CERN team claims measurement of neutrino speed >c

[harrylin]

By my calculations, this potential source of error is even smaller than the GR correction. Just for an order of magnitude estimate, the velocity of an object at rest on the Earth's equator, relative to the ECI frame, is about 450 m/s, or about 1.5 x 10^-6 c. That gives an SR correction due to the relativistic gamma factor on the order of 10^-12, which is two orders of magnitude smaller than the GR correction Vanadium 50 gave in post #177.

I did not consider gamma but the first order effect (Sagnac effect) which is not clearly mentioned in the paper. Anyway, I just followed up on it and made a calculation and compared it with the paper: I found that the effect of the speed of the surface of the Earth at that location (at most several 100 m/s) is at least one order of magnitude less than the cause that we are after. Indeed the >7 km/s that they found is really huge, one order of magnitude more than the speed of the equator!

Note that I still think that a GR correction should work the other way: light should be slightly slower at lower potential, right?

 

[PeterDonis]

I did not consider gamma but the first order effect (Sagnac effect) which is not clearly mentioned in the paper. Anyway, I just followed up on it and made a calculation and compared it with the paper: I found that the effect of the speed of the surface of the Earth at that location (at most several 100 m/s) is at least one order of magnitude less than the cause that we are after. Indeed the >7 km/s that they found is really huge, one order of magnitude more than the speed of the equator!

You're right, the Sagnac effect would be larger; I hadn't considered that because DaleSpam posted about it already a while back in this thread.

Note that I still think that a GR correction should work the other way: light should be slightly slower at lower potential, right?

I wasn't thinking about the time but the distance; whether GR corrections to the metric could significantly affect the computation of the distance traveled (and therefore the computation of the light travel time) from the GPS positions of the endpoints. (And as post #177 makes clear, this potential correction is much too small to matter.)

 

[JamesEitz]

With the data available. I have calculated the speed of the Nu = 299792465.4
m/s

 

[jbar18]

(1) Based on current understanding it means that the neutrino has negative mass. It could possibly account for the expansion of the universe if it's found that each neutrino has a very slightly repulsive effect on all matter in the universe.

This doesn't make much sense to me. I'm no expert, but if c is the real universal speed limit, surely anything traveling faster than c would still be measured to any observer to be traveling slower than c (symmetrically around c maybe?). I am very possibly wrong here.

Based on my own intuition, I would imagine that the measurement is probably some systematic error, but if not I would put my money on c being slightly faster than the speed of light. Since that would apparently change GR, which has held up pretty well so far, I would guess that it is almost certainly some systematic error.

 

[jobyts]

This morning's Dario Autiero seminar video link was posted on post 187, but it was apparently not available at that time. Here it is again.

http://cdsweb.cern.ch/record/1384486"

It should answer a lot of questions asked here concerning methodology, etc,. in addition to the paper.

As per the second slide in the presentation, the OPERA collaboration includes 160 physicists from 30 institutions. Does that mean, the 160 scientists are unable to find a fault with the procedure yet? If so,it is highly likely that there is no issue with the experiment.

 

[unusualname]

neutrinos do not have negative mass, there is no retrocausality and Einstein's SR is not proved wrong - I hope :-D

Most of the believable explanations have been put forward, but let us not assume that the experimental team are incompetent - perhaps they have in fact found that neutrinos are closer to the invariant speed required by SR than photons.

So why the wrong delay from light-year distant objects? Well, perhaps neutrinos interact with matter in a much more weaker way than photons (probabilistically weak) so that the delay effect is not noticeable over a few hundred kilometers but is noticeable over light-years.

photons travel VERY close to the invariant speed, but neutrinos travel EVEN CLOSER to the invariant speed (neither travel EXACTLY AT the invariant speed)

 

[harrylin]

As per the second slide in the presentation, the OPERA collaboration includes 160 physicists from 30 institutions. Does that mean, the 160 scientists are unable to find a fault with the procedure yet? If so,it is highly likely that there is no issue with the experiment.

I would agree if they accounted for the biggest possible causes that people here came up with. However if, like me, you cannot find where for example they accounted for the Sagnac effect (which is not obviously negligible), then there is good reason to suspect that they could very well also have overlooked something even bigger.

 

[xts]

160 physicists from 30 institutions. Does that mean, the 160 scientists are unable to find a fault with the procedure yet?

Of those 160, 140 are on the list, but did not work on the article and got even more surprised by it than we are. And remaining 20 - well - they are already biased by 5 years spent on the experiment.

There are even bigger ensembles of people, unable to internally discover their faults. To avoid drastic examples - think about 500 pers engineering team of [car-manufacturer-not-to advertise-it] who were unable to find that the brakes they designed are likely to block, or engineering team of aircraft manufacturer, who learned or air crash they made school-like error designing some detail of the construction.

Such experiment is a big, awfully complicated engineering task, where any of a million of details may cause a fault.

 

[ghwellsjr]

As I read it, the clocks are synchronized using GPS. Just having them identical doesn't account for time dilation if they are at different altitudes (i.e., different levels of gravitational potential), which I believe they are. There has to be some mechanism for correcting their rates to a common standard of simultaneity. That's what the GPS part is for (and it looks like it requires pretty hefty GPS equipment to get that kind of accuracy for the corrections).

Also, I see very precise measurements of distance, but they are all based on GPS location fixes, as far as I can tell. I see a reference to a "common analysis in the ETRF2000 reference frame", but there are no details, just a pointer to a reference at the end of the paper that isn't online. So I can't see if the reference frame they used for their computation of the distance, based on all the measurements, took into account that distance, as well as time, gets distorted when the altitude (i.e., gravitational potential) changes. I would think it would, since they talk about a geodetic survey, which is all about accurate measurements of equipotential surfaces. But it would be nice to have more details.

Isn't it the case that clocks at different altitudes will tick at different rates which means they can never be synchronized, so how can there be a single value for the time difference between these two locations? So how can they precisely define the one-way speed of light between these two locations?

 

[PAllen]

A lot of the measurement builds on the accumulated technology behind the GPS, which the paper authors assume without going into detail. For example, the sagnac effect is fully accounted for within the GPS measurements themselves:

http://relativity.livingreviews.org/Articles/lrr-2003-1/

 

[aleazk]

I was thinking along the same lines as jbar18. The most obvious explanation is that light travels slower than c. Photon mass won't do it (as per my previous post) but some other mechanism might.

So I was wondering if we have any very accurate measurement of c which is independent of measuring the the speed of light. Can we for example measure the deflection of light by the sun accurately enough to determine c to better than one part in 50,000? I don't know but I would be surprised. The best test might be in particle accelerators. The decay times of highly relativistic particles depend on actual c, not the speed of light. Does anyone know what limits that puts on c?

I totally agree, I think the limit velocity is real but I never asked myself at what extent we know that the numerical value of c is that of the speed of light. For this reason, I think the best way to attack the causality proposition experimentally it is by measuring the consecuences, like the particle-antiparticle mass equality, or the existence of black holes.
Maybe in the future we could measure the velocity of gravity waves, which is the actual c. I think we don't have yet a sufficiently precise value of the deflection of light by the Sun to calculate the actual c.

 

[Vanadium 50]

How are they insuring that the neutrinos in Gran Sasso are the same neutrinos from CERN?

You turn the accelerator on, and you see neutrinos in Opera. You turn it off, and they stop. Looks pretty convincing to me. Figure 11 in the paper shows that on a very short time scale.

 

[seerongo]

As per the second slide in the presentation, the OPERA collaboration includes 160 physicists from 30 institutions. Does that mean, the 160 scientists are unable to find a fault with the procedure yet? If so,it is highly likely that there is no issue with the experiment.

Curtain #1: Neutrinos are faster than c.
Curtain #2.: A group of very careful, smart human beings have not yet found a subtle problem in a very complex system (or study design).

While the jury is still out, if I were a betting man, I think I'd bet on curtain #2. Either curtain is sure to have something interesting, though

 

[PAllen]

Isn't it the case that clocks at different altitudes will tick at different rates which means they can never be synchronized, so how can there be a single value for the time difference between these two locations? So how can they precisely define the one-way speed of light between these two locations?

The clock rates are all adjusted via signals (all part of GPS). Effectively, the local clocks are adjusted not to measure local time but synchronized time. They don't measure one way speed of light - they assume it, and the distance is actually a measure time * c (indirectly, using triangulation, all based on radio signals). If neutrinos traveled at exactly c, then (assuming no errors), by construction, they would measure the same 'time' between source and detector as the radio signals imply.

I think there must be a mistake somewhere, but it is clear to me they have not simply forgotten any of the following:

- gravitational time dilation
- sagnac effect
- speed time dilation
- atmospheric effects

 

[Pengwuino]

http://cdsweb.cern.ch/record/1384486

The webcast recording of the seminar involving the results is online.

 

[dgwsoft]

I totally agree, I think the limit velocity is real but I never asked myself at what extent we know that the numerical value of c is that of the speed of light. For this reason, I think the best way to attack the causality proposition experimentally it is by measuring the consecuences, like the particle-antiparticle mass equality, or the existence of black holes.
Maybe in the future we could measure the velocity of gravity waves, which is the actual c. I think we don't have yet a sufficiently precise value of the deflection of light by the Sun to calculate the actual c.

In fact we only know G to one part in 10^4 so using the deflection of light by the sun to accurately measure c is a non-starter. But I am sure there must be measurements at CERN that put a value on c entirely independent of the speed of electromagnetic waves. I would be very interested to know the error bars on that.

And to change tack - I hope this is not overly speculative - I rather like the idea that neutrinos are tachyons (which has been seriously proposed, as other posts here have pointed out). Also, this article may be relevant:

http://en.wikipedia.org/wiki/Lorentz-violating_neutrino_oscillations

any thoughts?

 

[ghwellsjr]

Isn't it the case that clocks at different altitudes will tick at different rates which means they can never be synchronized, so how can there be a single value for the time difference between these two locations? So how can they precisely define the one-way speed of light between these two locations?

The clock rates are all adjusted via signals (all part of GPS). Effectively, the local clocks are adjusted not to measure local time but synchronized time. They don't measure one way speed of light - they assume it, and the distance is actually a measure time * c (indirectly, using triangulation, all based on radio signals). If neutrinos traveled at exactly c, then (assuming no errors), by construction, they would measure the same 'time' between source and detector as the radio signals imply.

I think there must be a mistake somewhere, but it is clear to me they have not simply forgotten any of the following:

- gravitational time dilation
- sagnac effect
- speed time dilation
- atmospheric effects

I know they cannot measure the one way speed of light, that's why I asked how they can define it since the clocks at both ends will be running at different speeds. To boil the problem down, they look at what time it is on the local clock when the neutrinos are emitted and then they look at what time it is on the other local clock when the neutrinos are detected and the difference in time is how long it took the neutrinos to make the trip, but if time is running at a different pace at each location (and presumably all along the trip) then how can they make any sense of the times on the two clocks running at different rates?

I don't see how clocks adjusted by GPS can get around the concern that I have. Consider the atomic clocks at Greenwich and Boulder running at different elevations and therefore running at different rates. If we measured the round-trip speed of light at both locations using their own atomic clocks, we'd get the correct answer of c. But if we used a common time generated by GPS, we will no longer get the correct answer of c at both locations, correct?

 

[PAllen]

I know they cannot measure the one way speed of light, that's why I asked how they can define it since the clocks at both ends will be running at different speeds. To boil the problem down, they look at what time it is on the local clock when the neutrinos are emitted and then they look at what time it is on the other local clock when the neutrinos are detected and the difference in time is how long it took the neutrinos to make the trip, but if time is running at a different pace at each location (and presumably all along the trip) then how can they make any sense of the times on the two clocks running at different rates?

But they adjust each clock to not run at local time, but instead run at the time rate and value of the GPS synthetic frame. Constant re-synchronization is done to achieve this. This is the core of how GPS works - all clocks involved are adjusted for time dilation of all flavors; atmosphere signal delays; sagnac effect; etc. read the paper I linked a few posts back.

I don't see how clocks adjusted by GPS can get around the concern that I have. Consider the atomic clocks at Greenwich and Boulder running at different elevations and therefore running at different rates. If we measured the round-trip speed of light at both locations using their own atomic clocks, we'd get the correct answer of c. But if we used a common time generated by GPS, we will no longer get the correct answer of c at both locations, correct?

Yes, that's true, but not relevant to what they are doing.

 

[DevilsAvocado]

Well, if this doesn't invalidate Special Relativity (which pretty much does), the neutrino speed surprasses so slowly the speed of light that in order for you to send a message to the previous day, it would take longer than your lifetime.

True, they have chosen not to interpret in terms of new physics, which is wise at this stage. However if the results survive further testing... and FlexGunship’s (2) is the answer, we have a problem with causality, no matter how small... IMHO

There is no way to be "a little pregnant" (according to current knowledge).

 

[SeventhSigma]

Wait why did the neutrinos of that supernova arrive 3 hours earlier if we're going to assume this experiment is wrong and c is still the ultimate speed limit? How can neutrinos beat light if they have mass?

 

[hamster143]

This combined with the fact that the neutrino pulse from supernova 1987A would have shown up years earlier than the exploding star's flash of light (at speeds seen by OPERA). Instead, http://en.wikipedia.org/wiki/SN_1987A#Neutrino_emissions"...

So why are the speeds seen by OPERA not achievable by the SN 1987A neutrinos?

I don’t know...

Neutrinos come in multiple mass eigenstates. Strictly speaking, all we know is that _some_ SN 1987A neutrinos arrived within hours of the flash of light. The Lorentz-violating eigenstate could have arrived during the Middle Ages, for all we know. Or arrived two years in advance, but dispersed over the period of 6 months and undetectable above background noise. There is a theory that it arrived 5 hours earlier, but it was only seen by one detector and discounted as a statistical fluke.

Wait why did the neutrinos of that supernova arrive 3 hours earlier if we're going to assume this experiment is wrong and c is still the ultimate speed limit? How can neutrinos beat light if they have mass?

The light from the supernova only appears a few hours after the explosion, because it is initially blocked by cool shock front that is ejected during collapse.

 

[gambit7]

No. The energies are too small by a factor of 1000. The neutrinos cannot be efficiently produced nor efficiently detected at these energies.

Okay so:
a) How were the 1987 neutrinos detected in the 1st place?
b) If we're talking efficiency, utilizing the accuracy of OPERA, isn't it safe to say this is mostly a tagging and timing issue?

If we can tag low-energy neutrinos (whether produced naturally or artificially) then that means we can effectively detect them after traveling the linear baseline no?

So basically, set your initial detector to find low energy neutrinos that are on flight paths toward the 2nd downstream detector, tag them, perfect the metrologies, and then look for them in said 2nd detector. Maybe even piggyback a high-energy beam as a marker? Then filter.

 

[Vanadium 50]

The detectors that detect MeV neutrinos are very different than the ones that detect GeV neutrinos. The operate on an entirely different technology, one that also happens to have less good timing. You can't make Opera do this.

 

[SeventhSigma]

I did the math too and was able to confirm Vanadium's result of a 4 year expected difference if the neutrinos were in fact faster than light. I think I trust that supernova over this experiment. Were there any other differences in the neutrinos themselves (I tried reading the paper but it's too dense for me)?

 

[hamster143]

What bothers me most about this result is not so much the claim of v > c, as the magnitude of the effect. It is way, way too strong.

It is incompatible with QG-inspired Lorentz-violating dispersion relations (it's too strong, by something like 13 orders of magnitude, compared to what we'd expect.) It is incompatible with tachyons (for tachyons, speed goes up as energy goes down, and that would be hard to miss - for starters, MINOS would've seen the arrival time anomaly of ~2000 ns.) The energy scale implied by this value of (v-c)/c is in the MeV range. I could accept a slightly superluminal mass eigenstate with negative m^2 on the same order as mass differences measured in neutrino oscillations; or even a value like those produced in tritium beta decay experiments (where m^2 values down to ~-100 eV^2 have been reported). But none of these values would come even close to producing a 10^-5 effect in the speed of travel at 17 GeV.

It has to be an unaccounted-for systematic error, a large-extra-dimensions effect that increases the strength of QG-induced Lorentz violation, or something completely unexpected. I'm leaning towards a systematic error.

 

[kmarinas86]

I did the math too and was able to confirm Vanadium's result of a 4 year expected difference if the neutrinos were in fact faster than light. I think I trust that supernova over this experiment. Were there any other differences in the neutrinos themselves (I tried reading the paper but it's too dense for me)?

For one, the method of acceleration is different. So if any difference is real, then the geometry and strength and dynamical motion of the magnetic field might play a role. This could lead to a different oscillation signature. It is possible that a difference in the speed may be real while us not having any immediate reasons as to why that may be. It is too early to speculate much further than on Physics Forums though. Answers to this (accounting for whether it is a statistically significant difference or not) cannot be complete or valid at this time.

 

[turbo]

I did the math too and was able to confirm Vanadium's result of a 4 year expected difference if the neutrinos were in fact faster than light. I think I trust that supernova over this experiment. Were there any other differences in the neutrinos themselves (I tried reading the paper but it's too dense for me)?

Put yourself in the position of an observational astronomer, and then extrapolate to the position of a theorist in stellar evolution. Do we know that photons and neutrinos are emitted at the same time in a SN? Do we know that if there is a differential in the emission of copious amounts of photons and neutrinos that it can be constrained to minutes, hours, months, years? Are any of these time-frames relevant if we don't know what happens when a star self-destructs?

We have a lot of stars to look at and supernovas of all types get lots of attention. Still, we don't know all that we need to about the birth, life, and death of stars. We have some compelling models, but our lives are very short and the lives of stars are very long, so there is a sampling problem...

 

[ghwellsjr]

I don't see how clocks adjusted by GPS can get around the concern that I have. Consider the atomic clocks at Greenwich and Boulder running at different elevations and therefore running at different rates. If we measured the round-trip speed of light at both locations using their own atomic clocks, we'd get the correct answer of c. But if we used a common time generated by GPS, we will no longer get the correct answer of c at both locations, correct?

Yes, that's true, but not relevant to what they are doing.

Is it not relevant because the errors caused by the different time rates are too small to matter in this experiment or because the experimenters took the different rates into account?

 

[peefer]

'Just standing back, ignoring the particle physics, looking a this from a nuts and bolts perspective ...

60ns. 18m. This seems too crazy-big to be a systematic error, right? What about this:

GPS-based distance measurements are made at the Earth's surface. Then, most significantly at the OPERA detector, adjustments are made for the detector's position relative to the GPS receiver. So, if the neutrino detector is 1400m underground, and 50m toward CERN, the correction is about -50m. Right? Wrong.

Since the Earth isn't flat like it used to be (sorry, I can't cite a reference for this offhand), two deep holes some distance apart are not parallel. They converge toward the Earth's center. The bottom of the 1400m deep hole at OPERA is in fact 26m closer to CERN than the top of the hole where the GPS receiver is, if you work out the numbers. (The extreme case would be a 1400m hole in New Delhi, India, which is about 1400m closer to New York. With OPERA and LHC only 730km apart, the effect is much smaller, but relevant.)

26 metres. That would quite nicely explain the 60ns premature neutrino detection within statistical error.

Of course, the scientists already must have considered this, right? It sure would be embarrassing if they didn't.

 

[hamster143]

'Just standing back, ignoring the particle physics, looking a this from a nuts and bolts perspective ...

60ns. 18m. This seems too crazy-big to be a systematic error, right? What about this:

GPS-based distance measurements are made at the Earth's surface. Then, most significantly at the OPERA detector, adjustments are made for the detector's position relative to the GPS receiver. So, if the neutrino detector is 1400m underground, and 50m toward CERN, the correction is about -50m.

This is logical, but wrong. :) OPERA is not exactly "underground" (as in, "in an abandoned mine".) It sits just off a 10-km highway tunnel through the mountain. They took two GPS units and measured locations of both ends of the tunnel, and then tracked back from the entrances to the facility to determine its exact coordinates.