The h Bar

12:19 AM

hi physicist from (T)CS world

"we" (Computer Science) recently had a question asking about software on the LHC supercollider. once had a similar question on Theoretical Computer Science that got closed also.

2

Q: Which programming language is used at the Large Hadron Collider?

First off I am new to computer science (first semester) and do not know anything about physics, so I am appologizing up front for this not very scientific question. I've been wondering how data is processed at the LHC. I am sure CERN doesn't just use one programming language to write code but I ...

the software/hardware on the LHC seems interesting... do you guys have any questions on it? know any good refs?

btw congrats on this recent "hot question"...

7

Q: What is probability current in quantum mechanics?

What is probability current in quantum mechanics? Why define such a thing? I mean the meaning of probability current. I know the formula for it but I just don't get the idea of a flow of probability over time. Does it imply that the probability of finding a particle along dx changes over time su...

quantum-mechanics wavefunction probability

have always thought that "probability current" concepts might be taken as circumstantial evidence supporting a LHV (local hidden variable) theory...

dmckee

1:42 AM

@vzn Short version, major use is made of the tools ROOT and GEANT4 which are both primarily written in C++. Other tools are written may be written in other language. Especially for things like supporting grid submission, and access to siloed data. I've not worked at CERN, but Fermilab uses python these days for a lot of tools.

We used to use a lot of fortran 77 (that was the language of CERNLIB and GEANT3), and I think a few of the event generators that written in it are still around.

vzn

2:48 AM

interesting, thx much!

interested in the analysis software for particle tracks/events... is any of that done in geant? wonder how those "algos" work

David Z

5:00 AM

@vzn not really, probability current is simply a convenient mathematical definition. Its mere existence doesn't prove anything. Plus LHV theories are effectively ruled out by Bell's theorem.

David Z

5:53 AM

@jinawee indeed... I'll get working on processing the info that came out of our community closing experiment

Speaking of which, @Manishearth (or anyone else) did you ever manage to compile any relevant statistics from that week?

ManishEarth

@DavidZ nope, was too busy :/

Anthony

6:52 AM

Ayyyyyyyyyyyyyyo

Physics Meta

7:17 AM

0

Q: Questions with no accepted answers?

I am a new user and I have already answered 19 questions. What annoys me is that in most cases the person who has asked the question does not revert back to accept an answer. I am pretty sure that if not all, most of my answers were pretty helpful. This is most relevant especially in cases where ...

discussion unanswered-questions

Physics Meta

7:30 AM

0

Q: Which answer should a questioner accept if two answers seem completely right?

It seems 2 answers given from two viewpoints are completely right. Also it seems anyone who visits that question in the future must read both answers to get a complete understanding of the subject concerning that question and future visitors read only answers that are accepted and so a questioner...

support questions accepted-answer

user43004

10:18 AM

Hello,
My question is regarding effect of sun's gravity on earth. I want to know that if sun's gravity can change earth's landscape in long duration (i.e. billion of years ) or not?? Means if earth is dead planet and all other planets are not present in the solar system and earth is as near as sun as mercury currently orbiting. And also suppose earth's surface is not smooth (because of hitting with other small bodies at regular interval) then can earth's surface can be smooth in long duration by gravity of sun or not??

dmckee

1:42 PM

@vzn Geant is simulation. Analysis is generally done in tools based on ROOT (which is a big framework built from a horrible class tree including TObject from which almost but not quite everything else derives). Individual experiments will each have their own analyzers an these can be quite substantial software projects on their own (hundreds of thousands of lines of code, scores or hundreds of contributing programmers, and a medium to pretty good score on the Joel test).

Mind you, on big collider projects the analysis makes heavy use of the results from simulation (as in this is a second source of Big Data) to design and understand the selection procedures and efficiencies associated with them.

I should also add that I have started seeing more Javascript tools acting as web-based front-ends in recent years.

APARAJITA

3:12 PM

@user43003 , that's a good question but I don think that will happen, wen mercury's surface is itself not smooth, earth's cannot be either.. Gravity of sun effects all the planets equally, whether there is only earth in the solar system or not.

vzn

dmc thx what is the "joel test"?

is a lot of the processing done offline or is there a significant realtime component?

my understanding is that in some experiments there is major filtering of snapshots of tracks to exclude/include those under study. ("selection"?)

dmckee

Joel test: joelonsoftware.com/articles/fog0000000043.html

vzn

wondering how the filtering works, it seems similar to an AI type process...

dmckee

There is considerable data reduction between the apparatus and disk this is controlled by the "trigger" which usual happens in several steps starting with hardware (lots of FPGAs in recent years). This is a high-speed processing challenge.

After that there is some, minimal, near real-time crunching for monitoring purposes (so the shifters can see any problems developing and take corrective action).

But the "real" analysis is all post-facto: it can take months to develop an analysis and days or weeks to run it over a significant portion of the data.

What kind of filtering schemes are used is a very complicated questions and depends on the machine and the physics of interest.

vzn

re JS criteria, nice... very solid/mature software engr/methodology there...

dmckee

3:19 PM

Various branches of AI research are used as the basis for some of this. Pattern recognitions is very big early in the process as tracks and showers must be identified in a host of discrete data points.

vzn

yes had heard a lot of the data analysis was offline

what kind of pattern recognition algos are used?

dmckee

The end of the processing chain is mostly worried about statistical methods to understand how well things have actually been measured/detected/ruled-out.

vzn

so the "trigger" attempts to exclude most "uninteresting" events from initial archival?

dmckee

@vzn What experiment? What instrument? What channel are they looking for. Particle physics is a very diverse field.

vzn

yes & alas there seem not to be a lot of surveys about the software aspects given the physics centric motivation

eg surely there was massive software engr aspects of Higgs but it tends to get glossed over sometimes

dmckee

3:23 PM

@vzn Well, it attempts to include all the "interesting" stuff, while sampling the discarded bits and keeping the data rate manageable.

vzn

in media accts etc

"sampling discarded bits"?

dmckee

@vzn You need enough data to be able to show that you know what was thrown away.

This will have been simulated up one side and down the other during design, but you have to be sure.

vzn

reminds me of null hypothesis testing. (havent used it only heard theory)

dmckee

I don't do collider physics--I've done fixed target and low-background neutrino stuff--so my understanding of exactly what they are up to now comes from watching talks given by colleagues.

vzn

ok

dmckee

3:26 PM

And from the bits of their stuff we're are adapting to use with large scale TPCs (time projection chambers, a kind of 3D tracking system).

vzn

←like "big physics" lots

found this recently & refd CS questioner to it

9

Q: Use of Monte-Carlo simulation in High-energy Physics

I've been doing some research into the analysis used in particle physics when determining the significance of a finding (e.g. the recent Higgs candidate was announced as a boson in the 125-126 GeV/$c^{2}$ mass range with a $5\sigma$ significance). I believe this confidence level is determined by...

particle-physics experimental-physics simulation statistics data-analysis

also found 2 nice questions re intersection CS/physics

planning to blog on it sometime sooner/later

have you used Geant/ROOT?

dmckee

@vzn Yes. I've written tens of thousands of lines of code (accepted for production) against ROOT and thousands against Geant4. I also used CERNLIB and Geant3. And I've written event generators from scratch when there wasn't the right support in Geant.

I've worked on approximately six large, collaborative codes.

Used CVS, SVN and git for version control on those projects. Used three or four different bug trackers. Several build systems. Three different silo systems. I think four different cluster submission schemes, but I may have lost track as they mostly look just similar enough to cause confusion.

vzn

have used CVS/SVN etc also

re joels criteria have seen more continuous build systems recently also

vzn

3:54 PM

11 hours ago, by David Z

@vzn not really, probability current is simply a convenient mathematical definition. Its mere existence doesn't prove anything. Plus LHV theories are effectively ruled out by Bell's theorem.

ah yes the decades old bell test....

have researched that quite a bit...

found interesting interpretation once in

"Understanding quantum physics, a users manual" by morrison

slightly unorthodox.

re probability current

a great quote might dig it up sometime

re LHV theories recently renewed interest in solitons...

know its highly contrarian to say least but find the theory extremely interesting/compelling

here are some wild musings

solitions, CA, QM, & disagreeing with scott aaronson Tmachine blog

anyway thx for feedback guys cool stuff

gotta blog on some of my favorite particle physics books sometime maybe

so many good ones

APARAJITA

4:27 PM

@vzn, can u name some good books on particle physics ??

vzn

4:59 PM

oh ok sure!

there are so many it depends on your tastes.

mine tend to run towards pop sci.

two standout ones among many

amazon.com/The-Particle-Odyssey-Journey-Matter/dp/0198504861

amazon.com/Taming-Atom-Hans-Christian-Baeyer/dp/0486414477

von baeyer is esp interesting in his descr of (single) atomic traps, revolutionary new technology at the time now somewhat more widespread

Transcript for