“We can rewind to a previous scene or skip several scenes ahead,” explained Miguel Navascués of the Austrian Academy of Sciences, of an achievement that heretofore was the province of science fiction. His team, in collaboration with researchers at the University of Vienna, has experimentally shown protocols that allow a quantum system to play out forwards or backwards in time without having knowledge of its history.
Central to the accomplishment is a device called a quantum switch. Processes in classical physics follow a set order, such as the frames of a movie being played out in succession. The quantum switch uses superposition to put the order of operations itself into a quantum state, allowing many different causal sequences to exist together. In their “rewind protocol,” the researchers passed a single photon through a process that changed its quantum state, then used the switch to restore it to precisely the state it was in beforehand reversing with more than 95 percent accuracy. The key was to do so “blind,” without measuring the intermediate state, which would have irreversibly altered it.
This capability to undo a quantum state without earlier knowledge is solving an essential challenge in quantum computing: error correction. Qubits, the quantum equivalents of bits, are notoriously sensitive to noise and decoherence. Conventional quantum error correction codes, like surface codes or repetition codes, flag and correct faults by redundantly encoding information across numerous qubits, usually at the expense of high hardware overhead. A rewind protocol might, in principle, serve as an immediate “undo” for specific unwanted evolutions, supplementing such techniques and obviating the necessity for enormous redundancy.
The researchers also showed a “fast-forward” protocol that speeded up a system’s evolution by siphoning time from identical systems. In one experiment, they caused a target system to “age” ten years in a single year, by draining one year of evolution from nine others. The idea is analogous to variational fast-forwarding (VFF) algorithms for quantum simulation, wherein a unitary evolution operator ( e^{-iH\Delta t} ) is approximated into a diagonal form that can be exponentiated to large times without deepening the circuit. In VFF, a shallow-depth circuit ( WDW^\dagger ) substitutes for multiple Trotter steps, enabling simulation times much beyond the hardware’s coherence time while maintaining error growth under control.
The physics behind rewind and fast-forward are connected to quantum superposition and the lack of a definite “arrow of time” at the microscopic scale. Macroscopic processes such as melting ice are irreversible because entropy increases, but quantum mechanics’s fundamental equations are time-reversal symmetric. By designing superpositions of forward and backward evolutions, as researched in superposition-of-time-direction experiments, it is possible to interfere such trajectories so that they are biased to lead to reversal or acceleration.
In practice, the application of such protocols demands precise control over quantum operations and states. The Austrian group’s single-photon arrangement circumvented intermediate measurement, maintaining coherence throughout the process. This reflects approaches in next-generation quantum error correction hardware, where the continuous parity measurements and feedback loops need to be engineered to avoid measurement-induced dephasing and crosstalk. The non-invasive nature of the rewind protocol is especially appealing for near-term quantum processors, where each extra measurement threatens to collapse the computation.
Possible uses go far beyond error correction. In quantum simulation, fast-forwarding might condense the equivalent of thousands of gate operations into a constant-depth circuit, allowing investigations of long-time dynamics in materials or chemical systems that are out of reach because of decoherence. For algorithms such as quantum phase estimation, where precision scales with the largest simulated time, compression might cut gate numbers by orders of magnitude.
There are limitations. As Navascués warned, replaying even one second of a person’s quantum state would take astronomic amounts of resources and have a “very, very low” success probability. The information content in macroscopic systems is enormous, and these protocols function within the well-controlled, low-entropy domain of only a few qubits or photons. Nevertheless, in that regime, they provide a basically novel control knob: the capacity to reshape the temporal profile of a quantum evolution without explicit information about its path.
Philip Walther, who led the experimental effort, called it “one of the most difficult experiments we’ve ever built for a single photon.” Its success suggests that time, in the quantum world, is not just a parameter to be endured but a resource to be engineered rewound, fast-forwarded, and perhaps one day, precisely edited for the needs of quantum technology.
