arXiv subjects

Paul Richmond

Publications and source records attributed to Paul Richmond.

At least 19 recordsLinked to original sources

arXiv2026

FeynTune: Large Language Models for High-Energy Theory

We present specialized Large Language Models for theoretical High-Energy Physics, obtained as 20 fine-tuned variants of the 8-billion parameter Llama-3.1 model. Each variant was trained on arXiv abstracts (through August 2024) from different combinations of hep-th, hep-ph and gr-qc. For a comparative study, we also trained models on datasets that contained abstracts from disparate fields such as the q-bio and cs categories. All models were fine-tuned using two distinct Low-Rank Adaptation fine-tuning approaches and varying dataset sizes, and outperformed the base model on hep-th abstract completion tasks. We compare performance against leading commercial LLMs (ChatGPT, Claude, Gemini, DeepSeek) and derive insights for further developing specialized language models for High-Energy Theoretical Physics.

cs.CL↗

arXiv2023

GPU-parallelisation of wavelet-based grid adaptation for fast finite volume modelling: application to shallow water flows

Wavelet-based grid adaptation driven by the "multiresolution analysis" (MRA) of the Haar wavelet (HW) allows to devise an adaptive first-order finite volume (FV1) model (HWFV1) that can readily preserve the modelling fidelity of its reference uniform-grid FV1 counterpart. However, the MRA incurs a high computational cost as it involves "encoding" (coarsening), "decoding" (refining), analysing and traversing modelled data across a deep hierarchy of nested, uniform grids. GPU-parallelisation of the MRA is needed to reduce its computational cost, but its algorithmic structure (1) hinders coalesced memory access on the GPU, and (2) involves an inherently sequential tree traversal problem. This work redesigns the algorithmic structure of the MRA in order to parallelise it on the GPU, addressing (1) by applying Z-order space-filling curves and addressing (2) by adopting a parallel tree traversal algorithm. This results in a GPU-parallelised HWFV1 model (GPU-HWFV1). GPU-HWFV1 is verified against its CPU predecessor (CPU-HWFV1) and its GPU-parallelised reference uniform-grid counterpart (GPU-FV1) over five shallow water flow test cases. GPU-HWFV1 preserves the modelling fidelity of GPU-FV1 while being up to 30 times faster. Compared to CPU-HWFV1, it is up to 200 times faster, suggesting the GPU-parallelised MRA could be used to speed up other FV1 models.

cs.CE↗

arXiv2022

6D (2,0) Bootstrap with soft-Actor-Critic

We study numerically the 6D (2,0) superconformal bootstrap using the soft-Actor-Critic (SAC) algorithm as a stochastic optimizer. We focus on the four-point functions of scalar superconformal primaries in the energy-momentum multiplet. Starting from the supergravity limit, we perform searches for adiabatically varied central charges and derive two curves for a collection of 80 CFT data (70 of these data correspond to unprotected long multiplets and 10 to protected short multiplets). We conjecture that the two curves capture the A- and D-series (2,0) theories. Our results are competitive when compared to the existing bounds coming from standard numerical bootstrap methods, and data obtained using the OPE inversion formula. With this paper we are also releasing our Python implementation of the SAC algorithm, BootSTOP. The paper discusses the main functionality features of this package.

hep-th↗

arXiv2022

Five-Dimensional Path Integrals for Six-Dimensional Conformal Field Theories

In this paper we derive Ward-Takahashi identities from the path integral of supersymmetric five-dimensional field theories with an $SU(1,3)$ spacetime symmetry in the presence of instantons. We explicitly show how $SU(1,3)$ is enhanced to $SU(1,3)\times U(1)$ where the additional $U(1)$ acts non-perturbatively. Solutions to such Ward-Takahashi identities were previously obtained from correlators of six-dimensional Lorentzian conformal field theories but where the instanton number was replaced by the momentum along a null direction. Here we study the reverse procedure whereby we construct correlation functions out of towers of five-dimensional operators which satisfy the Ward-Takahashi identities of a six-dimensional conformal field theory. This paves the way to computing observables in six dimensions using five-dimensional path integral techniques. We also argue that, once the instanton sector is included into the path integral, the coupling of the five-dimensional Lagrangian must be quantised, leaving no free continuous parameters.

hep-th↗

arXiv2021

The Supersymmetry of Higher-Derivative Supergravity in AdS$_4$ Holography

An action for the higher-derivative corrections to minimal gauged supergravity in four dimensions has been recently proposed. We demonstrate that the supersymmetric solutions of this model are those of the two-derivative action, and investigate some of their properties. In particular, we prove a formula for the renormalised on-shell action in terms of contributions from fixed points of a $U(1)$ action, and confirm that it is invariant under deformations which preserve the boundary almost contact structure.

hep-th↗

arXiv2021

Instanton Worldlines in Five-Dimensional $Ω$-Deformed Gauge Theory

We discuss the Bosonic sector of a class of supersymmetric non-Lorentzian five-dimensional gauge field theories with an $SU(1,3)$ conformal symmetry. These actions have a Lagrange multiplier which imposes a novel $Ω$-deformed anti-self-dual gauge field constraint. Using a generalised 't Hooft ansatz we find the constraint equation linearizes allowing us to construct a wide class of explicit solutions. These include finite action configurations that describe worldlines of anti-instantons which can be created and annihilated. We also describe the dynamics on the constraint surface.

hep-th↗

arXiv2020

Five-Dimensional Non-Lorentzian Conformal Field Theories and their Relation to Six-Dimensions

We study correlation functions in five-dimensional non-Lorentzian theories with an $SU(1,3)$ conformal symmetry. Examples of such theories have recently been obtained as $Ω$-deformed Yang-Mills Lagrangians arising from a null reduction of six-dimensional superconformal field theories on a conformally compactified Minkowski space. The correlators exhibit a rich structure with many novel properties compared to conventional correlators in Lorentzian conformal field theories. Moreover, identifying the instanton number with the Fourier mode number of the dimensional reduction offers a hope to formulate six-dimensional conformal field theories in terms of five-dimensional Lagrangian theories. To this end we show that the Fourier decompositions of six-dimensional correlation functions solve the Ward identities of the the $SU(1,3)$ symmetry, although more general solutions are possible. Conversely we illustrate how one can reconstruct six-dimensional correlation functions from those of a five-dimensional theory, and do so explicitly at 2- and 3-points. We also show that, in a suitable decompactification limit $Ω\to 0$, the correlation functions become those of the DLCQ description.

hep-th↗

arXiv2020

Boundary Conditions in Topological AdS$_4$/CFT$_3$

We revisit the construction in four-dimensional gauged $Spin(4)$ supergravity of the holographic duals to topologically twisted three-dimensional $\mathcal{N}=4$ field theories. Our focus in this paper is to highlight some subtleties related to preserving supersymmetry in AdS/CFT, namely the inclusion of finite counterterms and the necessity of a Legendre transformation to find the dual to the field theory generating functional. Studying the geometry of these supergravity solutions, we conclude that the gravitational free energy is indeed independent from the metric of the boundary, and it vanishes for any smooth solution.

hep-th↗

arXiv2020

Simulating Crowds and Autonomous Vehicles

Understanding how people view and interact with autonomous vehicles is important to guide future directions of research. One such way of aiding understanding is through simulations of virtual environments involving people and autonomous vehicles. We present a simulation model that incorporates people and autonomous vehicles in a shared urban space. The model is able to simulate many thousands of people and vehicles in real-time. This is achieved by use of GPU hardware, and through a novel linear program solver optimized for large numbers of problems on the GPU. The model is up to 30 times faster than the equivalent multi-core CPU model.

cs.RO↗

arXiv2020

Agent-based modelling of pedestrian responses during flood emergency: mobility behavioural rules and implications for flood risk analysis

An agent-based model (ABM) for simulating flood-pedestrian interaction is augmented to particularly explore more realistic responses of evacuating pedestrians during flooding. Pedestrian agents within the ABM follow navigation rules of governing their movement in dry areas. When in floodwater, pedestrian agents are assigned extra behavioural rules to factor in their states of stability and walking speed, and their different body height and weight. The ABM is applied to replicate a synthetic test case of a flooded shopping centre, considering increasingly sophisticated configuration modes for the behavioural rules of the evacuating pedestrians. Simulation results are analysed based on spatial and temporal indicators informing on the dynamic variations of flood risk states of flooded pedestrians in terms of a commonly used flood Hazard Rating (HR) metric, variable walking speed, and instability due to toppling and/or sliding. Our analysis reveal significantly prolonged evacuation times and risk exposure levels as stability and walking speed behavioural rules become more sophisticated. It also allows to identify more conservative HR thresholds due to unstable pedestrians, and a new formula to directly estimate walking speed states as function of HR for stable pedestrian in floodwater. Accompanying details for software accessibility are provided.

physics.soc-ph↗

arXiv2020

Bosonic Symmetries of $(2,0)$ DLCQ Field Theories

We investigate symmetries of the six-dimensional $(2,0)$ theory reduced along a compact null direction. The action for this theory was deduced by considering M-theory on $AdS_7 \times S^4$ and reducing the $AdS_7$ factor along a time-like Hopf fibration which breaks one quarter of the supersymmetry and reduces the isometry group from $SO(6,2)$ to $SU(3,1)$. The boundary theory was previously shown to have 24 supercharges and a Lifshitz scaling symmetry. In this paper, we show that it has four boost-like symmetries and an additional conformal symmetry which furnish a representation of $SU(3,1)$ when combined with the other bosonic symmetries, providing a nontrivial check of the holographic correspondence.

hep-th↗

arXiv2020

Agent-based simulator of dynamic flood-people interactions

This paper presents a new simulator for dynamic modelling of interactions between flooding and people in crowded areas. The simulator is developed in FLAMEGPU (a Flexible Large scale Agent-based Modelling Environment for the GPU), which allows to model multiple agent interactions while benefitting from the speed-up of GPUs. Flooding variables including terrain data are represented by a hydrodynamic Agent-Based Model (ABM) that is based on a non-sequential implementation of a robust Finite Volume (FV) solver of the Shallow Water Equations (SWEs). People movements are represented by a pedestrian ABM adopting force-based walking rules. The hydrodynamic ABM is coupled to the pedestrian ABM according to risk-to-life thresholds reported by the UK Environment Agency (EA). A hypothetical case study of a crowded shopping centre is proposed and used to assess the dynamic coupling ability of the simulator. Flooding into the shopping centre is induced based on realistic inflow conditions, and the simulator is applied considering two scenarios: evacuation without advanced warning and intervention with an advanced warning of 12 hours. Results show that the simulator can produce detailed statistics of spatiotemporal people states during evacuation, and is useful to plan safe and effective people intervention to deploy a sandbag-based temporary barriers.

physics.soc-ph↗

arXiv2020

AdS$_7$ Black-Hole Entropy and 5D $\mathcal{N}=2$ Yang-Mills

We generalise the work of 1810.11442 for the case of AdS$_7$/CFT$_6$. Starting from the 2-equivalent charge, 3-equivalent rotation non-extremal black-hole solution in 7D gauged supergravity, we consider the supersymmetric and then the extremal limit and evaluate the associated thermodynamic quantities. Away from extremality, the black-hole solution becomes complex. The entropy is then given by the Legendre transform of the on-shell action with respect to two complex chemical potentials subject to a constraint. At the conformal boundary we derive the dual background and evaluate the corresponding partition function for the $A_{N-1}$ 6D (2,0) theory at large $N$ in a Cardy-like limit. This is carried out via a 5D $\mathcal N=2$ super Yang-Mills calculation on $S^5$. The gravitational on-shell action is found to be exactly reproduced by the boundary partition function at large $N$. We argue that this agreement puts strong constraints on the form of possible higher-derivative corrections to the 5D gauge theory that is used in the $S^5$ evaluation.

hep-th↗

arXiv2019

Topological AdS/CFT and the $Ω$ deformation

In this note, we define a holographic dual to four-dimensional superconformal field theories formulated on arbitrary Riemannian manifolds equipped with a Killing vector. Moreover, assuming smoothness of the bulk solution, we study the variation of the holographically renormalized supergravity action in the class of metrics on the boundary four-manifold with a prescribed isometry.

hep-th↗

arXiv2019

Fast Simulation of Crowd Collision Avoidance

Real-time large-scale crowd simulations with realistic behavior, are important for many application areas. On CPUs, the ORCA pedestrian steering model is often used for agent-based pedestrian simulations. This paper introduces a technique for running the ORCA pedestrian steering model on the GPU. Performance improvements of up to 30 times greater than a multi-core CPU model are demonstrated. This improvement is achieved through a specialized linear program solver on the GPU and spatial partitioning of information sharing. This allows over 100,000 people to be simulated in real time (60 frames per second).

cs.RO↗

arXiv2019

Topological AdS/CFT

We define a holographic dual to the Donaldson-Witten topological twist of $\mathcal{N}=2$ gauge theories on a Riemannian four-manifold. This is described by a class of asymptotically locally hyperbolic solutions to $\mathcal{N}=4$ gauged supergravity in five dimensions, with the four-manifold as conformal boundary. Under AdS/CFT, minus the logarithm of the partition function of the gauge theory is identified with the holographically renormalized supergravity action. We show that the latter is independent of the metric on the boundary four-manifold, as required for a topological theory. Supersymmetric solutions in the bulk satisfy first order differential equations for a twisted $Sp(1)$ structure, which extends the quaternionic Kahler structure that exists on any Riemannian four-manifold boundary. We comment on applications and extensions, including generalizations to other topological twists.

hep-th↗

arXiv2019

Non-Lorentzian M5-brane Theories from Holography

M-theory on $AdS_7 \times S^4$ admits a description where the $AdS_7$ factor is constructed as a timelike Hopf fibration over a non-compact three dimensional complex projective space $\tilde{\mathbb{CP}}^3$. We consider the worldvolume theory for M5-branes at a fixed $\tilde{\mathbb{CP}}^3$ radius which, after reduction along the timelike fibre, is given by an $Ω$-deformed Yang-Mills theory with eight supercharges. Taking the radius to infinity then induces a classical RG flow. We construct the fixed point action which has an enhanced 24 supercharges and which can be understood as the $(2,0)$ theory of M5-branes on flat space reduced along a compact null Killing direction.

hep-th↗

arXiv2019

Two-Dimensional Batch Linear Programming on the GPU

This paper presents a novel, high-performance, graphical processing unit-based algorithm for efficiently solving two-dimensional linear programs in batches. The domain of two-dimensional linear programs is particularly useful due to the prevalence of relevant geometric problems. Batch linear programming refers to solving numerous different linear programs within one operation. By solving many linear programs simultaneously and distributing workload evenly across threads, graphical processing unit utilization can be maximized. Speedups of over 22 times and 63 times are obtained against state-of-the-art graphics processing unit and CPU linear program solvers, respectively.

cs.DC↗