Hacker News new | past | comments | ask | show | jobs | submit login
Zeptoseconds: New world record in short time measurement (goethe-university-frankfurt.de)
210 points by Tomte 12 days ago | hide | past | favorite | 92 comments

From the article: "They measured how long it takes for a photon to cross a hydrogen molecule: about 247 zeptoseconds for the average bond length of the molecule. This is the shortest timespan that has been successfully measured to date."


Where the mind truly starts to boggle is remembering that a zeptosecond still falls incredibly short of the Planck time/distance at 5×10−44.

It's certainly exciting to speculate how much closer to an actual Planck /measurement/ we may yet achieve.

For an idea of scale, 247 zepto seconds is as close to the Planck Time (on a log scale) as it is to about two weeks (13.1 days)[1]

[1]: https://www.wolframalpha.com/input/?i=5.39e-44%2F247e-21%3D2...

Alternative wording: the ratios of these pairs (13 days / 247zs) and (247zs / t_P) are approximately the same.

t_P being the Planck time here.

Another alternate wording: 247 zs is the geometric mean of t_P and 13.1 days.

Planck time, 247 zeptoseconds, and 13.1days are equally far apart on log scale.

Planck time is really small.

Thanks, I can understand this one :P

So we're still pretty far ! As I imagine an order of magnitude smaller requires an order of magnitude more money and complex equipments.

Thank you

And for anyone that wanted how large a zeptosecond is; it's 10^-21.

That is so amazing.

If we had clocks with this level of accuracy, you could measure the weight of someone walking in the room because of their time dilation.

I would imagine in practice even if the measurement were accurate, there would be so much noise from other large moving objects that this would still not be possible.

Woah. That sounds interesting; can you elaborate?

Not OP, and I don't have time (hah) to do the calculation but a rough rule of thumb is that if you take a "normal size" of a quantity (mass, speed, etc) then relativistic effects become measurable when you scale it by c^2, ie about 10^17. So e.g. We have had clocks for a while that are accurate to a part in 10^18, so they can measure time dilation at walking speed. Or detect gravitational time dilation when you lift something (the clock!) up one meter in the earths gravitational field.

Yet another way of saying how amazing this new result is: A second is much "closer" to the age of the universe than it is to a zeptosecond.

It is worth stating this explicitly - NIST has in fact done the practical experiments to demonstrate time dilation at both 10ms⁻¹ relative velocity and 1m relative height in Earth's gravitational well:


I'm going to guess that the comment about measuring the weight of a person was not referring to measuring his walking speed nor his height. It's more likely referring to his personal gravity well and the time dilation it causes as he approaches and you go further into it. Still a guess, though.

time dilation formula looks like 1/sqrt(1-v^2/c^2). For v = 1.3 m/s, c=3e8, it means if I travel across my room for 1s at 1.3 m/s, to an inertial observer the amount of time that has passed was actually 1.0000000000000000093889s.

.0000000000000000093889s = 9389 zeptoseconds, and this clock can measure time in mere hundreds of zeptoseconds.

For clarity, both people would need clocks though, correct? The top comment sort of implies you could measure this effect simply with one clock in the room, somehow, but you'd need two.

Even if you had 2 clocks that accurate, would it even be possible to properly synchronize them? If you used an electric signal to tell the clocks to measure, wouldn't that wire length need to be measured to the precision of a hydrogen atom?

In electronics, there's a trick to measure the delay between two signals when the length of your cable is unknown. First, Connect Cable 1 to Input 1, Cable 2 to Input 2, record the delay between Signal 1 and Signal 2. Next, swap Cable 1 and Cable 2, record the delay again. Then, add two delays together, delays in both cables are cancelled, finally divide it by two to get the true signal delay. You can calculate the delay of both cables if you want, and calibrate it out for your later measurements.

It works because it assumes swapping the cable has no effect on the delay. But the same cannot be said for your hypothetical atomic clock - creating an electronic switch with matched delay is an obvious challenge, but then, even minor physical movement of the cable will change its delay.

That is correct, you need two clocks.

If you think about it, this is true for any time measurement. If you only had one clock, how would you know if it was properly calibrated?

Yeah, this.

Google Einstein theory of relativity. One of man's greatest achievements.

Certainly one of that man's greatest achievements.

What does this even mean?

Sounds like some cool tech gadget from 2100 :)

To summarize the experiment - they measured the interference effects of electron waves from a hydrogen molecule hit by xrays (high energy photons) and the electron wave interference pattern carried with it information on the relative timing of photon absorption at two ends of the molecule. So using this interference-based "amplification" they could resolve and infer the propagation of the photon across the molecule length scale. Or at least this is how I interpreted it from a coarse read :)

There was no clocks and detectors involved that you might think of at first.

Thanks for your explanation. This makes a lot of sense.

A carrier signal is generated (as high as 10^27) [1] - modulated by some observable - and then recorded and analyzed. The highest resolution event that will be visible in the waveform will be 1/(2f) where f is the frequency of the carrier signal. The 1/2 factor is due to the sampling theorem [2].

[1] https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampli... [2] https://www.quora.com/What-was-the-highest-frequency-that-ma...

> (10-21 seconds)

You’d imagine that a CMS for a university website would support formatting of something as simple as an exponent. Instead what you get is a claim that a zeptosecond is somewhere between 10 and 21 seconds.

you can even just do it with unicode

10⁻²¹ seconds

(tho tbh it looks a bit janky -- but better than "10-21"!)

This is somebodies scope creep

“Of course it should communicate detailed chemical bonds and protein folding!”

I instantly knew it was incorrect but it took me way to long to realize it's suppose to be 1E-21 or 10^-21 seconds.

Is the 1e-21 notation ever used in handwriting? I have never seen it but I'm curious. It was not used in my schooling.

I've not seen it used in handwriting I think it's mostly used for digital inputs (including some calculators). The most common usage in handwriting I've seen is like 1.5×10³² (=1.5e32).

Exactly, what surprised me is that in school this syntax was never taught or explained. Oversight on their part.

Highschool: "They'll teach this in collge"

College: "They definitely learned this in highschool"

Way too many zeptoseconds ;-)

I had a job in college that was basically keeping articles in a CMS for a couple of departments up to date.

The only formatting I got was whatever I could do with HTML and CSS back in 2008. There was a sort of rich text editor you could also use, but it was really hard to make it do what you wanted.

The <sup> HTML tag has been available just about forever. Firefox version 1 supported it and I can't find a version of IE that didn't.

At the same time, a lot of CMSes specifically block HTML input for security purposes - the article author may have had access to stripped down markdown or possibly a docx => html converter built into the CMS.

Seems to originate in HTML 3.2 (January 1997)

Can also use the javascript .sup() function for post-html bloggers.


<ol> <rt>. <u>?

Not much, <sup> with you?

10-21 seconds would be the shortest time I spend on social media in one session.

So the actual measurement here is measuring the orientation of the molecule and the interference pattern of the two electrons ejected from the molecule by their experimental setup exciting them to deduce the amount of time it took for light to go from the first electron to the second, is that right?

“We observed for the first time that the electron shell in a molecule does not react to light everywhere at the same time."

Question: does this not violate the "quantum" in quantum physics? I thought that a quantum particle had to be in either one state or another simultaneously in all places in the universe, because otherwise, it would have a gradient between one and the other, and its energy would no longer be quantised.

The quantization of energy is mostly a 1920's observation that created the name Quantum Mechanics, due to a lot of interference effects among wave functions in various spherical or cylindrical symmetric arrangements that result in a lot of macroscopical effects seemingly being quantized, especially compared to the classical physics that preceded QM. But most are not fundamental effects, they are emergent.

When you go into smaller length and timescales, these quantization effects kind of lose or change their meaning.

So an electron doesn't instantaneously change between "shells" in an atom upon reception of a photon if you look closely enough, there are indeed smooth intermediary states and maybe the researchers are saying they detected subtleties in these states (I didn't read the original research article yet).

We can go quite higher. We are limited by inverse of the highest frequency carrier signal we can generate. Right now, that is 10^27ish for photons at the LHC. [1]

[1] https://www.quora.com/What-was-the-highest-frequency-that-ma...

Wow! This is the first time I notice our local Universities Physics Department in the News.

Congratulations to all involved!

Allows for extreme precision tests of spacetime (thanks Einstein!)

Amazing work

For scale, how many zeptosecond to execute a register access?

About a trillion zeptoseconds == 1-nanosecond.

Or 250 Billion zeptoseconds == 0.25 nanoseconds (aka: typical 4GHz desktop CPU these days). Many CPU operations can be applied to a register in a singular clock tick (Add, Sub, XOR, AND), multiplication can happen once-per-clock tick but still has a latency of a few cycles (maybe 4 to 5 clocks, or ~1-nanosecond)

1 second is to 1 picosecond ~ 1 nanosecond is to one zeptosecond

For comparison, the mean lifetime of the W and Z bosons is 0.3 ys - yoctoseconds, which is the next multiple down. (10E-24)

>A zeptosecond is a trillionth of a billionth of a second (10-21 seconds).

Wow. Seriously impressive work.

In what situations would one use this measurement?

I am curious about applications too. I wonder if this is somehow related to being able to do protein folding

Not sure, but theoretically it's precise enough to measure the time dilation caused by the gravitational field of a baseball. Pretty mind-blowing.

I'd guess quantum research

How many Planck times is that?

I mentioned in another comment, but Planck time is to this as this is to 2 weeks.

A Planck time is about 5 x 10^-44 seconds, so... a lot.

"They measured how long it takes for a photon to cross a hydrogen molecule: about 247 zeptoseconds for the average bond length of the molecule."

24-7? Say no more.

See: Apophenia (clustering illusion)

The replication crisis does not mean all science is false and the title of that paper is itself false.

No, I agree that not all science is not false.

But if you were a betting man, you would bet on this being false, seeing as 70% of studies cannot be reproduced.

And this is a study that no one else can even hope to reproduce. Its just a claim.

And there is the comedy 'gag' of 24-7. Ie, the time they measured was all the the time. Hilarious.

That is what i mean when i reference the clustering illusion. you have seen this pattern of numbers before in a different context and now believe this paper must be a forgery, because a number in it matches an unrelated pattern.

Naming something doesn't mean you have transcended it. People go around saying "I have solved the problem of recognizing patterns incorrectly...through recognizing this pattern!"

Whereas you believe it to be true.

Even though the odds are on my side. According to science!

you are conflating different things

How did they start & stop the measurement? (Since I suppose the photon needs to travel to/from the clock to turn it on/off, and this takes time)

I don't believe there was a clock at all which makes sense. We don't have clocks accurate enough to measure things at this time scale. From the link:

> “Since we knew the spatial orientation of the hydrogen molecule, we used the interference of the two electron waves to precisely calculate when the photon reached the first and when it reached the second hydrogen atom,"

They should switch to logarithmic scale and just say "an order-21 time interval". These "zepto" prefixes are gibberish.

Not gibberish at all, in my opinion. They were carefully and creatively chosen, and derived in many cases from logical linguistic sources. I think it's just that we are used to milli- and micro- and nano-, but actual usage of "zepto" has been basically nil until now. But that will change.

On the other side, we're all very used to mega- and giga- and tera-, and for data center folks, peta- as well, and we'll adapt to the higher-order prefixes as they enter more common use.

Monday is exascale day. https://www.exascaleproject.org/

Correction: it's Sunday (today)

Sure, but those are much smaller exponents than 21. I can see "peta" coming into use as storage sizes change, just as "kilo" has already disappeared from your list, but asking folks to keep track of 14 different prefixes is a little much. If it's outside of +/-9, I'd support the use of an exponent.

Remember that it's not 14 prefixes, or at least not 14 new ones. We're just talking about a couple new ones. There's only a new prefix for every three orders of magnitude, and everyone is already very familiar with the small ones. And there are mnemonics.

And we're not asking the layman to know all these, either. Zepto- will never be in common use; that's for particle physicists and basically nobody else, so far at least.

Everyone already knows 10^12=tera, so we're covered there.

-21, -18, -15, -12, -9, -6, -3, 3, 6, 9, 12, 15, 18, 21 is 14 prefixes.

I think the number of people who could (correctly!) name even half of those is much smaller than you think.

You are missing - 24, -2, -1, 1, 2, and 24 for a total of 20.

±1 doesn't require a prefix, and while in Europe "deci" is used in common parlance for units of power ±2 it's not really used in scientific or engineering terminology globally. I'm also not sure what real quantity you're referring to which requires the power ±24 prefix (yes, the term "yottabyte" exists but no measurement of data capacity uses it).

±2 are centi and hecto with centimetre and hectopascal probably being the widest used ones, ±1 are deci and deca with decibel probably being the widest used one. For deca I can also not think of any real usage.

Yeah, but the number of people who actually need to use those units do know how to name them.

I like this for writing but its much easier to say a prefix.

We use Zepta every time we discuss ZFS.

I think you meant to write "Zetta" there.

No they shouldn't, it's not rest of the world's fault that you use the imperial system and don't learn decimal prefixes.

How many Planks in a Zepto?

A planck time is 10E-44, a Zeptosecond is 10E-21 so 10E23 - the zeptosecond is further from a planck time than from a second (logarithmically).

i started studying astrology as a covid project and got distracted. one of the texts called surya siddhantha speaks of units of time. the smallest unit of time is 'truti' which is 10^-7.

the text divides time as 'murta' and 'amurta'...the first translates to 'embodied' and the latter translates to 'unembodied'. it means 'unreal time'. in unreal time, there is no destruction. all 'murta' time units are based upon 'one respiration' or 'one prana'. (living time or 'murta' unit)

one truta is 1/37500 of a second. i dont know if it goes smaller than that..but it builds up from that to a kalpa of brahma which is 4.32 billion human solar years. two kalpas make a day for brahma. 30 days make a month of brahma and he lives for 100 years. currently, we are in the 50th year of brahma, first month and in kali yuga(4 yugas of which kali is the last make one chatur yuga..1000 chatur yugas make one kalpa.)...kali yuga started ...acc to these texts started in 3102 BCE and will end after a run 432,000 years. current kali yuga began roughly 5,121 years ago and has 426,879 years remaining as of 2020 CE.

i wasnt paying attention to the quantum scale and was fascinated by the cosmological time. at some point, i expect to return to the lunisolar calender time that was really the original purpose of my reading, but i got carried away and went down the rabbit hole.

its going to be difficult to get back to astrology study because turns out there are multiverses and multiple brahmas who 'bubble' in the ocean and disappear as they create and destroy their worlds and themselves. and then there is time dilations for various creatures each one perceiving time on a different scale. this with the accompanying mythology is better than any sci fi i have ever read.


Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact