The LHC has recently completed its Run 3 data taking, marking the transition toward the High-Luminosity LHC era. While searches for physics beyond the Standard Model have already excluded many models at TeV scales, further gains from luminosity alone become increasingly limited. This motivates the development of new experimental and analysis techniques that can access signatures and phase-space regions beyond the reach of conventional searches. This review summarizes recent CMS searches for exotic physics, and then focuses on a selection of analyses that employ particularly novel strategies. These include level-1 trigger data scouting at the full 40 MHz collision rate for heavy long-lived charged particles, dedicated reconstruction techniques for highly collimated lepton pairs in heavy gauge-boson searches, and high-level trigger data scouting for searches for soft unclustered energy patterns, a signature that had not previously been explored at the LHC. These novel techniques reviewed here provide examples of promising directions for future searches during the HL-LHC era.
This work investigates the thermodynamics of non-stationary anti-de Sitter (adS) black holes and their gauge-gravity duals. To model a non-equilibrium process, we employ a time-dependent Vaidya-adS metric to describe the collapse of a null fluid and the subsequent thermalization of the dual field theory. We introduce an effective version of free energy as a quasilocal probe to be used in a dynamical setup. By analyzing how this quantity evolves, we show that a first-order phase transition occurs when the final black-hole mass exceeds the limit established by the Hawking–Page temperature. Furthermore, we demonstrate that the final state only approaches the thermal behavior of the N=4 supersymmetric Yang–Mills theory for mass states much larger than the adS radius. Finally, we compute the boundary energy–momentum tensor using holographic renormalization and compare the resulting energy density with the strong-coupling limit. This demonstrates that the dual description provides a consistent framework for both the equilibrium and non-equilibrium thermodynamics of strongly coupled gauge theories.
We present GalSpecCNN, a probabilistic deep-learning framework for estimating fiber-scale stellar mass (M*) and specific star formation rate (sSFR) from LAMOST galaxy spectra. The model uses a one-dimensional convolutional neural network to extract spectral features, followed by a probabilistic regression head that predicts a diagonal bivariate Gaussian distribution. Aleatoric uncertainty is modeled through the predicted variances, while epistemic uncertainty is approximated using Monte Carlo Dropout. On the test set, GalSpecCNN achieves residual scatters of 0.226 dex for M* and 0.319 dex for sSFR, with small mean residuals. The empirical coverages of the nominal 68% predictive intervals are 68.60% for M* and 71.58% for sSFR, indicating good overall calibration with mildly conservative coverage for sSFR. We apply the trained model to approximately 1.16×105 LAMOST DR7 galaxy spectra and construct a catalog of M* and sSFR estimates, together with their associated predictive intervals. This work provides a practical framework for uncertainty-aware estimation of galaxy physical properties from large spectroscopic surveys.
Direct kinematic measurements of cosmic acceleration via the cosmological redshift drift (the Sandage–Loeb effect) provide a powerful, model-independent avenue to map the expansion history of the Universe. In this work, which extends the forecasting framework developed in a recent study, we evaluate the capability of the Square Kilometre Array (SKA) to constrain three physically motivated, quantum-gravity inspired holographic dark energy (HDE) paradigms: Standard HDE (with a future event horizon cutoff), Tsallis HDE (THDE), and Rényi HDE (RHDE). Utilizing simulated neutral hydrogen (HI) 21 cm emission lines from a census of over 109 galaxies up to z=1.0 and 21 cm absorption lines in 1.8×103 damped Lyman-α (DLA) systems, we simulate mock velocity drift observations (v˙) at high spectral resolutions (Δν=0.001 Hz and 0.002 Hz) over a semi-annual (Δt=0.5 year) cadence baseline. We contrast these forecasts with current real-world constraints obtained from the joint Markov chain Monte Carlo (MCMC) likelihood analysis of the Pantheon SNe Ia and DESI DR2 BAO compilations. Our results reveal a profound limitation of integrated geometric probes: while current SNe Ia and BAO data suffer from severe parameter degeneracies—leaving the non-additive entropy scaling exponents (δ and α) completely unconstrained as open vertical bands—the simulated SKA redshift drift successfully breaks these degeneracies. From emission-line observations at 0.001 Hz spectral resolution, we obtain marginalized 1σ constraints of σc=±0.021 for the standard holographic parameter, σδ=±0.035 for the Tsallis entropy index, and σα=±0.007 for the Rényi parameter. Crucially, we identify and resolve a critical duplicate-plotting mathematical error present in the existing redshift-drift literature by providing the mathematically correct physical scaling and peak structures for both the dimensionless redshift drift (Sz) and the physical velocity drift (Sv). We conclude that the SKA will serve as a premier instrument for testing the holographic principle and non-extensive thermodynamics at cosmological scales.
Integrated-light spectral energy distribution modeling is widely used to infer the star-formation and assembly histories of galaxies that cannot be resolved into individual stars. However, existing approaches are often discussed primarily in terms of particular fitting codes, star-formation-history parameterizations, or inference algorithms. In this Review, we formulate the recovery of galaxy evolution histories from integrated spectral energy distributions as a unified inverse problem. We separate the physical spectral-generation operator from the observational operator and examine the resulting information loss through non-identifiability, singular-value structure, null directions, effective resolution, regularization, and model discrepancy. We then classify parametric and nonparametric star-formation histories, PCA, MOPED, VESPA, non-negative matrix factorization, deep learning, and simulation-based inference within a common five-component framework consisting of the representation space, forward operator, physical or statistical constraints, inference method, and uncertainty assessment. On this basis, we introduce information-driven adaptive representation as a general design principle in which the complexity of the recovered history is matched to the information supported by the observations. Finally, we extend the framework from star-formation histories to coupled galaxy-evolution states involving chemical enrichment, dust evolution, interstellar-medium conditions, and radiative transfer, and outline a three-layer research program linking controlled mock experiments, inverse-problem theory, and physical forward modeling.
