Are you one of the authors of this paper? Email us from your institutional or work email address mentioning this article's arXiv ID and we'll verify you and give you edit access to this page.
Are you an author who sees something we missed or got wrong? Write to us — the subject line already carries the paper id. We read every note and correct the write-up.
Are you an author of this paper and would prefer our analysis not to be published? Email us from your institutional address; once confirmed, we will take the page down.
A smart filter rejects spurious flashes and helps spot the rarest event in liquid xenon.
📄 Showing the "Simple" version — "Advanced" is not ready yet. Add it to favorites to help prioritize it.
In a dark mine under the Italian Alps, a tank holds three tons of liquid xenon. When a particle hits it, it knocks faint flashes of light and electron clouds out of the atoms. Following these tracks, physicists hope to see an ultra-rare decay: a xenon-136 nucleus turning into barium-136, emitting two electrons and no neutrinos. Such a process is forbidden in ordinary physics, and spotting it would upend our understanding of neutrinos.
The trouble is that similar tracks are left by gamma rays from the detector's own materials — steel, copper, cables. These background flashes mask the signal you're after, like crowd noise drowning out a single voice. Previously, the fix was crude: just throw away events with energies in a certain range. But that also discarded some useful decays.
The neural network looks not just at energy but at event shape: how light is distributed across the detector, how smeared the electron cloud is. Gamma background leaves a more ragged track than a neat double decay.
The new model, A-CNN, was trained on thousands of simulated and real XENONnT events. As a result, it rejects 60% of background gamma flashes while keeping 90% of genuine signals. For the future XLZD detector with a hundred tons of xenon, that means a sensitivity boost of about 40% for the elusive decay.
So the neural network becomes an ally in the hunt for an event no instrument has seen yet. It doesn't make the detector bigger: it teaches it to see more clearly.
Liquid xenon at −108 °C is transparent to its own light, so particle flashes can be picked up by photodetectors across the entire volume.
The search for neutrinoless double beta decay echoes the plot of Solaris: the planet's ocean responds to humans with rare signals that are hard to tell from noise, and scientists try to extract meaning from them.