On Fri, Mar 30, 2012 at 3:07 PM, iovane@inwind.it iovane@inwind.it wrote:

There was a meeting in Gran Sasso on Wednesday. You can see some of
the slides at http://agenda.infn.it/materialDisplay.py?materialId=slides&confId=4896

I found particularly interesting the ones by Maximiliano Sioli, where
he explained the two mistakes found in the OPERA data acquisition
chain and how, after correcting for their best estimate of their
effects, the time of flight is compatible with a speed of c.

I saw the webcast of the event. Some people did give the OPERA
spokesman a hard time, and he admitted to not having fully checked
everything they could have. Ah well, everyone makes mistakes. There
will be another run with neutrinos spaced by 100 ns in May. If all
four experiments in LNGS give the same result this time, I suppose the
case will be closed. It will also be very interesting to see the MINOS
results.

In any event, from a time-nut point of view this is quite exciting. It
is the first time neutrino speed is measured with this precision. I
think this will pave the way for future experiments using precision
geodesy and time transfer.

Cheers,

Javier

El 30/03/2012 16:55, Javier Serrano escribió:

Thanks for sharing them

The devil is ever in the small things :) It is truly very interesting

This is true. And also perhaps the use of neutrinos for time transfer
purposes :) Nowadays nort their generation in a controlled way either
the detection is easy, but who knows in the future...

Regards,

Javier

Hi Javier,

Thanks very much for posting the link to the presentations.

For those of you who just want a summary of the resolution of
the "neutrino faster than light" problem, here's what happened:

1. For several years an optical cable connector was loose. I have
   attached photos from pages 7 and 8 of the G._Sirri.pptx where
   you can see the actual connector and waveforms, before/after.

2. They used a Vectron OCXO to generate timestamps within each
   0.6 second measurement cycle. This oscillator was found to be high
   in frequency by 0.124 ppm. Thus, depending on where within this
   0.6 s interval the timestamp was made a timing bias of 0 to 74 ns
   would occur.

Javier -- if you have contacts there, it looks to me like they forgot
to include OCXO frequency drift effects into their analysis. What
they did was compensate for linear time drift (which assumes a
fixed frequency offset). They call the 124.1 ns/s time drift "stable"
since 2008. What evidence do they have for this? We know that
OCXO will drift in frequency over time; the time drift is quadratic.
The time drift rate may be 124e-9 today, but it probably wasn't last
month or last year, etc.

/tvb

I have to say that in general I have been staying clear of this thread.
But its really a surprise that they are that sloppy and basing the results
on a Vectron OCXO. Not that I have ever had a complaint about those.
It just seems like the stunt I would do in the basement on my surplus
accelerator.
Regards
Paul.

On Fri, Mar 30, 2012 at 12:57 PM, Tom Van Baak tvb@leapsecond.com wrote:

On 03/30/2012 04:55 PM, Javier Serrano wrote:

As in any research, even a failure is worth reporting, as people can
learn from it. Many forget this and tend to select their results which
can be a danger as you can go into bad science that way.

Will be quite interesting to follow.

Cheers,
Magnus

There are failures and there are failures.

A negative result is a failure that is worth reporting.

A failure due to an improperly mated connector...  not so much.

When doing a complex experiment, you have to be an absolute SOB about
everything. "You cannot inspect in quality".

YMMV,

-John

================

On Fri, Mar 30, 2012 at 7:42 PM, J. Forster jfor@quikus.com wrote:

That is saying to anyone who wants to do a similar experiment in the
future that they need at least two redundant systems. I believe that
is quite a valuable lesson.

Concerning the OCXO, one has to bear in mind that this experiment was
not meant to measure time of flight, but rather neutrino oscillations.
The message for me here is that it's good to publish all your designs,
including gateware sources, as soon as possible, but I don't know how
compatible that is with today's highly competitive scientific world.

So I think there is an important lesson behind each one of the two
issues. Of course this is easy to see from outside and after the fact.

Cheers,

Javier

Not quite, IMO. You need to do "sanity checks". When you are doing science
in unknown territory, you need to eliminate everything you can, as a
source of error.

Remember the problems Perkin Elmer had with the Hubble mirror?

Was there any real competition to this experiment? Seems like they have a
lot of very, very big, expensive, unique hardware that can't exactly be
bought at Radio Shack.

Which is more important? Getting it fast, or getting it right?

-John

=============

I find it very curious that the 'master clock' computer is very clearly a
desktop class machine rather than something in a sever rack with controlled
environmentals, power, etc.

Bob

On Fri, Mar 30, 2012 at 4:03 PM, Javier Serrano <
javier.serrano.pareja@gmail.com> wrote: