This paper reports a set of recent measurements and searches from the LHCb experiment that probe both the Standard Model and possible new particles. Using proton–proton collision data collected in different years, LHCb measured properties of the Z and W bosons and of the top quark in the detector’s forward region. The collaboration also searched for two types of hypothetical “mediators” that could connect ordinary matter to a dark sector: axion‑like particles (ALPs) and heavy neutral leptons (HNLs). These searches found no signals and set new exclusion limits.

What the researchers did: LHCb used several datasets. A Z‑boson mass was measured from Z→µ+µ− decays in 2016 data with 1.7 inverse femtobarns, giving mZ = 91,185.7 ± 8.3 (stat) ± 3.9 (syst) MeV. A model‑independent proof‑of‑concept W‑boson mass determination was performed using 100 inverse picobarns at 5.02 TeV, reporting mW = 80,369 ± 130 (exp) ± 33 (theory) MeV; the paper notes this does not supersede the earlier 13 TeV result but demonstrates a novel method. Production cross sections for W bosons were measured at 13 TeV using 5.1 inverse femtobarns, and top‑quark production was measured for the first time in the forward region using 5.4 inverse femtobarns of 2016–2018 data.

How the measurements work at a high level: LHCb’s detector covers particles produced at small angles to the beam, the so‑called forward region. That acceptance gives access to partons (the quarks and gluons inside the proton) at unusually low or high momentum fraction, which helps constrain parton distribution functions (PDFs) that feed into many theoretical predictions. The Z mass was extracted from the two muon final state after careful momentum calibration using known resonances. The W mass study used differential muon distributions and a strategy that separates detector effects from theoretical signal modelling so the signal model can be updated independently. The top analysis identified b‑quark jets with a new deep‑learning tagger that improves b‑ and c‑tagging efficiencies by 11–53% compared to older methods.