This paper presents the observation of four-top-quark ($$t\bar{t}t\bar{t}$$ t t ¯ t t ¯ ) production in proton-proton collisions at the LHC. The analysis is performed using an integrated luminosity of 140 $$\hbox {fb}^{-1}$$ fb - 1 at a centre-of-mass energy of 13 TeV collected using the ATLAS detector. Events containing two leptons with the same electric charge or at least three leptons (electrons or muons) are selected. Event kinematics are used to separate signal from background through a multivariate discriminant, and dedicated control regions are used to constrain the dominant backgrounds. The observed (expected) significance of the measured $$t\bar{t}t\bar{t}$$ t t ¯ t t ¯ signal with respect to the standard model (SM) background-only hypothesis is 6.1 (4.3) standard deviations. The $$t\bar{t}t\bar{t}$$ t t ¯ t t ¯ production cross section is measured to be $$22.5^{+6.6}_{-5.5}$$ 22 . 5 - 5.5 + 6.6  fb, consistent with the SM prediction of $$12.0 \pm 2.4$$ 12.0 ± 2.4 fb within 1.8 standard deviations. Data are also used to set limits on the three-top-quark production cross section, being an irreducible background not measured previously, and to constrain the top-Higgs Yukawa coupling and effective field theory operator coefficients that affect $$t\bar{t}t\bar{t}$$ t t ¯ t t ¯ production.

A search for forward proton scattering in association with light-by-light scattering mediated by an axion-like particle is presented, using the ATLAS Forward Proton spectrometer to detect scattered protons and the central ATLAS detector to detect pairs of outgoing photons. Proton-proton collision data recorded in 2017 at a centre-of-mass energy of $$ \sqrt{s} $$ s = 13 TeV were analysed, corresponding to an integrated luminosity of 14.6 fb−1. A total of 441 candidate events were selected. A search was made for a narrow resonance in the diphoton mass distribution, corresponding to an axion-like particle (ALP) with mass in the range 150–1600 GeV. No excess is observed above a smooth background. Upper limits on the production cross section of a narrow resonance are set as a function of the mass, and are interpreted as upper limits on the ALP production coupling constant, assuming 100% decay branching ratio into a photon pair. The inferred upper limit on the coupling constant is in the range 0.04–0.09 TeV−1 at 95% confidence level.

We present Turbo-Sim, a generalised autoencoder framework derived from principles of information theory that can be used as a generative model. By maximising the mutual information between the input and the output of both the encoder and the decoder, we are able to rediscover the loss terms usually found in adversarial autoencoders and generative adversarial networks, as well as various more sophisticated related models. Our generalised framework makes these models mathematically interpretable and allows for a diversity of new ones by setting the weight of each loss term separately. The framework is also independent of the intrinsic architecture of the encoder and the decoder thus leaving a wide choice for the building blocks of the whole network. We apply Turbo-Sim to a collider physics generation problem: the transformation of the properties of several particles from a theory space, right after the collision, to an observation space, right after the detection in an experiment.

Conditional generation is a subclass of generative problems where the output of the generation is conditioned by the attribute information. In this paper, we present a stochastic contrastive conditional generative adversarial network (InfoSCC-GAN) with an explorable latent space. The InfoSCC-GAN architecture is based on an unsupervised contrastive encoder built on the InfoNCE paradigm, an attribute classifier and an EigenGAN generator. We propose a novel training method, based on generator regularization using external or internal attributes every n-th iteration, using a pre-trained contrastive encoder and a pre-trained classifier. The proposed InfoSCC-GAN is derived based on an information-theoretic formulation of mutual information maximization between input data and latent space representation as well as latent space and generated data. Thus, we demonstrate a link between the training objective functions and the above information-theoretic formulation. The experimental results show that InfoSCC-GAN outperforms the "vanilla" Eigen-GAN in the image generation on AFHQ and CelebA datasets. In addition, we investigate the impact of discriminator architectures and loss functions by performing ablation studies. Finally, we demonstrate that thanks to the EigenGAN generator, the proposed framework enjoys a stochastic generation in contrast to vanilla deterministic GANs yet with the independent training of encoder, classifier, and generator in contrast to existing frameworks. Code, experimental results, and demos are available online at github.com/vkinakh/InfoSCC-GAN.