We introduce a potentially powerful new method of searching for new physics at the LHC, using autoencoders and unsupervised deep learning. The key idea of the autoencoder is that it learns to map "normal" events back to themselves, but fails to reconstruct "anomalous" events that it has never encountered before. The reconstruction error can then be used as an anomaly threshold. We demonstrate the effectiveness of this idea using QCD jets as background and boosted top jets and RPV gluino jets as signal. We show that a deep autoencoder can significantly improve signal over background when trained on backgrounds only, or even directly on data which contains a small admixture of signal. Finally we examine the correlation of the autoencoders with jet mass and show how the jet mass distribution can be stable against cuts in reconstruction loss. This may be important for estimating QCD backgrounds from data. As a test case we show how one could plausibly discover 400 GeV RPV gluinos using an autoencoder combined with a bump hunt in jet mass. This opens up the exciting possibility of training directly on actual data to discover new physics with no prior expectations or theory prejudice.

No abstract

No abstract

The ACME collaboration has recently announced a new constraint on the electron EDM, |d e |< 1.1×10 −29 e cm, from measurements of the ThO molecule. This is a powerful constraint on CP-violating new physics: even new physics generating the EDM at two loops is constrained at the multi-TeV scale. We interpret the bound in the context of different scenarios for new physics: a general order-of-magnitude analysis for both the electron EDM and the CP-odd electronnucleon coupling; 1-loop SUSY, probing sleptons above 10 TeV; 2-loop SUSY, probing multi-TeV charginos or stops; and finally, new physics that generates the EDM via the charm quark or top quark Yukawa couplings. In the last scenario, new physics generates a "QULE operator" (q f σ µνū f ) · ( σ µνē ), which in turn generates the EDM through RG evolution. If the QULE operator is generated at tree level, this corresponds to a previously studied leptoquark model. For the first time, we also classify scenarios in which the QULE operator is generated at one loop through a box diagram, which include SUSY and leptoquark models. The electron EDM bound is the leading constraint on a wide variety of theories of CP-violating new physics interacting with the Higgs boson or the top quark. We argue that any future nonzero measurement of an electron EDM will provide a strong motivation for constructing new colliders at the highest feasible energies. arXiv:1810.07736v1 [hep-ph] 17 Oct 2018 there are many motivations for searching for physics beyond the Standard Model, three of the most important are the matter-antimatter asymmetry of our universe, the existence of dark matter, and the fine-tuning puzzle of the Higgs boson mass. The matter-antimatter asymmetry clearly indicates a need for new CP-violating physics, which could first be detected through its indirect effect on the electron EDM. As we will discuss below, EDMs also have interesting connections with WIMP dark matter (in specific models) and with the fine-tuning problem.The possibility of testing heavy new physics through electric dipole moment measurements has been studied extensively; reviews include [11][12][13][14]. Here we attempt to briefly summarize some of the important history of the topic, with apologies for inevitable omissions. Some early theoretical studies of lepton EDMs appeared already in the 1970s [15,16]. Many of the early studies of CP violation in supersymmetric theories focused on the neutron EDM [17][18][19], but studies of the electron EDM in supersymmetry commenced [20] shortly after a suggestion of Gavela and Georgi that lepton EDMs could be effective probes of new physics [21]. Subsequently, a variety of additional sources of EDMs were studied, such as 3-gluon operators [22] or two-loop diagrams mediated by electroweak bosons [23,24]. A variety of new physics scenarios have been shown to predict interesting EDMs, including: stops in SUSY [25]; electroweakinos in SUSY [26] and specifically split SUSY [27,28]; two Higgs doublet models [24,[29][30][31]; SUSY beyond the MSSM [32,33]; and fermionic...

The chiral anomaly provides smoking-gun evidence of a new confining gauge theory. Motivated by a reported event excess in a diphoton invariant mass distribution at the LHC, we discuss a scenario that a pseudo-Nambu-Goldstone (PNG) boson of a new QCD-like theory is produced by gluon fusion and decays into a pair of the standard model gauge bosons. Despite the strong dynamics, the production cross section and the decay widths are determined by an anomaly matching condition. The excess can be explained by the PNG boson with mass of around 750 GeV. The model also predicts exotic hadrons such as a color-octet scalar and baryons. Some of them are within the reach of the LHC experiment.

No abstract

Abstract:Winter-forest processes affect global and local climates. The interception-sublimation fraction (F) of snowfall in forests is a substantial part of the winter water budget (up to 40%). Climate, weather-forecast and hydrological modellers incorporate increasingly realistic surface schemes into their models, and algorithms describing snow accumulation and snow-interception sublimation are now finding their way into these schemes. Spatially variable data for calibration and verification of wintertime dynamics therefore are needed for such modelling schemes. The value of F was determined from snow courses in open and forested areas in Hokkaido, Japan. The value of F was related to species and canopystructure measures such as closure, sky-view fraction (SVF ) and leaf-area index (LAI ). Forest structure was deduced from fish-eye photographs. The value of F showed a strong linear correlation to structure: F D 0Ð44 0Ð6 ð SVF for SVF < 0Ð72 and F D 0 for SVF > 0Ð72, and F D 0Ð11 LAI. These relationships seemed valid for evergreen conifers, larch trees, alder, birch and mixed deciduous stands. Forest snow accumulation S F could be estimated from snowfall in open fields S o and to LAI according to S F D S o (1 0Ð11 LAI ) as well as from SVF according to S F D S o (0Ð56 C 0Ð6 SVF) for SVF < 0Ð72. The value of S F was equal to S o for SVF values above 0Ð72. The value of sky-view fraction was correlated to the normalized difference snow index (NDSI ) using a Landsat-TM image for observation plots exceeding 1 ha. Variables F and S F were related to NDSI for these plots according to: F D 0Ð37NDSI C 0Ð29 and S F D S o 0Ð81 C 0Ð37NDSI . These relationships are somewhat hypothetical because plot-size limitation only allowed one sparse-forest observation of NDSI to be used. There is, therefore, a need to confirm these relationships with further studies.

No abstract