Showing 1–18 of 18 results for author: Winer, M

Deep Neural Nets as Hamiltonians

Authors: Mike Winer, Boris Hanin

Abstract: Neural networks are complex functions of both their inputs and parameters. Much prior work in deep learning theory analyzes the distribution of network outputs at a fixed a set of inputs (e.g. a training dataset) over random initializations of the network parameters. The purpose of this article is to consider the opposite situation: we view a randomly initialized Multi-Layer Perceptron (MLP) as a… ▽ More Neural networks are complex functions of both their inputs and parameters. Much prior work in deep learning theory analyzes the distribution of network outputs at a fixed a set of inputs (e.g. a training dataset) over random initializations of the network parameters. The purpose of this article is to consider the opposite situation: we view a randomly initialized Multi-Layer Perceptron (MLP) as a Hamiltonian over its inputs. For typical realizations of the network parameters, we study the properties of the energy landscape induced by this Hamiltonian, focusing on the structure of near-global minimum in the limit of infinite width. Specifically, we use the replica trick to perform an exact analytic calculation giving the entropy (log volume of space) at a given energy. We further derive saddle point equations that describe the overlaps between inputs sampled iid from the Gibbs distribution induced by the random MLP. For linear activations we solve these saddle point equations exactly. But we also solve them numerically for a variety of depths and activation functions, including $\tanh, \sin, \text{ReLU}$, and shaped non-linearities. We find even at infinite width a rich range of behaviors. For some non-linearities, such as $\sin$, for instance, we find that the landscapes of random MLPs exhibit full replica symmetry breaking, while shallow $\tanh$ and ReLU networks or deep shaped MLPs are instead replica symmetric. △ Less

Submitted 5 April, 2025; v1 submitted 31 March, 2025; originally announced March 2025.

Comments: 19+7 pages

A Jammed Parisi Ansatz

Authors: Michael Winer, Aidan Herderschee

Abstract: Constraint Satisfaction Problems are ubiquitous in fields ranging from the physics of solids to artificial intelligence. In many cases, such systems undergo a transition when the ratio of constraints to variables reaches some value $α_{\textrm{crit}}$. Above this critical value, it is exponentially unlikely that all constraints can be mutually satisfied. We calculate the probability that constrain… ▽ More Constraint Satisfaction Problems are ubiquitous in fields ranging from the physics of solids to artificial intelligence. In many cases, such systems undergo a transition when the ratio of constraints to variables reaches some value $α_{\textrm{crit}}$. Above this critical value, it is exponentially unlikely that all constraints can be mutually satisfied. We calculate the probability that constraints can all be satisfied, $P(\textrm{SAT})$, for the spherical perceptron. Traditional replica methods, such as the Parisi ansatz, fall short. We find a new ansatz, the jammed Parisi ansatz, that correctly describes the behavior of the system in this regime. With the jammed Parisi ansatz, we calculate $P(\textrm{SAT})$ for the first time and match previous computations of thresholds. We anticipate that the techniques developed here will be applicable to general constraint satisfaction problems and the identification of hidden structures in data sets. △ Less

Submitted 24 March, 2025; v1 submitted 14 March, 2025; originally announced March 2025.

Failure of the large-N expansion in a Bosonic Tensor Model

Authors: Aidan Herderschee, Michael Winer

Abstract: We study the tensor model generalization of the quantum $p$-spherical model in the large-$N$ limit. While the tensor model has the same large-$N$ expansion as the disordered quantum $p$-spherical model, its ground state is superextensive, in contradiction with large-$N$ perturbation theory. Therefore, the large-$N$ expansion of this model catastrophically fails at arbitrarily large-$N$, without an… ▽ More We study the tensor model generalization of the quantum $p$-spherical model in the large-$N$ limit. While the tensor model has the same large-$N$ expansion as the disordered quantum $p$-spherical model, its ground state is superextensive, in contradiction with large-$N$ perturbation theory. Therefore, the large-$N$ expansion of this model catastrophically fails at arbitrarily large-$N$, without any obvious signal in perturbation theory. △ Less

Submitted 3 March, 2025; v1 submitted 9 October, 2024; originally announced October 2024.

The Magnetic Maze: A System With Tunable Scale Invariance

Authors: Tian-Gang Zhou, Michael Winer, Brian Swingle

Abstract: Random magnetic field configurations are ubiquitous in nature. Such fields lead to a variety of dynamical phenomena, including localization and glassy physics in some condensed matter systems and novel transport processes in astrophysical systems. Here we consider the physics of a charged quantum particle moving in a ``magnetic maze'': a high-dimensional space filled with a randomly chosen vector… ▽ More Random magnetic field configurations are ubiquitous in nature. Such fields lead to a variety of dynamical phenomena, including localization and glassy physics in some condensed matter systems and novel transport processes in astrophysical systems. Here we consider the physics of a charged quantum particle moving in a ``magnetic maze'': a high-dimensional space filled with a randomly chosen vector potential and a corresponding magnetic field. We derive a path integral description of the model by introducing appropriate collective variables and integrating out the random vector potential, and we solve for the dynamics in the limit of large dimensionality. We derive and analyze the equations of motion for Euclidean and real-time dynamics, and we calculate out-of-time-order correlators. We show that a special choice of vector potential correlations gives rise, in the low temperature limit, to a novel scale-invariant quantum theory with a tunable dynamical exponent. Moreover, we show that the theory is chaotic with a tunable chaos exponent which approaches the chaos bound at low temperature and strong coupling. △ Less

Submitted 9 September, 2024; v1 submitted 3 September, 2024; originally announced September 2024.

Neural Networks as Spin Models: From Glass to Hidden Order Through Training

Authors: Richard Barney, Michael Winer, Victor Galitski

Abstract: We explore a one-to-one correspondence between a neural network (NN) and a statistical mechanical spin model where neurons are mapped to Ising spins and weights to spin-spin couplings. The process of training an NN produces a family of spin Hamiltonians parameterized by training time. We study the magnetic phases and the melting transition temperature as training progresses. First, we prove analyt… ▽ More We explore a one-to-one correspondence between a neural network (NN) and a statistical mechanical spin model where neurons are mapped to Ising spins and weights to spin-spin couplings. The process of training an NN produces a family of spin Hamiltonians parameterized by training time. We study the magnetic phases and the melting transition temperature as training progresses. First, we prove analytically that the common initial state before training--an NN with independent random weights--maps to a layered version of the classical Sherrington-Kirkpatrick spin glass exhibiting a replica symmetry breaking. The spin-glass-to-paramagnet transition temperature is calculated. Further, we use the Thouless-Anderson-Palmer (TAP) equations--a theoretical technique to analyze the landscape of energy minima of random systems--to determine the evolution of the magnetic phases on two types of NNs (one with continuous and one with binarized activations) trained on the MNIST dataset. The two NN types give rise to similar results, showing a quick destruction of the spin glass and the appearance of a phase with a hidden order, whose melting transition temperature $T_c$ grows as a power law in training time. We also discuss the properties of the spectrum of the spin system's bond matrix in the context of rich vs. lazy learning. We suggest that this statistical mechanical view of NNs provides a useful unifying perspective on the training process, which can be viewed as selecting and strengthening a symmetry-broken state associated with the training task. △ Less

Submitted 12 August, 2024; originally announced August 2024.

Comments: 18 pages, 9 figures

Spectral Statistics, Hydrodynamics and Quantum Chaos

Authors: Michael Winer

Abstract: One of the central problems in many-body physics, both classical and quantum, is the relations between different notions of chaos. Ergodicity, mixing, operator growth, the eigenstate thermalization hypothesis, and spectral chaos are defined in terms of completely different objects in different contexts, don't necessarily co-occur, but still seem to be manifestations of closely related phenomena. I… ▽ More One of the central problems in many-body physics, both classical and quantum, is the relations between different notions of chaos. Ergodicity, mixing, operator growth, the eigenstate thermalization hypothesis, and spectral chaos are defined in terms of completely different objects in different contexts, don't necessarily co-occur, but still seem to be manifestations of closely related phenomena. In this dissertation, we study the relation between two notions of chaos: thermalization and spectral chaos. We define a quantity called the Total Return Probability (TRP) which measures how a system forgets its initial state after time $T$, and show that it is closely connected to the Spectral Form Factor (SFF), a measure of chaos deriving from the energy level spectrum of a quantum system. The main thrust of this work concerns hydrodynamic systems -- systems where locality prevents charge or energy from spreading quickly, this putting a throttle on thermalization. We show that the detailed spacings of energy levels closely capture the dynamics of these locally conserved charges. We also study spin glasses, a phase of matter where the obstacle to thermalization comes not from locality but from the presence of too many neighbors. Changing one region requires changing nearby regions which requires changing nearby-to-nearby regions, until only catastrophic realignments of the whole system can fully explore phase space. In spin glasses we find our clearest analytic link between thermalization and spectral statistics. We analytically calculate the spectral form factor in the limit of large system size and show it is equal to the TRP. Finally, in the conclusion, we discuss some ideas for the future of both the SFF and the TRP. △ Less

Submitted 10 July, 2024; originally announced July 2024.

Comments: PhD Thesis. Long. Introduction and Conclusion are new. Chapters 2-5 are unmodified papers submitted to arxiv by this author and collaborators

The Glass Transition of Quantum Hard Spheres in High Dimensions

Authors: Michael Winer, Christopher L. Baldwin, Richard Barney, Victor Galitski, Brian Swingle

Abstract: We study the equilibrium thermodynamics of quantum hard spheres in the infinite-dimensional limit, determining the boundary between liquid and glass phases in the temperature-density plane by means of the Franz-Parisi potential. We find that as the temperature decreases from high values, the effective radius of the spheres is enhanced by a multiple of the thermal de Broglie wavelength, thus increa… ▽ More We study the equilibrium thermodynamics of quantum hard spheres in the infinite-dimensional limit, determining the boundary between liquid and glass phases in the temperature-density plane by means of the Franz-Parisi potential. We find that as the temperature decreases from high values, the effective radius of the spheres is enhanced by a multiple of the thermal de Broglie wavelength, thus increasing the effective filling fraction and decreasing the critical density for the glass phase. Numerical calculations show that the critical density continues to decrease monotonically as the temperature decreases further, suggesting that the system will form a glass at sufficiently low temperatures for any density. △ Less

Submitted 8 November, 2023; originally announced November 2023.

Directly observing replica symmetry breaking in a vector quantum-optical spin glass

Authors: Ronen M. Kroeze, Brendan P. Marsh, David Atri Schuller, Henry S. Hunt, Alexander N. Bourzutschky, Michael Winer, Sarang Gopalakrishnan, Jonathan Keeling, Benjamin L. Lev

Abstract: Spin glasses are quintessential examples of complex matter. Although much about their order remains uncertain, abstract models of them inform, e.g., the classification of combinatorial optimization problems, the magnetic ordering in metals with impurities, and artificial intelligence -- where they form a mathematical basis for neural network computing and brain modeling. We demonstrate the ability… ▽ More Spin glasses are quintessential examples of complex matter. Although much about their order remains uncertain, abstract models of them inform, e.g., the classification of combinatorial optimization problems, the magnetic ordering in metals with impurities, and artificial intelligence -- where they form a mathematical basis for neural network computing and brain modeling. We demonstrate the ability of an active quantum gas microscope to realize a spin glass of a novel driven-dissipative and vector form. By microscopically visualizing its glassy spin states, the technique allows us to directly measure replica symmetry breaking and the resulting ultrametric hierarchical structure. Ultrametricity is known to be emergent in models of evolution, protein folding, climate change, and infinite-range equilibrium spin glasses; this work shows it to be directly observable in a physically realized system. △ Less

Submitted 20 November, 2024; v1 submitted 7 November, 2023; originally announced November 2023.

Comments: main text 6 pages and 4 figures plus references, supplement 40 pages and 26 figures plus references

A Bosonic Model of Quantum Holography

Authors: Brian Swingle, Michael Winer

Abstract: We analyze a model of qubits which we argue has an emergent quantum gravitational description similar to the fermionic Sachdev-Ye-Kitaev (SYK) model. The model we consider is known as the quantum $q$-spin model because it features $q$-local interactions between qubits. It was previously studied as a model of a quantum spin glass, and while we find that the model is glassy for $q=2$, $q=3$, and lik… ▽ More We analyze a model of qubits which we argue has an emergent quantum gravitational description similar to the fermionic Sachdev-Ye-Kitaev (SYK) model. The model we consider is known as the quantum $q$-spin model because it features $q$-local interactions between qubits. It was previously studied as a model of a quantum spin glass, and while we find that the model is glassy for $q=2$, $q=3$, and likely $q=4$, we also find evidence for previously unexpected SYK-like behavior for the quenched free energy down to the lowest temperatures for $q \geq 5$. This SYK-like physics includes power-law correlation functions and an extensive low temperature entropy, so we refer to the model as Spin SYK. The model is generic in that it includes all possible $q$-body couplings, lacks most symmetries, and has no spatial structure, so our results can be construed as establishing a certain ubiquity of quantum holography in systems dominated by many-body interactions. Furthermore, we discuss a generalized family of models which includes Spin SYK and which provably exhibit SYK-like physics in the solvable limit of large local Hilbert space dimension. We also comment on implications of a bosonic system with SYK-like properties for the study of holography, Hamiltonian complexity, and related topics. △ Less

Submitted 2 November, 2023; originally announced November 2023.

Reappearance of Thermalization Dynamics in the Late-Time Spectral Form Factor

Authors: Michael Winer, Brian Swingle

Abstract: The spectral form factor (SFF) is an important diagnostic of energy level repulsion in random matrix theory (RMT) and quantum chaos. The short-time behavior of the SFF as it approaches the RMT result acts as a diagnostic of the ergodicity of the system as it approaches the thermal state. In this work we observe that for systems without time-reversal symmetry, there is a second break from the RMT r… ▽ More The spectral form factor (SFF) is an important diagnostic of energy level repulsion in random matrix theory (RMT) and quantum chaos. The short-time behavior of the SFF as it approaches the RMT result acts as a diagnostic of the ergodicity of the system as it approaches the thermal state. In this work we observe that for systems without time-reversal symmetry, there is a second break from the RMT result at late time around the Heisenberg time. Long after thermalization has taken hold, and after the SFF has agreed with the RMT result to high precision for a time of order the Heisenberg time, the SFF of a large system will briefly deviate from the RMT behavior in a way exactly determined by its early time thermalization properties. The conceptual reason for this second deviation is the Riemann-Siegel lookalike formula, a resummed expression for the spectral determinant relating late time behavior to early time spectral statistics. We use the lookalike formula to derive a precise expression for the late time SFF for semi-classical quantum chaotic systems, and then confirm our results numerically for more general systems. △ Less

Submitted 30 July, 2023; v1 submitted 26 July, 2023; originally announced July 2023.

Spectral statistics of a minimal quantum glass model

Authors: Richard Barney, Michael Winer, Christopher L. Baldwin, Brian Swingle, Victor Galitski

Abstract: Glasses have the interesting feature of being neither integrable nor fully chaotic. They thermalize quickly within a subspace but thermalize much more slowly across the full space due to high free energy barriers which partition the configuration space into sectors. Past works have examined the Rosenzweig-Porter (RP) model as a minimal quantum model which transitions from localized to chaotic beha… ▽ More Glasses have the interesting feature of being neither integrable nor fully chaotic. They thermalize quickly within a subspace but thermalize much more slowly across the full space due to high free energy barriers which partition the configuration space into sectors. Past works have examined the Rosenzweig-Porter (RP) model as a minimal quantum model which transitions from localized to chaotic behavior. In this work we generalize the RP model in such a way that it becomes a minimal model which transitions from glassy to chaotic behavior, which we term the "Block Rosenzweig-Porter" (BRP) model. We calculate the spectral form factors of both models at all timescales. Whereas the RP model exhibits a crossover from localized to ergodic behavior at the Thouless timescale, the new BRP model instead crosses over from glassy to fully chaotic behavior, as seen by a change in the slope of the ramp of the spectral form factor. △ Less

Submitted 27 June, 2023; v1 submitted 1 February, 2023; originally announced February 2023.

Comments: 41 pages, 10 figures

Journal ref: SciPost Phys. 15, 084 (2023)

Emergent Spectral Form Factors in Sonic Systems

Authors: Michael Winer, Brian Swingle

Abstract: We study the spectral form factor (SFF) for hydrodynamic systems with a sound pole, a large class including any fluid with momentum conservation and energy conservation, or any extended system with spontaneously broken continuous symmetry. We study such systems in a finite volume cavity and find that the logarithm of the hydrodynamic enhancement to the SFF is closely related to the spectral form f… ▽ More We study the spectral form factor (SFF) for hydrodynamic systems with a sound pole, a large class including any fluid with momentum conservation and energy conservation, or any extended system with spontaneously broken continuous symmetry. We study such systems in a finite volume cavity and find that the logarithm of the hydrodynamic enhancement to the SFF is closely related to the spectral form factor of a quantum particle moving in the selfsame cavity. Depending upon the dimensionality and nature of the effective single-particle physics, these systems exhibit a range of behaviors including an intricate resonance phenomenon, emergent integrability in the SFF, and anomalously large fluctuations of the SFF. △ Less

Submitted 16 August, 2023; v1 submitted 16 November, 2022; originally announced November 2022.

Comments: 31 pages

The Loschmidt Spectral Form Factor

Authors: Michael Winer, Brian Swingle

Abstract: The Spectral Form Factor (SFF) measures the fluctuations in the density of states of a Hamiltonian. We consider a generalization of the SFF called the Loschmidt Spectral Form Factor, $\textrm{tr}[e^{iH_1T}]\textrm{tr} [e^{-iH_2T}]$, for $H_1-H_2$ small. If the ensemble average of the SFF is the variance of the density fluctuations for a single Hamiltonian drawn from the ensemble, the averaged Losc… ▽ More The Spectral Form Factor (SFF) measures the fluctuations in the density of states of a Hamiltonian. We consider a generalization of the SFF called the Loschmidt Spectral Form Factor, $\textrm{tr}[e^{iH_1T}]\textrm{tr} [e^{-iH_2T}]$, for $H_1-H_2$ small. If the ensemble average of the SFF is the variance of the density fluctuations for a single Hamiltonian drawn from the ensemble, the averaged Loschmidt SFF is the covariance for two Hamiltonians drawn from a correlated ensemble. This object is a time-domain version of the parametric correlations studied in the quantum chaos and random matrix literatures. We show analytically that the averaged Loschmidt SFF is proportional to $e^{iλT}T$ for a complex rate $λ$ with a positive imaginary part, showing in a quantitative way that the long-time details of the spectrum are exponentially more sensitive to perturbations than the short-time properties. We calculate $λ$ in a number of cases, including random matrix theory, theories with a single localized defect, and hydrodynamic theories. △ Less

Submitted 5 October, 2022; v1 submitted 1 June, 2022; originally announced June 2022.

Comments: 16 pages, comments welcome

Spectral Form Factor of a Quantum Spin Glass

Authors: Michael Winer, Richard Barney, Christopher L. Baldwin, Victor Galitski, Brian Swingle

Abstract: It is widely expected that systems which fully thermalize are chaotic in the sense of exhibiting random-matrix statistics of their energy level spacings, whereas integrable systems exhibit Poissonian statistics. In this paper, we investigate a third class: spin glasses. These systems are partially chaotic but do not achieve full thermalization due to large free energy barriers. We examine the leve… ▽ More It is widely expected that systems which fully thermalize are chaotic in the sense of exhibiting random-matrix statistics of their energy level spacings, whereas integrable systems exhibit Poissonian statistics. In this paper, we investigate a third class: spin glasses. These systems are partially chaotic but do not achieve full thermalization due to large free energy barriers. We examine the level spacing statistics of a canonical infinite-range quantum spin glass, the quantum $p$-spherical model, using an analytic path integral approach. We find statistics consistent with a direct sum of independent random matrices, and show that the number of such matrices is equal to the number of distinct metastable configurations -- the exponential of the spin glass "complexity" as obtained from the quantum Thouless-Anderson-Palmer equations. We also consider the statistical properties of the complexity itself and identify a set of contributions to the path integral which suggest a Poissonian distribution for the number of metastable configurations. Our results show that level spacing statistics can probe the ergodicity-breaking in quantum spin glasses and provide a way to generalize the notion of spin glass complexity beyond models with a semi-classical limit. △ Less

Submitted 6 July, 2022; v1 submitted 23 March, 2022; originally announced March 2022.

Comments: 38 pages, comments welcome

Spontaneous Symmetry Breaking, Spectral Statistics, and the Ramp

Authors: Michael Winer, Brian Swingle

Abstract: Ensembles of quantum chaotic systems are expected to exhibit energy eigenvalues with random-matrix-like level repulsion between pairs of energies separated by less than the inverse Thouless time. Recent research has shown that exact and approximate global symmetries of a system have clear signatures in these spectral statistics, enhancing the spectral form factor or correspondingly weakening level… ▽ More Ensembles of quantum chaotic systems are expected to exhibit energy eigenvalues with random-matrix-like level repulsion between pairs of energies separated by less than the inverse Thouless time. Recent research has shown that exact and approximate global symmetries of a system have clear signatures in these spectral statistics, enhancing the spectral form factor or correspondingly weakening level repulsion. This paper extends those results to the case of spontaneous symmetry breaking, and shows that, surprisingly, spontaneously breaking a symmetry further enhances the spectral form factor. For both RMT-inspired toy models and models where the symmetry breaking has a description in terms of fluctuating hydrodynamics, we obtain formulas for this enhancement for arbitrary symmetry breaking patterns, including $Z_n$, $U(1)$, and partially or fully broken non-Abelian symmetries. △ Less

Submitted 14 June, 2021; originally announced June 2021.

Hydrodynamic Theory of the Connected Spectral Form Factor

Authors: Michael Winer, Brian Swingle

Abstract: One manifestation of quantum chaos is a random-matrix-like fine-grained energy spectrum. Prior to the inverse level spacing time, random matrix theory predicts a `ramp' of increasing variance in the connected part of the spectral form factor. However, in realistic quantum chaotic systems, the finite time dynamics of the spectral form factor is much richer, with the pure random matrix ramp appearin… ▽ More One manifestation of quantum chaos is a random-matrix-like fine-grained energy spectrum. Prior to the inverse level spacing time, random matrix theory predicts a `ramp' of increasing variance in the connected part of the spectral form factor. However, in realistic quantum chaotic systems, the finite time dynamics of the spectral form factor is much richer, with the pure random matrix ramp appearing only at sufficiently late time. In this article, we present a hydrodynamic theory of the connected spectral form factor prior to the inverse level spacing time. We start from a discussion of exact symmetries and spectral stretching and folding. We then derive a general formula for the spectral form factor of a system with almost-conserved sectors in terms of return probabilities and spectral form factors within each sector. Next we argue that the theory of fluctuating hydrodynamics can be adapted from the usual Schwinger-Keldysh contour to the periodic time setting needed for the spectral form factor, and we show explicitly that the general formula is recovered in the case of energy diffusion. We also initiate a study of interaction effects in this modified hydrodynamic framework and show how the Thouless time, defined as the time required for the spectral form factor to approach the pure random matrix result, is controlled by the slow hydrodynamics modes. △ Less

Submitted 11 March, 2022; v1 submitted 2 December, 2020; originally announced December 2020.

Comments: v3: Removed didactic sections, added in section on Floquet dynamics

An exponential ramp in the quadratic Sachdev-Ye-Kitaev model

Authors: Michael Winer, Shao-Kai Jian, Brian Swingle

Abstract: A long period of linear growth in the spectral form factor provides a universal diagnostic of quantum chaos at intermediate times. By contrast, the behavior of the spectral form factor in disordered integrable many-body models is not well understood. Here we study the two-body Sachdev-Ye-Kitaev model and show that the spectral form factor features an exponential ramp, in sharp contrast to the line… ▽ More A long period of linear growth in the spectral form factor provides a universal diagnostic of quantum chaos at intermediate times. By contrast, the behavior of the spectral form factor in disordered integrable many-body models is not well understood. Here we study the two-body Sachdev-Ye-Kitaev model and show that the spectral form factor features an exponential ramp, in sharp contrast to the linear ramp in chaotic models. We find a novel mechanism for this exponential ramp in terms of a high-dimensional manifold of saddle points in the path integral formulation of the spectral form factor. This manifold arises because the theory enjoys a large symmetry group. With finite nonintegrable interaction strength, these delicate symmetries reduce to a relative time translation, causing the exponential ramp to give way to a linear ramp. △ Less

Submitted 26 June, 2020; originally announced June 2020.

Comments: 5+6 pages

Journal ref: Phys. Rev. Lett. 125, 250602 (2020)

Observation of relativistic corrections to Moseley's law at high atomic number

Authors: Duncan C. Wheeler, Emma Bingham, Michael Winer, Janet M. Conrad, Sean P. Robinson

Abstract: Transitions between low-lying electron states in atoms of heavy elements lead to electromagnetic radiation with discrete energies between about 0.1~keV and 100~keV (x rays) that are characteristic of the element. Moseley's law --- an empirical relation first described by Moseley in 1913 which supported predictions of the then-new Bohr model of atomic energy levels while simultaneously identifying… ▽ More Transitions between low-lying electron states in atoms of heavy elements lead to electromagnetic radiation with discrete energies between about 0.1~keV and 100~keV (x rays) that are characteristic of the element. Moseley's law --- an empirical relation first described by Moseley in 1913 which supported predictions of the then-new Bohr model of atomic energy levels while simultaneously identifying the integer atomic number $Z$ as the measure of nuclear charge --- predicts that the energy of these characteristic x rays scales as $Z^2$. The foundational nature of Moseley's experiment has led to the popularity of Moseley's law measurements in undergraduate advanced laboratory physics courses. We report here observations of deviations from Moseley's law in the characteristic $K_α$ x-ray emission of 13 elements ranging from $Z=29$ to $Z=92$. %: Cu, Rb, Mo, Ag, In, Ba, Tb, Ta, W, Pb, Au, Pt, and U. While following the square-law predictions of the Bohr model fairly well at low $Z$, the deviations become larger with increasing $Z$ (negligible probability of the Bohr model fitting data by a $χ^2$ test). We find that relativistic models of atomic structure are necessary to fit the full range of atomic numbers observed (probability value of $0.20$ for the relativistic Bohr-Sommerfeld model). As has been argued by previous authors, measurements of the relativistic deviations from Moseley's law are both pedagogically valuable at the advanced laboratory level and accessible with modern but modest apparatus. Here, we show that this pedagogical value can be be extended even further --- to higher $Z$ elements, where the effects are more dramatically observable --- using apparatus which is enhanced relative to more modest versions, but nevertheless still accessible for many teaching laboratories. △ Less

Submitted 17 December, 2018; v1 submitted 11 September, 2018; originally announced September 2018.