Digital Twin Teaches a Laser to See express
A laser strikes a tiny glass grating, and electrons passing through it are accelerated. You can't see inside, so you judge what's happening by the energy of the electrons after acceleration, like judging the shape of a stone by the splash it makes.
Scientists built a digital twin: a computer model that mimics the behavior of the real accelerator. They ran thousands of laser pulse variations through it and obtained thousands of energy spectra. Then the neural network learned the inverse — to determine what the pulse was from the spectrum.
Now, a single snapshot of electron energy reveals the tilt of the laser pulse front to within a degree, and the phase shift to within 0.36 radians (that's a difference of 0.4 femtoseconds in arrival time). The neural network does this in milliseconds, keeping up with 50 pulses per second. Thus, the accelerator becomes its own sensor, seeing what is hidden from direct measurements.
A laser pulse can be compared to a wave at a stadium: spectators stand up and sit down, passing the wave along. The tilt of the front is like the wave moving not straight but diagonally. The phase is the moment when an electron meets the crest of the wave. Both parameters are critical for acceleration, and now they can be controlled in real time.
Original abstract
as written by the authors arXiv:2603.28549
We present the development of a digital twin-based reconstruction framework for dielectric laser acceleration (DLA) based on machine-learning-assisted inversion of single-shot electron energy spectra. DLA as a promising candidate for compact electron accelerator designs using optical nearfields in dielectric nanostructures lacks on direct diagnostics on the laser-electron interaction. Thus, the outgoing electron energy distribution is one of the few experimentally accessible observables. To exploit this information, DLA interaction and mapping on the downstream spectrometer are treated as nonlinear measurement device whose response is described by the symplectic sixdimensional tracking code DLAtrack6D. This forward simulation model serves as a digital twin mapping laser-electron interaction parameters onto resulting energy spectra. For diagnostics, we are interested in the inverse mapping represented by a neural network trained by synthetic datasets generated with the forward simulation model. The reconstruction performance and parameter identifiability of the inverse model are evaluated for parameter ranges relevant to planned experiments at the ARES linac in the SINBAD facility at DESY. Simulation studies demonstrate that the method can recover pulse front tilt angles with an accuracy of about 1 deg and phase offsets with an RMSE of about 0.36 rad corresponding to a difference of 0.4 fs in arrival time. Training with noisy spectra further improves robustness against spectrometer noise. The trained surrogate model evaluates in the millisecond range, enabling shot-to-shot parameter estimation compatible with the 50 Hz repetition rate of ARES. The approach effectively transforms the DLA interaction region into a virtual in situ diagnostics for otherwise inaccessible laser parameters during experiment and provides a foundation for real-time monitoring and control of future DLA experiments.
Dielectric laser accelerators promise to be thousands of times more compact than traditional ones: instead of kilometer-long tunnels, a chip the size of a palm. But so far, they accelerate electrons only to a fraction of the energy of big machines.
In the series 'The Expanse,' ships are propelled by giant accelerators, but the future may belong to tiny chips that control light as precisely as the neural network from this work.