Connect with us

NEWS

Physicists Find Time Reversal Stops Where Light Remembers

Aharonov and Elitzur find quantum time reversal lasts about a femtosecond, the time light takes to remember a decay.

Published

on

A sealed cloud of decaying atoms can look as if time ran backward, but only for about a femtosecond. That cap sits in 18-page November preprint by Yakir Aharonov and Avshalom C. Elitzur, and it is set by the light the atoms emit, not by time itself.

Aharonov, Institute for Quantum Studies director at Chapman University in Orange, California, and Elitzur, of Iyar, the Israel Institute for Advanced Research in Zichron Ya’akov, Israel, asked how far their old quantum time-translation machine can push an irreversible process. The answer, they write, is as far as the environment’s memory, “and beyond it only at a cost that grows exponentially with the depth of the reversal.”

They Came Back to a 1990 Time Machine

On 18 June 1990, Aharonov, Jeeva Anandan, Sandu Popescu and Lev Vaidman published a 1990 time-translation paper in Physical Review Letters. A clock whose path is split in a gravitational field, or whose speed is split, spends a different proper time on each branch. After the branches are recombined, a final filter called post-selection can make the clock behave as if it had run faster than either path, frozen, or backward.

The trick is a weak value: a number that can sit outside the range of the two clock rates you actually prepared. Aharonov had introduced weak values two years earlier. In the new work the weak value is the time-dilation factor, written αw, and it can exceed 1, vanish, or go negative. Two-path versions of that superposition have already been run with coupled nuclear spins. Those tests reversed a rotation, not a decay.

Thirty-six years later the pair returned with a sharper question. If the thing inside the machine is not a spin but an irreversible process, how far back can the apparent reversal go? They treat the whole construction as a gedankenexperiment. The numbers, they say, are absurd, and they print them anyway because that is how you see the conditions under which the effect exists at all.

Atoms That Decay, Then Un-Decay

A freshly excited atom is a poor witness. Its excited population is a peak, so a later count that looks “more excited” could still be ordinary forward motion. Atoms that have already started to decay can certify a negative proper time, because no forward evolution puts more of them back into the excited state.

Imagine 100 excited atoms, waited on until 80 remain excited. After the theoretical run, a count above 80 would mean some of the decay had been undone. For the light to return to the atoms, it cannot have escaped. The test system is therefore a small sealed cell that carries the atoms, the radiation field they emit into, and the absorbing walls as one quantum object, split, held, recombined, and filtered whole.

HOW THE WITNESS WORKS

  • The starting point: Use atoms that have already begun to decay, not a fresh excited sample.
  • The package: Keep atoms, emitted light and absorbing walls inside one cell, because a photon left behind would label the path and kill the interference.
  • The split: Put that cell into a superposition of two clock rates, then recombine it.
  • The filter: Keep only the rare outcomes in which the control does not reveal which branch ran.
  • The readout: Count excited atoms; a rise above the inserted value is a projective sign that no forward run can mimic.

The rule the authors state in italics is blunt. A process can be reversed only if everything it emits is carried by what is recombined. Leave a photon outside and the environment has already recorded which branch you were in.

Why Does the Reversal Last About a Femtosecond?

Decay into a realistic light field has two stages. For a short memory time τc, set by the bandwidth of the field, the excited population falls as a quadratic, and the emission is still coherent with the atom. After that, the fall is exponential, the process is Markovian, and the photon has left for good.

With only two branches, post-selection amplifies the tiny difference between the cell’s contents on each path. That difference is set by the full energy spread of the atoms and their field, not by the narrow linewidth of the glow. Once the decay is older than the memory time, the amplification eats the signal. An exact plot in the paper, with the run lasting one-tenth of the pre-decay wait, keeps the witness alive up to about 1.2 memory times, and loses it badly by 100 memory times.

THE TWO-BRANCH WINDOW

  • Memory time: For atoms radiating into free space it is of the order of the optical period, about a femtosecond.
  • What that means: In that interval the emitted field has not separated from the atom by more than a wavelength, and no wall is involved.
  • What is reversed: The coherent onset of the emission, not the emission after the photon has left.
  • Deeper undo: Reversing a nanosecond of optical decay would need of order a million branches, plus a strictly bounded spectrum, with a success chance that falls exponentially in that number.

The authors’ own summary puts the finding in one sentence.

The obstacle is bandwidth, not entropy, and it is the same bandwidth that makes the emission irreversible: the arrow of time enters exactly where the machine stops.

Yakir Aharonov and Avshalom C. Elitzur, arXiv:2611.12971

The paper has not been peer reviewed. Ancillary code for the two-branch calculation rides with the posting, which lists both quantum physics and general relativity as subjects.

Four Clock Rates and One Success in 1048 Runs

Nothing in the decay analysis depends on how you superpose the clock rate. The machine needs a control that gives the branches different proper times and as little else as possible. Section 6 of the paper walks through four controls, from the original gravitational shell to a moving cell, and Table 1 puts numbers on the rate gap δ and on the extra ways a branch can leak which-path information.

FOUR CONTROLS FOR A SUPERPOSED CLOCK

Control What differs between branches Rate gap δ Extra leak
Classical hollow shells A constant clock rate, no field inside 4×10-25 for a ton at a meter None
Heights on Earth Depth in a shared uniform field 5×10-17 per meter None; the field is common
Rotating disk Depth in the centrifugal potential 10-12 at 500 m/s rim speed Uneven strain in the cell
Inertial speeds Speed, with no field in either branch 3×10-12 at 1 km/s The turnaround acceleration

Going down that list, the price of a frozen clock (αw = 0) falls by 25 orders of magnitude, from one successful run in 1048 for the shells, through 3×1032 for a meter of height on Earth, 2×1024 for the disk, and 1023 for a kilometer-per-second twin-paradox interferometer. The shell remains the purest control, because the interiors differ in one number and nothing acts on the cell. The kinematic versions buy probability by accepting a leak that has to be estimated, not wished away.

A gram-scale cell with a proper-time difference of 10-20 seconds already carries a rest-energy phase of order 1028 radians. Those phases are known, so the formalism can absorb them into which output port is dark. Their stability is one more reason the authors call the whole construction a thought experiment.

Eddington’s Rule Still Collects Its Fee

Arthur Eddington wrote in 1928 that a theory found against the second law of thermodynamics should collapse in deepest humiliation. The new scheme does not contradict that law, because the reversal is seen only in a post-selected slice of runs.

If your theory is found to be against the Second Law of Thermodynamics I can give you no hope; there is nothing for it but to collapse in deepest humiliation.

Arthur Eddington, The Nature of the Physical World, 1928

The control is measured first, the atoms only afterward, so the reversal is certified in hindsight. Averaged over every outcome of the filter, the atoms show exactly the expected decay and the walls take exactly the expected energy. “Post-selection sorts runs; it changes nothing on average,” the authors write. “Nothing is resurrected.”

Conditioning on the rare successful record lowers the entropy of the atoms and walls at the cost of that record and of the discarded runs, the same bookkeeping that follows Maxwell’s demon. The second law constrains ensembles. It does not constrain a sub-ensemble picked by a final boundary condition. What limits the machine is the spectral width of what must be reversed, through the which-branch information that width carries. A broad environment forgets quickly, and it is past its memory time that the photon leaves for good.

What Would a Superposed Shell Add?

In all four controls above, gravity or velocity stays classical. What is superposed is the cell’s worldline. The 1990 idea went further: a hollow massive shell itself in a superposition of radii, so the gravitational field inside is superposed. If that version showed the same witness, the field would have been shown to enter a superposition and to act coherently on a clock.

Mean-field semiclassical gravity predicts the opposite, a single average field and no superposed clock rates. The superposed shell would therefore also be a test of whether gravity is quantum, with spontaneous emission rather than an interference phase as the witness. Its price is the one the four classical controls were chosen to avoid: a body of tons held in a coherent superposition of configurations.

Aharonov, a Wolf Prize winner in 1998 and a Fellow of the Royal Society elected in 2024, has spent decades on weak values, superoscillations and the Aharonov-Bohm effect. Elitzur, who completed a 1999 doctorate on time’s passage under Aharonov, is known for the 1993 Elitzur-Vaidman bomb tester. Their new calculation does not build a corridor into the past. It locates, to about a femtosecond for ordinary light, the instant a radiating atom’s surroundings start keeping a record, which is the instant the one-way world we live in begins.

Frequently Asked Questions

What Is a Quantum Time-Translation Machine?

It is a 1990 thought device in which a system is prepared in a superposition of different proper times, then filtered by post-selection so that the surviving runs behave as if they evolved for a duration outside that range, including a negative duration. The 1990 write-up also noted a force-amplification example: ten thousand superposed states can, in principle, yield a hundredfold boost in about one run in 10,000, which is the same rarity-for-amplification trade that the 2026 decay analysis turns into an exponential cost.

Does a Femtosecond Time Reversal Break the Second Law?

No, because a Cauchy-Schwarz bound in the paper says the runs in which the atoms are more excited than when inserted are never more numerous than the ordinary survivors; the conditional chance looks large only because the filter itself is rare. Over the full ensemble the expected number of decays, and the expected energy dumped in the walls, stay exactly where ordinary thermodynamics put them.

Why Are Two Branches Not Enough Past the Memory Time?

Past the memory time the post-selected combination has to reproduce a backward evolution across the whole spectrum of the cell, which oscillates about Λτ0 times, so a superoscillation with that many wiggles needs at least that many branches. Any weight outside a strictly bounded band is amplified by the same huge coefficients, which is why an open optical line, with its far wings, kills the two-path witness as soon as the photon has left.

Who Are Yakir Aharonov and Avshalom Elitzur?

Aharonov, who also holds an emeritus title at Tel Aviv University, received the U.S. National Medal of Science in 2010 for weak measurement and the Aharonov-Bohm effect; Elitzur left school at 16 to work as a laboratory technician, later founded Iyar, and built his reputation on interaction-free measurement. Their new paper thanks Eliahu Cohen for pointing out the role of energy variance, which led to the memory-time analysis.

Harry is the editor of BUDGY APP, an independent title he owns and runs after ten years in journalism that began on a reporter's desk and ended up at the editor's. Numbers get particular attention here. A percentage in a business story is recomputed from the underlying figures before it goes live, a benchmark in a technology or gaming review is quoted with the conditions it was measured under, and a transfer fee or a lap time in the sports and auto pages is traced back to the club, the league or the timing sheet that published it. The same rule covers news, science, entertainment, lifestyle and travel: if a figure cannot be tied to a filing, a dataset, a transcript or a test Harry ran himself, it does not appear. Readers around the world see prices in the original currency with a conversion alongside. Errors are corrected in the open under a published corrections policy, with the change noted on the article. Questions about any figure reach him at support@budgyapp.com.

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending