Physicists Reversed a Quantum Computer's State, Not Time Itself

Facebook ↗WhatsApp ↗
Quantum Researchers Reversed a System’s Evolution, Not the Passage of Time

@tsarcyanide/MIPT

In March 2019 a team of physicists reported that they had made a small quantum computer "go back in time". What they actually did was more modest and, for physicists, more interesting: they ran a programme on IBM's public cloud quantum processor that let a few qubits drift from a simple, ordered state into a more complex one, then used a carefully designed operation to bring the qubits back to where they started. Time itself did not run backwards, nothing travelled into the past, and the clock in the lab kept ticking forward throughout.

The study, "Arrow of time and its reversal on the IBM quantum computer", was published in Scientific Reports by Gordey Lesovik of the Moscow Institute of Physics and Technology (MIPT) and colleagues, including Andrey Lebedev of MIPT and ETH Zurich and Valerii Vinokur of Argonne National Laboratory in the United States. Here is what they did, what the numbers mean, and why the headlines went too far.

Why physicists talk about an arrow of time

Most of the fundamental equations of physics work equally well forwards and backwards. Film two billiard balls bouncing off each other, play it in reverse, and nothing looks wrong: the reversed motion obeys the same laws. Yet everyday life clearly has a direction. A cue ball scatters a neat triangle of balls, but scattered balls never roll back into a triangle by themselves. Milk mixes into tea and never unmixes. Eggs break; they do not reassemble.

The usual explanation is the second law of thermodynamics. In an isolated system, entropy, roughly the amount of disorder, tends to increase. Ordered arrangements are rare compared with disordered ones, so a system left to itself almost always moves towards disorder. Going the other way is not strictly forbidden by the underlying equations, just overwhelmingly unlikely. This one-way tendency is what physicists call the arrow of time.

The open question is how this arrow appears in the quantum world, where single particles follow the Schrödinger equation, which also makes no distinction between past and future.

The electron thought experiment

The team began with a thought experiment. Picture a single electron, isolated in empty space, whose position is known quite precisely at first. Over time its wave packet spreads out: the region where it might be found grows, and our knowledge of its position becomes fuzzier. This spreading plays the role of increasing disorder.

Could that electron spontaneously "unspread" and become localised again, as if its recent history had been rewound? In principle a random fluctuation in its surroundings, such as the faint microwave background that fills space, could nudge it in exactly the right way. The researchers calculated how likely that would be. Their answer, as summarised by MIPT, was that even if you observed ten billion freshly localised electrons every second for the entire age of the universe, you would expect to see such a reversal only once, and the electron would return to its state from no more than about a ten-billionth of a second earlier.

In other words, nature essentially never does this on its own. That is the point the experiment was designed to contrast with.

What the quantum computer experiment did

A quantum computer can be programmed to do what nature will not do by chance. The team used IBM's five-qubit "ibmqx4" processor, which researchers could access over the cloud, and worked first with two superconducting qubits, then with three. The experiment had four steps, illustrated below with the same billiard-ball and electron analogies.

Quantum Researchers Reversed a System’s Evolution, Not the Passage of Time

Four stages of the experiment, compared with a billiard break and a spreading electron. Credit: @tsarcyanide/MIPT

  1. Order. All the qubits start in the simplest state, zero. This is like a neat rack of billiard balls, or an electron confined to a small region.
  2. Degradation. A short evolution programme runs, and the qubits' state turns into a complicated superposition of zeros and ones. This is the break shot, or the electron's wave packet spreading out.
  3. Reversal. A second, specially designed operation transforms the state so that it is set up to evolve "backwards". In the billiard analogy this is an impossibly precise kick to the table; for the electron, it is the lucky fluctuation that nature almost never supplies. Here it was deliberately engineered.
  4. Regeneration. The same evolution programme from step two runs again. If the reversal worked, the complexity unwinds and the qubits return to their starting state of zeros, like the balls rolling back into a triangle.

The paper reports that the two-qubit version returned to the starting state about 85 percent of the time (85.3 percent). With three qubits, success dropped to about 49 percent. The authors attributed the failures to imperfections in the hardware, such as noise and gate errors, which pile up as more qubits are involved.

Why this was not time travel

Headlines such as "physicists reverse time" made good clickbait, but they blur two very different ideas. One is reversing a system's evolution: steering it back to an earlier state. The other is reversing time itself: making events run backwards or sending something into the past. The experiment did the first, not the second.

The key detail is the external intervention. Step three is a precisely calculated operation applied from outside, using full knowledge of how the qubits had evolved. Without it, the system would simply have kept becoming more disordered. That fits the second law rather than breaking it, because the qubits were not an isolated system left to themselves. Rewinding a video does not mean the events in it happened backwards; similarly, rewinding a quantum state does not mean time ran in reverse.

The success percentages also need care. They measure how often a particular small circuit on a particular noisy device returned to its target state. They are not the probability of anything travelling in time, and they say nothing about large objects. Lebedev himself compared the electron's spreading with ordinary disorder in large systems, but scaling that kind of reversal up to anything macroscopic would require controlling an astronomical number of particles at once.

Why it still matters

Stripped of the hype, the study is a neat demonstration that the arrow of time, at least in a tiny quantum system, can be reversed when you have complete control over it, and that the chance of this happening spontaneously is vanishingly small. It links an abstract question about thermodynamics to something that can be tested on real hardware.

There are practical angles too. The authors suggested that reversal procedures like this could help test and improve quantum computers, for example by checking how faithfully a device can undo its own operations. Running a circuit forward and then backward is a sensitive way to reveal errors, because any imperfection stops the system from returning cleanly to its starting point.

Since 2019, quantum processors have grown much larger and more accurate, and related ideas, such as "echo" experiments that run evolutions forward and backward to study how information spreads, have become standard tools in the field. None of them involve sending anything into the past, but they do help physicists understand why the world we live in only ever seems to run one way.

Quick answers

Did scientists reverse time? No. They reversed the evolution of the state of two or three qubits, returning them to an earlier configuration, while time carried on as normal.

Which quantum computer did they use? IBM's publicly accessible five-qubit processor, known as ibmqx4.

Does this break the second law of thermodynamics? No. The reversal needed a carefully designed outside intervention, so the qubits were not an isolated system, and the paper's calculations show that spontaneous reversal in nature is extremely improbable.

Where was it published? In Scientific Reports, volume 9, article 4396 (2019); a preprint is on arXiv.

Sources: Lesovik et al., Scientific Reports 9, 4396 (2019); arXiv preprint 1712.10057; Phys.org / MIPT press release, March 2019. Illustrations: @tsarcyanide/MIPT.

COMMENTS