Photonic Quantum Computer Sets a Speed Record ⚡ экспресс
To overcome photon losses, the main obstacle in optical demonstrations of quantum computational advantage, new Gaussian boson sampling experiments are conducted using 1,024 highly efficient squeezed states injected into a hybrid spatiotemporal encoder of 8,176 modes on the programmable photonic quantum processor Jiuzhang 4.0. Up to 3,050 photon detection events were recorded. This result surpasses all classical simulation algorithms, including the matrix product states (MPS) method recently proposed to exploit losses for reducing simulation complexity. On the most powerful supercomputer El Capitan, a state-of-the-art MPS algorithm would require over 10^42 years to build the necessary tensor network, whereas Jiuzhang 4.0 produces a sample in 25.6 microseconds. This work sets a new milestone in quantum supremacy and charts a course toward fault-tolerant photonic quantum computers.
A classic Galton board with pegs and balls generates a pattern from randomness. In the quantum version—Jiuzhang 4.0—balls are replaced by photons. They travel through a maze of 8,176 waveguide paths, and photometry (precise flash counting) records the final pattern. The chip itself is cooled below interstellar temperatures to suppress thermal noise.
According to the Standard Model, each particle of light takes all paths at once—like a ball instantly smeared across the entire board. A regular supercomputer would have to check more possibilities than there are atoms in the universe, taking 10^42 years. Jiuzhang delivers the answer in 25.6 microseconds. Engineers used "squeezed" light to eliminate noise, and randomness turned into computational power—just as entropy, which usually brings disorder, here generates order. The term “quantum supremacy” was coined by [scientist:John Preskill]. This step brings closer photonic computers capable of designing drug molecules.
🎯 If the El Capitan supercomputer had started crunching numbers at the Big Bang, by now it wouldn't have completed even a fraction of what Jiuzhang 4.0 does in 25 microseconds.
🎬 Instant computation, like in "Star Trek", but in a real lab chip.