Search

We study $D=11$ supergravity solutions which are dual to one-dimensional superconformal defects in $d=3$ SCFTs. We consider defects in ABJM theory with monodromy for $U(1)^4\subset SO(8)$ global symmetry, as well as in $\mathcal{N}=2$ mABJM SCFT, which arises from the RG flow of a mass deformation of ABJM theory, with monodromy for $U(1)^3\subset SU(3)\times U(1)$ global symmetry. We show that the defects of the two SCFTs are connected by a line of bulk marginal mass deformations and argue that they are also related by bulk RG flow. In all cases we allow for the possibility of conical singularities at the location of the defect. Various physical observables of the defects are computed including the defects conformal weight and the partition function, as well as associated supersymmetric Renyi entropies., Comment: 71 pages, 6 figures

We study type IIB supergravity solutions that are dual to two-dimensional superconformal defects in $d=4$ SCFTs which preserve $\mathcal{N}=(0,2)$ supersymmetry. We consider solutions dual to defects in $\mathcal{N}=4$ SYM theory that have non-trivial monodromy for $U(1)^3\subset SO(6)$ global symmetry and we also allow for the possibility of conical singularities. In addition, we consider the addition of fermionic and bosonic mass terms that have non trivial dependence on the spatial directions transverse to the defect, while preserving the superconformal symmetry of the defect. We compute various physical quantities including the central charges of the defect expressed as a function of the monodromy, the on-shell action as well as associated supersymmetric Renyi entropies. Analogous computations are carried out for superconformal defects in the $\mathcal{N}=1$, $d=4$ Leigh-Strassler SCFT. We also show that the defects of the two SCFTs are connected by a line of bulk marginal mass deformations and argue that they are also related by bulk RG flow., Comment: 93 pages, 8 figures. References and figure added. Various discussions refined, including on the existence of defect solutions

We show that in-context learning emerges in transformers in discrete developmental stages, when they are trained on either language modeling or linear regression tasks. We introduce two methods for detecting the milestones that separate these stages, by probing the geometry of the population loss in both parameter space and function space. We study the stages revealed by these new methods using a range of behavioral and structural metrics to establish their validity.

The infinite-parent spatial Lambda-Fleming-Viot (SLFV) process is a model of random growth, in which a set evolves by the addition of balls according to points of an underlying Poisson point process, and which was recently introduced to study genetic diversity in spatially expanding populations. In this article, we give asymptotics for the location and depth of the moving interface, and identify the exact asymptotic scale of the transverse fluctuations of geodesics. Our proofs are based on a new representation of the infinite-parent SLFV in terms of chains of reproduction events, and on the study of the properties of a typical geodesic. Moreover, we show that our representation coincides with the alternative definitions of the process considered in the literature, subject to a simple condition on the initial state. Our results represent a novel development in the study of stochastic growth models, and also have consequences for the study of genetic diversity in expanding populations., Comment: 57 pages, 2 figures

To better understand complexity in neural networks, we theoretically investigate the idealised phenomenon of lossless network compressibility, whereby an identical function can be implemented with fewer hidden units. In the setting of single-hidden-layer hyperbolic tangent networks, we define the rank of a parameter as the minimum number of hidden units required to implement the same function. We give efficient formal algorithms for optimal lossless compression and computing the rank of a parameter. Losslessly compressible parameters are atypical, but their existence has implications for nearby parameters. We define the proximate rank of a parameter as the rank of the most compressible parameter within a small L-infinity neighbourhood. We give an efficient greedy algorithm for bounding the proximate rank of a parameter, and show that the problem of tightly bounding the proximate rank is NP-complete. These results lay a foundation for future theoretical and empirical work on losslessly compressible parameters and their neighbours., Comment: 9 pages paper, 33 pages total, 10 figures, 4 tables

We consider the tight-binding model of graphene with slowly spatially varying hopping functions. We develop a low energy approximation as a derivative expansion in a Dirac spinor that is perturbative in the hopping function deformation. The leading description is the Dirac equation in flat 2+1-d spacetime with (strain-)gauge field. Prior work considered subleading corrections written as non-trivial frame and spin connection terms. We previously argued that such corrections cannot be considered consistently without taking all the terms at the same order of approximation, which due to the unconventional power counting originating from the large gauge field, involve also higher covariant derivative terms. Here we confirm this, explicitly computing subleading terms. To the order we explore, the theory is elegantly determined by the gauge field and frame, both given by the hopping functions, the torsion free spin connection of the frame, together with coefficients for the higher derivative terms derived from lattice invariants. For the first time we compute the metric that the Dirac field sees - the `electrometric' - to quadratic order in the deformation allowing us to describe the subleading corrections to the dispersion relation for inhomogeneous deformations originating from corrections to the frame. Focussing on in-plane inhomogeneous strain, we use a simple model to relate the hopping functions to the strain field, finding the electrometric becomes curved at this quadratic order. Thus this lattice model yields an effective analog gravity description as a curved space Dirac theory, with large magnetic field, and Lorentz violating higher covariant derivative terms. We check this by comparison to numerical diagonalization. From this we conjecture a form for the effective theory for monolayer graphene in terms of the strain tensor, consistent up to quadratic order in the deformation., Comment: 39 pages, 7 figures, downloadable Mathematica notebook, abridged abstract; v2. typos corrected, references added

Understanding the learning process of artificial neural networks requires clarifying the structure of the parameter space within which learning takes place. A neural network parameter's functional equivalence class is the set of parameters implementing the same input--output function. For many architectures, almost all parameters have a simple and well-documented functional equivalence class. However, there is also a vanishing minority of reducible parameters, with richer functional equivalence classes caused by redundancies among the network's units. In this paper, we give an algorithmic characterisation of unit redundancies and reducible functional equivalence classes for a single-hidden-layer hyperbolic tangent architecture. We show that such functional equivalence classes are piecewise-linear path-connected sets, and that for parameters with a majority of redundant units, the sets have a diameter of at most 7 linear segments., Comment: 15 pages, 3 figures

In recent years we have explored using Haskell alongside a traditional mathematical formalism in our large-enrolment university course on topics including logic and formal languages, aiming to offer our students a programming perspective on these mathematical topics. We have found it possible to offer almost all formative and summative assessment through an interactive learning platform, using Haskell as a lingua franca for digital exercises across our broad syllabus. One of the hardest exercises to convert into this format are traditional written proofs conveying constructive arguments. In this paper we reflect on the digitisation of this kind of exercise. We share many examples of Haskell exercises designed to target similar skills to written proof exercises across topics in propositional logic and formal languages, discussing various aspects of the design of such exercises. We also catalogue a sample of student responses to such exercises. This discussion contributes to our broader exploration of programming problems as a flexible digital medium for learning and assessment., Comment: In Proceedings TFPIE 2021/22, arXiv:2207.11600

We construct a new class of supersymmetric $AdS_3\times Y_7$ solutions of type IIB supergravity, where $Y_7$ is an $S^5$ fibration over a spindle, which are dual to $d=2$, $\mathcal{N}=(0,2)$ SCFTs. The solutions are constructed in a sub-truncation of $D=5$, $SO(6)$ maximal gauged supergravity and they all lie within the anti-twist class. We show that the central charge computed from the gravity solutions agrees with an anomaly polynomial calculation associated with compactifying the $\mathcal{N}=1$, $d=4$ Leigh-Strassler SCFT on a spindle., Comment: 55 pages, 2 figures. Very minor changes - published version

It is often very challenging to manually design reward functions for complex, real-world tasks. To solve this, one can instead use reward learning to infer a reward function from data. However, there are often multiple reward functions that fit the data equally well, even in the infinite-data limit. This means that the reward function is only partially identifiable. In this work, we formally characterise the partial identifiability of the reward function given several popular reward learning data sources, including expert demonstrations and trajectory comparisons. We also analyse the impact of this partial identifiability for several downstream tasks, such as policy optimisation. We unify our results in a framework for comparing data sources and downstream tasks by their invariances, with implications for the design and selection of data sources for reward learning., Comment: ICML 2023. 9 pages main paper, 26 pages total, 3 figures

Undistorted monolayer graphene has energy bands which cross at protected Dirac points. It elastically deforms and much research has assumed the Dirac description persists, now in a curved space and coupled to a gauge field related to lattice strain. We show this is incorrect by using a real space gradient expansion to study how the Dirac equation derives from the tight binding model. Generic spatially varying hopping functions give rise to large magnetic fields which spoil the truncation in derivatives. In the perturbative regime, the only consistent truncation to Dirac is one with nontrivial gauge field but in flat space. One can instead fine tune the magnetic field to be small, and we derive the resulting differential condition that the hopping functions must satisfy to yield a consistent truncation to Dirac in curved space. We consider whether mechanical effects might impose this fine tuning, but find this is not the case for a simple elastic membrane model., Comment: v1. 5 pages; v2. journal version of article, text improvements and some discussion expanded, some references added, title changed to adhere to journal policy, 7 pages

Let $d\ge 3$ be a fixed integer, $p\in (0,1)$, and let $n\geq 1$ be a positive integer such that $dn$ is even. Let $\mathbb{G}(n, d, p)$ be a (random) graph on $n$ vertices obtained by drawing uniformly at random a $d$-regular (simple) graph on $[n]$ and then performing independent $p$-bond percolation on it, i.e. we independently retain each edge with probability $p$ and delete it with probability $1-p$. Let $|\mathcal{C}_{\text{max}}|$ be the size of the largest component in $\mathbb{G}(n, d, p)$. We show that, when $p$ is of the form $p=(d-1)^{-1}(1+\lambda n^{-1/3})$ for $\lambda\in \mathbb{R}$, and $A$ is large, \begin{align*} \mathbb{P}(|\mathcal{C}_{\text{max}}|>An^{2/3})\asymp A^{-3/2}e^{-\frac{A^3(d-1)(d-2)}{8d^2}+\frac{\lambda A^2(d-2)^2}{2d(d-1)}-\frac{\lambda^2 A(d-1)}{2(d-2)}}. \end{align*} This improves on a result of Nachmias and Peres. We also give an analogous asymptotic for the probability that a particular vertex is in a component of size larger than $An^{2/3}$., Comment: 54 pages, 1 figure