Multifractal scaling of quantum eigenfunctions in the thermodynamic limit

arXiv:2611.08087 · math-ph, math.MP · Submitted 2026-11-05 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematical Physics. The day in mathematical physics: rigorous models of physical systems.

Kai: We're looking at the paper: "Multifractal scaling of quantum eigenfunctions in the thermodynamic limit".

Mira: Here's the gist of the paper. MULTIFRACTAL SCALING OF QUANTUM EIGENFUNCTIONS IN THE THERMODYNAMIC LIMIT proves multifractal scaling of measures related to the eigenfunctions of Schrödinger operators with point-like bump potentials, in the thermodynamic limit, in dimension d ≥ 5 This is consistent with a longstanding expectation in the physics literature in these dimensions The fractal exponents are computed explicitly as a function of the parameter defining the moments.

Kai: The paper addresses the rigorous proof of localization-delocalization transitions in disordered systems, focusing on multifractal scaling of eigenfunctions in the thermodynamic limit.

Mira: The mathematical setup involves a sequence of operators HL = −∆ + Xω∈omegaL αω,LVr(· − ω), on tori Td L as L → +∞.

Kai: Key assumptions include a weak disorder condition and a strong coupling condition.

Paper summary: Kai: To recap the main thrust of "Multifractal scaling of quantum eigenfunctions in the thermodynamic limit," the paper focuses on proving multifractal scaling for measures related to Schrödinger operators with point-like bump potentials as the system size goes to infinity in dimension d greater than five >

Mira: They claim this is consistent with a long-standing expectation in physics literature for these dimensions, and they compute the fractal exponents explicitly as a function of the parameter defining those moments >

Lev: The core of their mathematical framework involves considering a sequence of operators HL L equals minus delta plus X omega in a set Omega L on tori T d L as L approaches infinity >

Kai: They establish this by assuming a weak disorder condition on the location of the scatterers and a strong coupling condition on the sequence of eigenvalues >

Mira: The multifractality itself is defined using Renyi entropy, specifically that for a sequence of probability measures mu N with sample space X N, it satisfies a power law for q from q one to q two where q two is greater than one >

Lev: If you look at the results in Theorem one point four, the fractal exponent Dq is given by d minus four over d times q to the power of q minus one for any moment parameter greater than d over four > <ref:2611.08087#pg3>

Kai: This formula shows a non-constant dependence on q, which is what proves multifractal scaling, confirming that these measures exhibit this complex scaling in this regime >

Mira: It’s interesting because they do this rigorously in dimension d greater than five, which is a regime where these phenomena are expected to occur >

Lev: The paper also touches on the coupling regimes based on how the parameter alpha L grows or shrinks, discussing fixed coupling versus small coupling scenarios >

Kai: And they provide some limit theorems about eigenfunctions, showing that convergence of eigenfunctions has very specific constraints regarding their Fourier transforms and integrability >

Conclusion: Kai: So looking at the title "Multifractal scaling of quantum eigenfunctions in the thermodynamic limit," it really just means they're studying how these quantum states behave as the system gets infinitely large in a high-dimensional space >

Mira: It’s about characterizing that behavior by showing that the probability measures associated with those states don't scale simply, but instead follow a multifractal scaling law defined by those fractal exponents >

Lev: In simple terms, what this changes for someone who just listens to the show is that it confirms a specific type of scaling exists in these high dimensions for these kinds of systems >

Kai: It’s a confirmation that the mathematical tools they used can rigorously prove this behavior without having to rely on simpler approximations >

Mira: They're showing that even with disorder and point-like scatterers, you get this rich structure in the scaling, and they're doing it in dimensions d greater than five >

Lev: So for us working on hardware, it tells us that the overhead associated with modeling these states will depend heavily on which specific regime of coupling you are actually operating in >

JONATHAN P. KEATING, HENRIK UEBERSCHAR¨

University of Oxford · Sorbonne Université and Université Paris Cité, CNRS, Institut Mathématique Jean Picard

math-ph, math.MP

Submitted: 2026-11-05

Updated: 2026-11-05

Comments: 25 pages, no figures

Code: https://github.com/openai/math

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 91/100

The gist: MULTIFRACTAL SCALING OF QUANTUM EIGENFUNCTIONS IN THE THERMODYNAMIC LIMIT proves multifractal scaling of measures related to the eigenfunctions of Schrödinger operators with point-like bump…

Key concepts

Multifractal Scaling
This describes how probability measures associated with quantum eigenfunctions behave under scaling. Instead of a single scaling exponent, it shows that different parts of the measure scale differently, characterized by a continuous spectrum of fractal exponents.
Fractal Exponent (Dq)
The Dq exponent quantifies multifractality. It is defined based on the Renyi entropy Hq(µN) divided by the logarithm of the scaling parameter N. A continuous range of these exponents indicates that the system exhibits complex, non-uniform scaling properties.
Renyi Entropy (Hq)
Renyi entropy is a mathematical tool used to measure the complexity and multifractality of a probability measure. By calculating Hq(µN), researchers can determine how the measure scales with respect to the parameter N, which directly yields the fractal exponents.

Terminology

Summary

MULTIFRACTAL SCALING OF QUANTUM EIGENFUNCTIONS IN THE THERMODYNAMIC LIMIT proves multifractal scaling of measures related to the eigenfunctions of Schrödinger operators with point-like bump potentials, in the thermodynamic limit, in dimension d ≥ 5 This is consistent with a longstanding expectation in the physics literature in these dimensions The fractal exponents are computed explicitly as a function of the parameter defining the moments. p. 0

Where the paper says it

We prove multifractal scaling of measures related to the eigenfunctions of Schr¨odinger operators with point-like bump potentials, in the thermodynamic limit, in dimension d ≥ 5. This is consistent with a longstanding expectation in the physics literature in these dimensions.

Background and Mathematical Framework

The paper addresses the rigorous proof of localization-delocalization transitions in disordered systems, focusing on multifractal scaling of eigenfunctions in the thermodynamic limit. The mathematical setup involves a sequence of operators HL = −∆ + Xω∈omegaL αω,LVr(· − ω), on tori Td L as L → +∞. p. 1

Where the paper says it

In this article, we consider the sequence of operators HL = −∆ + X ω∈ΩL αω,LVr(· − ω), aω,L ∈ C, (1. 1) on the tori T d L = R d/2πLZ d , d ≥ 5, as L → +∞, where V˜ ∈ C∞ c (R d , C), supp V˜ ⊂ B(0, 1), R V˜ = 1, is a smooth bump potential and Vr(x) = X k∈2πZd V˜ r(x − Lk), V˜ r(x) = r −dV˜ x r , is the periodized potential which lives on T d L .

Key assumptions include a weak disorder condition and a strong coupling condition. p. 5
Where the paper says it

We make the following assumptions: (WD) Assume that we satisfy a weak disorder condition in the sense that for any ω ∈ ΩLj we have X ω′∈ΩLj \{ω} |ω − ω ′ | −N ≤ 1 2CN . 14) (SC) Moreover, assume that the sequence of eigenvalues λj satisfies a strong coupling condition: |λjL 2 j − n ± j | ≳ 1, as j → +∞, (1.

The study focuses on the multifractal scaling of probability measures associated with the eigenfunctions, defined via Renyi entropy. p. 3
Where the paper says it

Our ultimate goal is to prove multifractal scaling laws for measures associated with the eigenfunctions of a sequence of toral Schr¨odinger operators. We will follow the approach outlined in [11] by studying probability measures associated with the eigenfunctions on the Fourier side.

Defining Multifractality

Multifractal scaling is characterized by a power law for the q-th moment of a sequence of probability measures µN, where N is the scaling parameter. The fractal exponent Dq is defined as the limit of the Renyi entropy Hq(µN) divided by log N. p. 6

Where the paper says it

Let us denote the qth moment of the averaged measure µj : Mq,j = X k∈L −1 j Zd∩B(0,Rj ) µj (k) q . 16) The Renyi entropy of µj is defined as Hq(µj ) = 1 1 − q log   X k∈L −1 j Zd∩B(0,Rj ) µj (k) M1,j q   = log Mq,j − q log M1,j 1 − q (1.

A continuous spectrum of scaling exponents indicates multifractal scaling, which depends on the location where the scaling occurs. p. 2
Where the paper says it

In the case of multifractal scaling, a continuous spectrum of scaling exponents is required. This means that the scaling exponent is local, because it depends on the location at which the scaling occurs.

Main Results in Dimension d ≥ 5

Theorem 1.4 states that for dimension d ≥ 5, under specific conditions (WD) and (SC), the fractal exponents are given by Dq = d−4/d q(q−1) for any q > d/4. p. 4

Where the paper says it

The scaling parameter is given by Nj = |Xj | ∼ cd(RjLj ) d . We note that the choice of the constant C > 0 plays no role in the computation of the fractal exponents Dq, because its choice does not affect the leading term in the asymptotic of log Nj .

This non-constant function of q confirms that the measures exhibit multifractal scaling in this regime. p. 5
Where the paper says it

Then, under assumptions (WD) and (SC), we have that Dq = d−4 d q q−1 for any q > d/4. The fact that Dq is not a constant function of q means that the measures µj exhibit multifractal scaling when d ≥ 5, in the sense defined above.

The proof relies on estimating the rescaled moments M'q j, which satisfy log M'1 j = (d − 4) log Rj + d log Lj + O(1).

Proof Outline and Limitations

The proof of Theorem 1.4 involves computing the asymptotic behavior of the Renyi entropy Hq(µj), which leads to Dq = q(d − 4)/(q − 1) / d. p. 5

Where the paper says it

While multifractal scaling is predicted in dimension d ≥ 3 in the physics literature, and the formula we find for Dq vanishes if we formally set d = 4, it is important to point out that this allows no conclusion to be drawn regarding the form of Dq in this case. We prove an asymptotic for the Renyi entropy which is based on asymptotic upper and lower bounds for the number of lattice points on a sphere of radius √ n in dimension d, the arithmetic function rd(n).

The authors note that their methods for obtaining asymptotics for lattice point sums are available in dimension d ≥ 5, but they are unable to adapt these methods for dimensions d = 3, 4. p. 5
Where the paper says it

Moreover, our ability to obtain an asymptotic for the Renyi entropy, as Lj → +∞, hinges on the fact that a leading growing term dominates another more complicated, potentially oscillatory term, which is only the case in dimension d ≥ 5. At present, we are unable to adapt our methods for dimensions d = 3, 4.

Coupling Regimes and Renormalization

Section 5 motivates different coupling regimes based on the growth or decay of the parameter αL. These regimes include fixed coupling/fixed support, small coupling/fixed support, and small coupling/small support. p. 6

Where the paper says it

These supplementary results are stated in section 4. In section 5, we motivate the different coupling regimes considered in section 4.

The relation (5.12) describes the renormalization of the coupling parameter in terms of rL, α0 and r0. p. 0
Where the paper says it

The fractal exponents are computed explicitly as a function of the parameter defining the moments. Our result requires a weak disorder condition on the location of the scatterers and a strong coupling condition on the sequence of eigenvalues.

Limit Theorems on Eigenfunctions

Theorem 4.2 establishes that if a sequence of eigenfunctions converges to a limit u in L 1(R d) on the Fourier side, then u = 0 unless P(k) = E. p. 6

Where the paper says it

In section 3 we give the proof of Theorem 1. While the methods introduced in this paper do not suffice to prove multifractality in dimension d = 3, 4, we are able to prove general limit theorems on the Fourier side for eigenfunctions in the thermodynamic limit.

Corollary 4.3 shows that if the eigenfunctions converge to a non-zero limit v in L inf(R d), then v is not integrable on R d. p. 1
Where the paper says it

We refer to the thermodynamic limit, if #ΩL ∼ ρLd , for ρ > 0, as L → +∞, and we consider eigenvalues lying in an energy window [E1, E2] for fixed 150 ≤ E1 < E2. (ii) We consider a strictly increasing sequence of length scales (Lj ) +∞ j=1, Lj ↗ +∞, and a sequence of eigenfunctions uj of the operators HLj , HLjuj = λjuj , such that the eigenvalues λj converge to a limit E ∈ [E1, E2], as Lj → +∞.

The general characterization of limits on the Fourier side is given by Theorem 4. p. 6
Where the paper says it

In section 3 we give the proof of Theorem 1. While the methods introduced in this paper do not suffice to prove multifractality in dimension d = 3, 4, we are able to prove general limit theorems on the Fourier side for eigenfunctions in the thermodynamic limit.

4.

Asymptotic Analysis of Lattice Sums

Section 7 provides asymptotic evaluations for lattice sums using summation by parts, yielding results such as X n<n−Lrd(n) n − λjL 2j ≲ L(d−1) j. The main integral term yields an asymptotic relation involving cd Lj r d−2. p. 4

Where the paper says it

We note that the choice of the constant C > 0 plays no role in the computation of the fractal exponents Dq, because its choice does not affect the leading term in the asymptotic of log Nj . Main result.

Conclusion on Localization

The analysis of the quasi-mode Ψ Lj shows that Fk = limj→+∞ ubj(ξj/Lj) vanishes unless k ∈ S d√−1E. This proves that the limit function is zero unless its Fourier transform is supported on the set of points where P(k) = E. p. 5

Where the paper says it

While multifractal scaling is predicted in dimension d ≥ 3 in the physics literature, and the formula we find for Dq vanishes if we formally set d = 4, it is important to point out that this allows no conclusion to be drawn regarding the form of Dq in this case. We prove an asymptotic for the Renyi entropy which is based on asymptotic upper and lower bounds for the number of lattice points on a sphere of radius √ n in dimension d, the arithmetic function rd(n).

Since P is a non-zero polynomial, the volume of this set is zero. p. 4
Where the paper says it

MULTIFRACTAL EIGENFUNCTIONS IN THE THERMODYNAMIC LIMIT 5 We recall our assumption (iii): rj ↘ 0, as Lj → +∞. This means the potentials Vrj are “point-like”, by which we mean that the diameter of the support of Vrj shrinks to zero in the thermodynamic limit.

The final result concludes that u = 0. p. 0
Where the paper says it

The fractal exponents are computed explicitly as a function of the parameter defining the moments. Our result requires a weak disorder condition on the location of the scatterers and a strong coupling condition on the sequence of eigenvalues.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improve spectral analysis of eigenfunctions by explicitly computing fractal exponents as a function of moments to distinguish multifractal from monofractal scaling. This allows for a mathematical criterion to capture complex behavior observed at critical disorder.

  2. Enable rigorous proof of Anderson localization and delocalization transitions in higher dimensions by applying the derived multifractal scaling results, which are consistent with longstanding expectations in dimension d >= 5.

  3. Develop new methods for proving multifractality in dimensions three and four by adapting techniques used for asymptotic upper and lower bounds on lattice point sums, as these are not available in those lower dimensions.

  4. Enhance the analysis of coupling regimes by providing a clear framework to distinguish between fixed coupling/fixed support, small coupling/fixed support, and small coupling/small support models based on the asymptotic behavior of the renormalization relation 1 / alphaL.

  5. Improve computational physics simulations by using the derived scaling laws Dq = d - 4 / (d q) for q > d / 4 as a benchmark for characterizing spectral statistics intermediate between Poisson and random-matrix statistics.

Abstract

We prove multifractal scaling of measures related to the eigenfunctions of Schrödinger operators with point-like bump potentials, in the thermodynamic limit, in dimension d at least 5. This is consistent with a longstanding expectation in the physics literature in these dimensions. The fractal exponents are computed explicitly as a function of the parameter defining the moments. Our result requires a weak disorder condition on the location of the scatterers and a strong coupling condition on the sequence of eigenvalues.

Related papers