 and hence the lowest action ( 2 ). This is the ‘low temperature’ limit of the partition function, in which energy strongly dominates over entropy. The matrix partition function can be evaluated by saddle point in this limit [ 1 ] , without using localisation. For the U(N) theory at large \Omega \left.\frac{\log Z_{\text{irred}}}{N}\right|_{\Omega\to\infty}\approx c_{N}+2\left(\frac{3\,\Omega^{2}N}{2^{8}}\right)^{2}+{\mathcal{O}}(1/N)\,. (4) We are primarily interested in the thermodynamic large N limit of the partition function. The constant c_{N}\equiv\frac{N}{2}\left(\frac{3}{2}+\log\frac{(2\pi)^{10}}{N}\right) . At large \Omega the maximally irreducible fuzzy sphere saddle describes a spherical, classical probe D1-brane in a background with nonvanishing NSNS flux and axion [ 1 ] . The D1-brane arises as the polarisation of N D-instantons by the background flux. Fluctuations of the D1-brane geometry are described by a Maxwell field and a scalar. In the opposite \Omega\to 0 limit, the action ( 2 ) becomes the same for each representation. ‘Entropic’ effects are therefore crucial in this high temperature regime. The moduli integrals in ( 1 ) are seen to each develop a 1/\Omega^{2} divergence in this limit, and hence the partition function is dominated by the maximally reducible representation which has the most moduli integrals. The constraint ( 3 ) allows N copies of the one-dimensional representation, so that n_{1}=N . The small \Omega divergences allow the partition function to be evaluated in this limit, again without using localisation [ 3 ] \left.\frac{\log Z_{\text{triv}}}{N}\right|_{\Omega\to 0}\approx c_{N}+\frac{1}{2}\log\frac{2e^{2}}{3\pi}+\log\frac{2^{8}}{3\,\Omega^{2}N}+{\mathcal{O}}(1/N)\,. (5) In this limit the dominant configurations should be thought of as N well-separated D-instantons. These are captured by approximately commuting matrices, with the simultaneous eigenvalues giving the location of the D-instantons. We will see in § 6 below that at larger values of \Omega the D-instantons in this ‘trivial’, maximally reducible, representation coalesce to form (p,q) fivebrane bound states of NS5- and D5-branes. The polarised IKKT model therefore exhibits the familiar behavior of a canonical partition function: a low temperature phase dominated by energy gives way to a high temperature phase dominated by entropy. In the thermodynamic large N limit, the phases may be distinguished by the expectation value of the \mathfrak{su}(2) Casimir. On a given representation the trace of the Casimir is \frac{1}{N}\text{Tr}_{\mathcal{R}}C_{2}\equiv\frac{1}{4N}\sum_{M}n_{M}M(M^{2}-1)\,. (6) This expression is proportional to the on-shell action ( 2 ). At large \Omega the above discussion implies that \langle\frac{1}{N}\text{Tr}\,C_{2}\rangle\approx\frac{1}{4}N^{2} , while at small \Omega it follows that \langle\frac{1}{N}\text{Tr}\,C_{2}\rangle=0 . It is natural to suspect that there is a phase transition between these two regimes. Comparing the limiting behaviours ( 4 ) and ( 5 ), one can anticipate that at large N there will be a first order phase transition at \frac{3\,\Omega^{2}N}{2^{8}}={\mathcal{O}}(1)\,. (7) In § 2 , immediately below, we will find evidence for such a first order transition by performing a numerical evaluation of the full partition function ( 1 ). In particular, the Casimir will be found to jump abruptly between the two phases. Phase transitions out of fuzzy sphere configurations have been seen previously in other matrix models. Such a transition was found by Monte Carlo simulation of a three-matrix integral [ 8 , 9 ] , and an analogous quantum phase transition in the ground state of a three-matrix quantum mechanics was found in [ 10 ] , using a neural network variational wavefunction. Monte Carlo simulations of the thermal partition function of the string-theoretic BMN matrix quantum mechanics also exhibit the desintegration of a fuzzy sphere saddle, which is usually interpreted as a deconfinement transition [ 11 , 12 , 13 , 14 ] . A similar mass-induced transition is seen [ 15 ] in the Lorentzian IKKT model introduced in [ 16 ] . The localisation formula ( 1 ) allows us to study

Read the complete text