Atomic clocks compared with astounding accuracy | Hacker News Reader

Atomic clocks compared with astounding accuracy

nature.com

70 pointszeristor41 comments

I got to visit JILA one time and talk to researchers working on a blue or ultraviolet atomic clock (possibly the strontium mentioned in the article - can't quite remember) and they had a great story about how sensitive their clocks were:

In a part of the testing and commissioning phase of this new clock, they were comparing its frequency with the nearby NIST clock and noticed that some days it ran slower than other days. Even after accounting for the usual thing you might expect (position of the Moon, the tides [which have an effect on these clocks even all the way up in Colorado], and even variations in atmospheric density) they still saw this unexplained variation now and then. Eventually they realized it was correlated with home games at the football stadium: the mass of the fans in the bleachers was gravitionally redshifting their clock!

I'm not 100% sure if that's a fully true story or whether it was embellished to impress prospective grad students, but I thought it was amazing. They also said the clocks went crazy any time there was a big earthquake in North America. The sensitivity of these devices is just incredible

If that was true, shouldn't they also see significant differences during night/day, and holidays?

Those cause dispersed changes on people's distribution on all directions around the clock, so no, they are probably not detectable.

Also the sibling comparing with the Earth's mass isn't in a good direction, because the Earth's mass is mostly stable, and the clock only suffers disturbances from changes... But if the story is true, it should have an easy time detecting vulcanism.

Tides cause clocks to change time... even the tail end of electromechanical clocks were precise enough to detect it.

It was detected in 1929 with the Shortt–Synchronome clock See section 3 of this: https://hgss.copernicus.org/articles/11/215/2020/

I would have though mass would be centred at commercial areas during day; and I would also guess that peoples cars would be more mass-significant than themselves, so higher traffic during day would also be significant.

During holidays, people would be more likely to go on vacation, and/or the streets less busy.

Oh, if you place the clock somewhere where commercial areas are mostly on one side of it and residential areas are mostly on the other side, people movement will influence it.

But most cities are not distributed that way, people cancel each other.

It's possible they had already measured and accounted for those variations.

They modelled the distribution of cars/people during holidays (if that's even possible), but then omitted one particular use case?

If we manage to increase the sensitivity a few more orders of magnitude, and put an array of these devices in a sphere, we can make gravitic sensors!

Gravimeters are already a thing and they're sensitive enough to detect accumulating snowfall on the roof of the building they're in.

Gravitational force can be measured very accurately by dropping a retroreflector in a vacuum chamber and measuring the acceleration with a laser interferometer. The clock comparisons explained in the article measure the difference in gravitational potential between two locations. This would normally require measuring and integrating the force all the way between the two locations.

There are better sensors that are already deployed, they were developed in the 1970s so that submarines could navigate undersea with zero emissions by using a gravity gradient map of the seafloor, part of the Trident II system.

https://en.wikipedia.org/wiki/Gravity_gradiometry#Lockheed_M...

I was curious and did some rough back of the envelope math.

The stadium seats about 50,000. Lets assume it's half full of 100 kg people (or entirely full of 50kg people), or a total of about 2,500,000 kg. Additionally the stadium is about 300 meters from JILA. Meanwhile Earth is 6×10^24 kg at 6000 km. That's about 2x10^18 times as massive at only 20,000 times as far. Accounting for squaring the distance, that's about 5x10^9 times as much effect from Earth than from the crowd. It's plausible their instruments could pick up the effect of a difference of that much mass, though I suspect they embellished a bit the degree to which the clock went slower on game days.

According to the article these clocks are accurate to 2 parts in 10^18, so...

Most people will drive to the stadium, so the local mass increase is a factor of ~10 higher. Assume 2 people per car and 1500kg/car (this is pretty conservative for Colorado, where SUVs and pickups are common), so each athlete, employee, and attendee would bring in ~800kg mass plus their own weight.

EDIT: I was curious about the parking situation; seems to be a bunch of lots scattered around with JILA effectively in the center: https://en.parkopedia.com/parking/stadium/folsom-field-co/?a...

> Accounting for squaring the distance

You actually don't want to square the distance, since you are not trying to compute the acceleration due to gravity, you are trying to compute the gravitational potential (since that's what affects clock rates), which goes like 1/r, not 1/r^2. So the Earth's effect should be about 10^14 larger than the effect of the people in the stadium. (Which is still several orders of magnitude larger than the sensitivity of the clocks, so it's entirely plausible that the clocks were detectably affected.)

I once worked with a guy who had worked on gravitational measurements in (I presume) grad school. They measured the attraction between spheres of lead. One of the adjustments they had to make was considering the change in the masses as lead atoms evaporated from the surface of the spheres.

The gravitational attraction between spheres of lead is strong enough to be measured in a high-school science lab (if you can remove the students and air currents); this is the famous Cavendish experiment of 1798.

I remember it blowing my mind to realize that although a lead sphere is tiny compared to the Earth, it's also right there: whereas the Earth acts like a point mass four thousand miles beneath your feet.

The stadium seems to be about 160 m from the closest part of JILA. Assuming that 100000 people arrive in a 2000 kg car each, the gravitational redshift would be around 1E-21. This is about 1000 smaller than the stability of the best JILA optical clocks.

I built a microscope and most of the time I have it looked at a test object. It's in the garage; when my kids in the house walk around about 20 feet away, on a loosely coupled floor, the image jiggles noticeably.

Michelson experiment was done deep in a building at night, with instrumentation floated on a pool of mercury to remove vibrations, but they still had problems due to horse-driven cargo a quarter mile away. Today, undergrad physics students do the experiment in a day on a tabletop (tools got much better).

The Ligo interferomoters are very sensitive to vibrations. I remember hearing on the tour they said they could detect waves lapping on the coast a few hundred miles away.

Yes, it's a holy temple of science. It's pretty much on a straight path from the michelson experiment.

My buddy in college was working on reducing thermal noise in the suspension of the mirrors by growing a single crystal suspension string as the thermal vibrations in a metallic wire were enough to throw the experiment completely out of wack (they went on to use different approaches but the story was alway fun as he had to fly with a giant briefcase to transport them to LIGO)

huh. many days I feel like I went into the wrong fields (biophysics and then CS). growing and installing single crystal suspension springs sounds awesome. (of course I'm sure 90% is miserable)

According to Einstein’s theory of relativity, Earth’s gravity causes the frequency of an atomic clock to depend on its altitude. Consequently, the height difference between two remote clocks can be determined by measuring their difference in frequency. At the level of measurement uncertainty achieved in the latest work, clock comparisons could resolve centimetre-sized height differences. Therefore, clocks could provide new tools for long-term environmental monitoring of, for example, ice sheets or ocean levels.

If you want to read more about atomic clocks, this is a great post from 2005: http://leapsecond.com/great2005/tour/

Link to paper: https://arxiv.org/pdf/2005.14694.pdf

The potential land surveying application of high-accuracy clocks is quite interesting. It would provide a measure of elevation that is (in the vertical scale, at least) completely independent of GNSS, at an accuracy comparable to what can be achieved by current field-deployed GNSS receivers, at ~0.02m

I wonder how far that could be stretched. Current permanent ground-station GNSS receivers have vertical accuracies in the mm scale. They achieve that by solving for measurements taken over very long periods (~6 months). Would atomic clock height measurement tools be able to achieve similar accuracies, if the datasets for a single station were made over a similar length of time?

It sounds like it would more be used for relative elevation differences. Is GNSS the gold-standard for that? I'd have thought you could do better with a laser and a level.

I'm not sure that we have levels that are accurate enough for that.

Really? I can't tell what accuracy a good theodolite gets, but this consumer device [1] claims 3/16ths of an inch at 30ft. Obviously that won't scale to 1km distance like they're doing in this, a professional one seems competitive with the quoted ~2cm accuracy.

[1] https://www.amazon.com/SKIL-Self-Leveling-Cross-Line-Laser/d...

Absolute gold standard for use in a laboratory, probably not. But GNSS is definitely the instrument of choice for 99% of surveying and navigational purposes.

GNSS and clock comparisons measure slightly different things and are complementary. GNSS gives the position of the receiver in space relative to some defined coordinate system. This allows accurate mapping of the earth's surface and measurements of continental drift. Clock comparisons measure (via the gravitational red shift) the difference in gravitational potential between the locations of the clocks. This is affected not only by the height difference, but also by changes in the mass distribution within the earth. Transportable optical clocks would allow real-time mapping of the earth's gravitational equipotential surface (geoid). Currently, the geoid has to be determined by painstakingly levelling accross the continents point by point. This can be surprisingly accurate (cm-level) but is obviously much too slow to observe fast variations, for example due to earthquakes.

GNSS and clock comparisons measure different things, one is a based on radio wave travel time between a receiver and the satellites in the constellation, the other is a physical measurement. That's what I was referring to wrt independence: a gravity measurement would avoid any potential error source in the whole chain of the GNSS system (which is quite lengthy).

The geoid has been partially satellite-derived for some time, see the GRAIL, GRACE, and GRACE-FO missions. The US is also running a nationwide airborne gravity program at the moment, too, which an impressive undertaking and should be done soon-ish.

Comparing a measurement of actual local gravity against the geoid-stated gravity should be enough to give the elevation relative to the geoid, which is the same output that you get from GNSS. That's my intuition, anyway.

It sounds like they need to put an array of full tensor 3d gravity gradiometers in Boulder to monitor the real time movement of mass above, below, and on the ground.

It wouldn't be that expensive, relatively speaking. ;-)

I ran several stratum 1 systems and was a member of the public NTP pool for quite awhile.

I have since moved on but maintain my own stratum 1 clock in my rack and a high resolution clock in my office.

Reply on news.ycombinator.com