Search

The universality of the logarithmic soft photon theorem in four dimensions can be traced to an infinite-dimensional asymptotic symmetry which acts as a local phase rotation on matter as we have shown in 2403.13053. Here we extend our earlier results for the charges associated to these superphaserotations to all orders in the coupling and prove that their conservation is exactly the classical logarithmic soft photon theorem discovered by Saha, Sahoo and Sen in 1912.06413. We furthermore generalize the formulae for the associated electromagnetic displacement memory and its tail from particles to scalar matter fields. This completes the classical superphaserotation infrared triangle., Comment: 35 pages, 1 figure

The universality of gravitational scattering at low energies and large distances encoded in soft theorems and memory effects can be understood from symmetries. In four-dimensional asymptotically flat spacetimes the infinite enhancement of translations, extending the Poincar\'e group to the BMS group, is the symmetry underlying Weinberg's soft graviton theorem and the gravitational displacement memory effect. Beyond this leading infrared triangle, loop corrections alter their nature by introducing logarithms in the soft expansion and late time tails to the memory, and this persists in the classical limit. In this work we give the first complete description of an `infrared triangle' where the long-range nature of gravitational interactions is accounted for. Building on earlier results in 2403.13053 where we derived a novel conservation law associated to the infinite dimensional enhancement of Lorentz transformations to superrotations, we prove here its validity to all orders in the gravitational coupling and show that it implies the classical logarithmic soft graviton theorem of Saha-Sahoo-Sen in 1912.06413. We furthermore extend the formula for the displacement memory and its tail from particles to fields, thus completing the classical superrotation infrared triangle., Comment: 46 pages, 1 figure

Incorporating higher curvature terms into gravity theories modifies the classical field equations, potentially leading to theoretical issues like Shapiro time advancements that violate the Camanho, Edelstein, Maldacena, and Zhiboedov (CEMZ) causality criterion. We explore this criterion within the context of Generalised Quadratic Gravity (GQG), a higher curvature theory with a distinct graviton three-point coupling from General Relativity (GR). By constructing an exact shock wave solution of GQG and calculating the Shapiro time shift for a massless probe graviton, we demonstrate that it can remain strictly positive within a classically allowed parameter space of couplings, ensuring the theory's adherence to the CEMZ criterion. This finding indicates that GQG can offer a causal extension beyond GR, paving the way for further exploration into the consistency of classical higher curvature gravity theories., Comment: 5 pages, 2 figures

The symmetries of the gravitational scattering are intimately tied to the symmetries which preserve asymptotic flatness at null infinity. In Penrose's definition of asymptotic flatness, a central role is played by the notion of asymptotic simplicity and the ensuing peeling behavior which dictates the decay rate of the Weyl tensor. However, there is now accumulating evidence that in a generic gravitational scattering the peeling property is broken, so that the spacetime is not asymptotically-flat in the usual sense. These obstructions to peeling can be traced back to the existence of universal radiative low frequency observables called "tails to the displacement memory". The universality of these tail modes is the statement of the classical logarithmic soft graviton theorem of Sahoo, Saha and Sen. Four-dimensional gravitation scattering therefore exhibits a rich infrared interplay between tail to the memory, loss of peeling, and universal logarithmic soft theorems. In this paper we study the solution space and the asymptotic symmetries for logarithmically-asymptotically-flat spacetimes. These are defined by a polyhomogeneous expansion of the Bondi metric which gives rise to a loss of peeling, and represent the classical arena which can accommodate a generic gravitational scattering containing tails to the memory. We show that while the codimension-two generalized BMS charges are sensitive to the loss of peeling at $\mathcal{I}^+$, the flux is insensitive to the fate of peeling. Due to the tail to the memory, the soft superrotation flux contains a logarithmic divergence whose coefficient is the quantity which is conserved in the scattering by virtue of the logarithmic soft theorem. In our analysis we also exhibit new logarithmic evolution equations and flux-balance laws, whose presence suggests the existence of an infinite tower of subleading logarithmic soft graviton theorems., Comment: 40 pages, published version, references added

Excessive memory requirements of key and value features (KV-cache) present significant challenges in the autoregressive inference of large language models (LLMs), restricting both the speed and length of text generation. Approaches such as Multi-Query Attention (MQA) and Grouped Query Attention (GQA) mitigate these challenges by grouping query heads and consequently reducing the number of corresponding key and value heads. However, MQA and GQA decrease the KV-cache size requirements at the expense of LLM accuracy (quality of text generation). These methods do not ensure an optimal tradeoff between KV-cache size and text generation quality due to the absence of quality-aware grouping of query heads. To address this issue, we propose Quality and Capacity-Aware Grouped Query Attention (QCQA), which identifies optimal query head groupings using an evolutionary algorithm with a computationally efficient and inexpensive fitness function. We demonstrate that QCQA achieves a significantly better tradeoff between KV-cache capacity and LLM accuracy compared to GQA. For the Llama2 $7\,$B model, QCQA achieves $\mathbf{20}$\% higher accuracy than GQA with similar KV-cache size requirements in the absence of fine-tuning. After fine-tuning both QCQA and GQA, for a similar KV-cache size, QCQA provides $\mathbf{10.55}\,$\% higher accuracy than GQA. Furthermore, QCQA requires $40\,$\% less KV-cache size than GQA to attain similar accuracy. The proposed quality and capacity-aware grouping of query heads can serve as a new paradigm for KV-cache optimization in autoregressive LLM inference.

Arkani-Hamed, Bai, He, and Yan (ABHY) discovered a convex realisation of the associahedron whose combinatorial and geometric structure generates tree-level amplitudes in bi-adjoint scalar theory. In this paper, we identify S-matrix of Yang-Mills theory with a scalar obtained by contracting the canonical form of ABHY associahedron with a multi-vector field (MVF) in the kinematic space. Components of this MVF are determined by the combinatorial structures that underlie the associahedron and Corolla polynomial that was introduced by Kreimer, Sars, and van Suijlekom (KSVS) in [2]. KSVS used the Corolla polynomial to obtain (at all orders in the loop expansion) the parametric representation of gauge theory Feynman integral from the corresponding Feynman integral in $\phi^{3}$ theory. Using the full power of Corolla polynomial, we then extend these results to obtain Yang-Mills one loop planar integrand by contracting the Corolla generated MVF with the canonical form defined by $\hat{D}_{n}$ polytope discovered by Arkani-Hamed, Frost, Plamondon, Salvatori, Thomas. We also demonstrate that KSVS representation of Corolla graph differential in the parametric space can be readily extended to "spin up" the curve integral formulae for $\textrm{Tr}\phi^{3}$ amplitude discovered in [3,4] and give an explicit construction of such formulae for tree-level and planar one loop gluon amplitudes., Comment: 68 pages, 12 figures

Gauge theories and perturbative gravity in four dimensions are governed by a tower of infinite-dimensional symmetries which arise from tree-level soft theorems. However, aside from the leading soft theorems which are all-loop exact, subleading ones receive loop corrections due to long-range infrared effects which result in new soft theorems with logarithmic dependence on the energy of the soft particle. The conjectured universality of these logarithmic soft theorems to all loop orders cries out for a symmetry interpretation. In this letter we initiate a program to compute long-range infrared corrections to the charges that generate the asymptotic symmetries in (scalar) QED and perturbative gravity. For late-time fall-offs of the electromagnetic and gravitational fields which give rise to infrared dressings for the matter fields, we derive finite charge conservation laws and show that in the quantum theory they correspond precisely to the first among the infinite tower of logarithmic soft theorems. This symmetry interpretation, by virtue of being universal and all-loop exact, is a key element for a holographic principle in spacetimes with flat asymptotics., Comment: 8 pages, fixed notation and sign convention to match 2412.16142 and 2412.16149

In an instance of the minimum eigenvalue problem, we are given a collection of $n$ vectors $v_1,\ldots, v_n \subset {\mathbb{R}^d}$, and the goal is to pick a subset $B\subseteq [n]$ of given vectors to maximize the minimum eigenvalue of the matrix $\sum_{i\in B} v_i v_i^{\top} $. Often, additional combinatorial constraints such as cardinality constraint $\left(|B|\leq k\right)$ or matroid constraint ($B$ is a basis of a matroid defined on $[n]$) must be satisfied by the chosen set of vectors. The minimum eigenvalue problem with matroid constraints models a wide variety of problems including the Santa Clause problem, the E-design problem, and the constructive Kadison-Singer problem. In this paper, we give a randomized algorithm that finds a set $B\subseteq [n]$ subject to any matroid constraint whose minimum eigenvalue is at least $(1-\epsilon)$ times the optimum, with high probability. The running time of the algorithm is $O\left( n^{O(d\log(d)/\epsilon^2)}\right)$. In particular, our results give a polynomial time asymptotic scheme when the dimension of the vectors is constant. Our algorithm uses a convex programming relaxation of the problem after guessing a rescaling which allows us to apply pipage rounding and matrix Chernoff inequalities to round to a good solution. The key new component is a structural lemma which enables us to "guess'' the appropriate rescaling, which could be of independent interest. Our approach generalizes the approximation guarantee to monotone, homogeneous functions and as such we can maximize $\det(\sum_{i\in B} v_i v_i^\top)^{1/d}$, or minimize any norm of the eigenvalues of the matrix $\left(\sum_{i\in B} v_i v_i^\top\right)^{-1} $, with the same running time under some mild assumptions. As a byproduct, we also get a simple algorithm for an algorithmic version of Kadison-Singer problem.

In an instance of the weighted Nash Social Welfare problem, we are given a set of $m$ indivisible items, $\mathscr{G}$, and $n$ agents, $\mathscr{A}$, where each agent $i \in \mathscr{A}$ has a valuation $v_{ij}\geq 0$ for each item $j\in \mathscr{G}$. In addition, every agent $i$ has a non-negative weight $w_i$ such that the weights collectively sum up to $1$. The goal is to find an assignment $\sigma:\mathscr{G}\rightarrow \mathscr{A}$ that maximizes $\prod_{i\in \mathscr{A}} \left(\sum_{j\in \sigma^{-1}(i)} v_{ij}\right)^{w_i}$, the product of the weighted valuations of the players. When all the weights equal $\frac1n$, the problem reduces to the classical Nash Social Welfare problem, which has recently received much attention. In this work, we present a $5\cdot\exp\left(2\cdot D_{\text{KL}}(\mathbf{w}\, ||\, \frac{\vec{\mathbf{1}}}{n})\right) = 5\cdot\exp\left(2\log{n} + 2\sum_{i=1}^n w_i \log{w_i}\right)$-approximation algorithm for the weighted Nash Social Welfare problem, where $D_{\text{KL}}(\mathbf{w}\, ||\, \frac{\vec{\mathbf{1}}}{n})$ denotes the KL-divergence between the distribution induced by $\mathbf{w}$ and the uniform distribution on $[n]$. We show a novel connection between the convex programming relaxations for the unweighted variant of Nash Social Welfare presented in \cite{cole2017convex, anari2017nash}, and generalize the programs to two different mathematical programs for the weighted case. The first program is convex and is necessary for computational efficiency, while the second program is a non-convex relaxation that can be rounded efficiently. The approximation factor derives from the difference in the objective values of the convex and non-convex relaxation.

Induced magnetospheres form around planetary bodies with atmospheres through the interaction of the solar wind with their ionosphere. Induced magnetospheres are highly dependent on the solar wind conditions and have only been studied with single spacecraft missions in the past. This gap in knowledge could be addressed by a multi-spacecraft plasma mission, optimized for studying global spatial and temporal variations in the magnetospheric system around Venus, which hosts the most prominent example of an induced magnetosphere in our solar system. The MVSE mission comprises four satellites, of which three are identical scientific spacecraft, carrying the same suite of instruments probing different regions of the induced magnetosphere and the solar wind simultaneously. The fourth spacecraft is the transfer vehicle which acts as a relay satellite for communications at Venus. In this way, changes in the solar wind conditions and extreme solar events can be observed, and their effects can be quantified as they propagate through the Venusian induced magnetosphere. Additionally, energy transfer in the Venusian induced magnetosphere can be investigated. The scientific payload includes instrumentation to measure the magnetic field, electric field, and ion-electron velocity distributions. This study presents the scientific motivation for the mission as well as requirements and the resulting mission design. Concretely, a mission timeline along with a complete spacecraft design, including mass, power, communication, propulsion and thermal budgets are given. This mission was initially conceived at the Alpbach Summer School 2022 and refined during a week-long study at ESAs Concurrent Design Facility in Redu, Belgium, Comment: 23 pages, 5 figures, Submitted to Acta Astronautica

The diffusion of gas through porous material is important to understand the physical processes underlying cometary activity. We study the diffusion of a rarefied gas (Knudsen regime) through a packed bed of monodisperse spheres via experiments and numerical modelling, providing an absolute value of the diffusion coefficient and compare it to published analytical models. The experiments are designed to be directly comparable to numerical simulations, by using precision steel beads, simple geometries, and a trade-off of the sample size between small boundary effects and efficient computation. For direct comparison, the diffusion coefficient is determined in Direct Simulation Monte Carlo (DSMC) simulations, yielding a good match with experiments. This model is further-on used on a microscopic scale, which cannot be studied in experiments, to determine the mean path of gas molecules and its distribution, and compare it against an analytical model. Scaling with sample properties (particle size, porosity) and gas properties (molecular mass, temperature) is consistent with analytical models. As predicted by these, results are very sensitive on sample porosity and we find that a tortuosity $q(\varepsilon)$ depending linearly on the porosity $\varepsilon$ can well reconcile the analytical model with experiments and simulations. Mean paths of molecules are close to those described in the literature, but their distribution deviates from the expectation for small path lengths. The provided diffusion coefficients and scaling laws are directly applicable to thermophysical models of idealised cometary material., Comment: accepted by MNRAS

In this note, we prove that the realization of associahedron discovered by Arkani-Hamed, Bai, He, and Yun (ABHY) is a positive geometry for tree-level S-matrix of scalars which have no color and which interact via cubic coupling. More in detail, we consider diffeomorphic images of the ABHY associahedron. The diffeomorphisms are linear maps parametrized by the right cosets of the Dihedral group on n elements. The set of all the boundaries associated with these copies of ABHY associahedron exhaust all the simple poles. We prove that the sum over the diffeomorphic copies of ABHY associahedron is a positive geometry and the total volume obtained by summing over all the dual associahedra is proportional to the tree-level S matrix of (massive or massless) scalar particles with cubic coupling. We then provide non-trivial evidence that the projection of the planar scattering forms parametrized by the Stokes polytope on these realizations of the associahedron leads to the tree-level amplitudes of scalar particles, which interact via quartic coupling. Our results build on ideas laid out in our previous works, leading to further evidence that a large class of positive geometries which are diffeomorphic to the ABHY associahedron defines an ``amplituhedron" for a tree-level S matrix of some local and unitary scalar theory. We also highlight a fundamental obstruction in applying these ideas to discover positive geometry for the one loop integrand when propagating states have no color., Comment: 33 Pages, 4 Figures

Determinant maximization provides an elegant generalization of problems in many areas, including convex geometry, statistics, machine learning, fair allocation of goods, and network design. In an instance of the determinant maximization problem, we are given a collection of vectors $v_1,\ldots, v_n \in \mathbb{R}^d$, and the goal is to pick a subset $S\subseteq [n]$ of given vectors to maximize the determinant of the matrix $\sum_{i \in S} v_iv_i^\top$, where the picked set of vectors $S$ must satisfy some combinatorial constraint such as cardinality constraint ($|S| \leq k$) or matroid constraint ($S$ is a basis of a matroid defined on $[n]$). In this work, we give a combinatorial algorithm for the determinant maximization problem under a matroid constraint that achieves $O(d^{O(d)})$-approximation for any matroid of rank $r\geq d$. This complements the recent result of~\cite{BrownLPST22} that achieves a similar bound for matroids of rank $r\leq d$, relying on a geometric interpretation of the determinant. Our result matches the best-known estimation algorithms~\cite{madan2020maximizing} for the problem, which could estimate the objective value but could not give an approximate solution with a similar guarantee. Our work follows the framework developed by~\cite{BrownLPST22} of using matroid intersection based algorithms for determinant maximization. To overcome the lack of a simple geometric interpretation of the objective when $r \geq d$, our approach combines ideas from combinatorial optimization with algebraic properties of the determinant. We also critically use the properties of a convex programming relaxation of the problem introduced by~\cite{madan2020maximizing}.

We study a novel asymptotic limit of massive scalar fields in nongravitational quantum field theories in four-dimensional flat space. We foliate the spacetime into a set of dS$_3$ slices that are spacelike to, and at a constant proper distance from, an arbitrarily chosen origin, and study the boundary dS$_3$ obtained in the infinite-distance limit. Massive bulk fields have an exponentially small tail in this limit, and by stripping off this tail we obtain observables that are intrinsic to the boundary dS$_3$. A single massive field in the bulk can be decomposed into an infinite set of dS$_3$ fields, and the Minkowski vacuum corresponds to the Euclidean vacuum for these fields. Our procedure for extrapolating bulk observables induces potential singularities in boundary correlators but we show how they can be cured in the free theory by smearing the boundary operators. We show that by integrating boundary operators with suitable smearing functions it is possible to reconstruct all local bulk operators in the free theory. We argue, using perturbation theory, that our extrapolation procedure continues to be well defined in the presence of interactions. We demonstrate a relationship between the width of the boundary smearing function and the localization of the bulk field. We study other interesting properties of the boundary algebra including the action of global translations and the manner in which local bulk interactions are encoded on the boundary., Comment: 54 pages; 4 figures

Determinant maximization problem gives a general framework that models problems arising in as diverse fields as statistics \cite{pukelsheim2006optimal}, convex geometry \cite{Khachiyan1996}, fair allocations\linebreak \cite{anari2016nash}, combinatorics \cite{AnariGV18}, spectral graph theory \cite{nikolov2019proportional}, network design, and random processes \cite{kulesza2012determinantal}. In an instance of a determinant maximization problem, we are given a collection of vectors $U=\{v_1,\ldots, v_n\} \subset \RR^d$, and a goal is to pick a subset $S\subseteq U$ of given vectors to maximize the determinant of the matrix $\sum_{i\in S} v_i v_i^\top $. Often, the set $S$ of picked vectors must satisfy additional combinatorial constraints such as cardinality constraint $\left(|S|\leq k\right)$ or matroid constraint ($S$ is a basis of a matroid defined on the vectors). In this paper, we give a polynomial-time deterministic algorithm that returns a $r^{O(r)}$-approximation for any matroid of rank $r\leq d$. This improves previous results that give $e^{O(r^2)}$-approximation algorithms relying on $e^{O(r)}$-approximate \emph{estimation} algorithms \cite{NikolovS16,anari2017generalization,AnariGV18,madan2020maximizing} for any $r\leq d$. All previous results use convex relaxations and their relationship to stable polynomials and strongly log-concave polynomials. In contrast, our algorithm builds on combinatorial algorithms for matroid intersection, which iteratively improve any solution by finding an \emph{alternating negative cycle} in the \emph{exchange graph} defined by the matroids. While the $\det(.)$ function is not linear, we show that taking appropriate linear approximations at each iteration suffice to give the improved approximation algorithm.

Building on the prior work in [1] we locate a family of positive geometries in the kinematic space which are a specific class of convex realisations of the associahedron. These realisations are obtained by scaling and translating the kinematic space associahedron discovered by by Arkani-Hamed, Bai, He and Yan (ABHY). We call the resulting polytopes, deformed realisations of the associahedron. The deformed realisations shed new light on the CHY formula. One of the striking discoveries in [2] was the fact that the CHY scattering equations generate diffeomorphism between the (compactified) CHY moduli space and the ABHY associahedron. As we argue, the deformed realisation of the associahedron can also be interpreted as an diffeomorphic image of the CHY moduli space under scattering equations that we call deformed scattering equations. The canonical form in the kinematic space is thus once again the push-forward of the Parke-Taylor form . A natural off-shoot of our analysis is the universality of the Parke-Taylor form as a CHY Integrand for a class of (tree-level and planar) multi-scalar field amplitudes. These ideas help us in proving the existence of positive geometries for certain specific multi-scalar interactions. We prove that in a field theory with a massless and a massive bi-adjoint scalar fields which interact via cubic interaction, the tree-level S-matrix with massless external states and at most one massive propagator is a weighted sum over the canonical forms defined by certain deformed realisations of the associahedron. Finally, we show that these ideas admit an extension to one-loop. In particular, the one loop S-matrix integrand with at most one massive propagator is a weighted sum over canonical forms of a family of deformed realisations of the type-D cluster polytope, discovered in [3,4]., Comment: 39 pages, 8 figures

In the commercial aviation domain, there are a large number of documents, like, accident reports (NTSB, ASRS) and regulatory directives (ADs). There is a need for a system to access these diverse repositories efficiently in order to service needs in the aviation industry, like maintenance, compliance, and safety. In this paper, we propose a Knowledge Graph (KG) guided Deep Learning (DL) based Question Answering (QA) system for aviation safety. We construct a Knowledge Graph from Aircraft Accident reports and contribute this resource to the community of researchers. The efficacy of this resource is tested and proved by the aforesaid QA system. Natural Language Queries constructed from the documents mentioned above are converted into SPARQL (the interface language of the RDF graph database) queries and answered. On the DL side, we have two different QA models: (i) BERT QA which is a pipeline of Passage Retrieval (Sentence-BERT based) and Question Answering (BERT based), and (ii) the recently released GPT-3. We evaluate our system on a set of queries created from the accident reports. Our combined QA system achieves 9.3% increase in accuracy over GPT-3 and 40.3% increase over BERT QA. Thus, we infer that KG-DL performs better than either singly., Comment: LREC 2022 Main Conference Accepted Paper

We study a unified approach and algorithm for constructive discrepancy minimization based on a stochastic process. By varying the parameters of the process, one can recover various state-of-the-art results. We demonstrate the flexibility of the method by deriving a discrepancy bound for smoothed instances, which interpolates between known bounds for worst-case and random instances.