I know that most of the datasets from ATLAS or other particle physics experiments are terabytes, but I was just curious is there any place where one can find them to download? And one more question, which software they are using to analyze data and do the simulations? Plus if there is any guide available where they show setting up the software and analyzing some smaller particle physics datasets, I would be glad. I kept searching the Internet for similar things, but unfortunately couldn't find anything.
$\begingroup$
$\endgroup$
6
-
1$\begingroup$ Related: physics.stackexchange.com/q/89233 physics.stackexchange.com/q/9569 physics.stackexchange.com/q/4063 $\endgroup$– dmckee --- ex-moderator kittenSep 13, 2014 at 23:07
-
5$\begingroup$ I've written before that you get access to data sets by asking for them. But frankly, unless you have some history as a particle physicist you're going to need months (or more likely years) just to get up to speed on what the data is, what the tools are, how to run them and how to interpret them. Even for another experimental particle physicist, the jargon that the collider types use is dense and requires considerable puzzling out. $\endgroup$– dmckee --- ex-moderator kittenSep 13, 2014 at 23:11
-
$\begingroup$ @dmckee Thanks for the info, but where specifically do you ask for those data? Is there any contact form or e-mail address for it? $\endgroup$– user20669Sep 13, 2014 at 23:13
-
1$\begingroup$ No, there is not. Because (a) the answer is going to be "No, you can't have it" until they're wrung it dry and (b) they assume that if you have the background to work on this kind of data you'll know how to make the approach (e.g. by asking someone on the collaborations executive committee) and have professional contacts to make the introduction if you don't already known one of those people directly. $\endgroup$– dmckee --- ex-moderator kittenSep 13, 2014 at 23:15
-
1$\begingroup$ May be of interest: kaggle.com/c/higgs-boson $\endgroup$– lemonSep 14, 2014 at 0:22
|
Show 1 more comment
4 Answers
$\begingroup$
$\endgroup$
1
All of the existing answers to this question are now out of date, because CERN has released data for all of its experiments through its Open Data program, including data taken as recently as 2016. There is sufficient data to, e.g. discover the Higgs boson for yourself, along with ample documentation.
-
$\begingroup$ +1 At first I wondered why this question from so long ago was dragged back up out of obscurity, then I saw this answer. This needs more attention as it's definitely the correct answer now, even if a bit dated for the original poster. $\endgroup$ Feb 26, 2020 at 9:35
$\begingroup$
$\endgroup$
2
Geant4 is probably the closest thing you could conceivably get. It's a detector simulation software package with a steep learning curve that's heavily used in the HEP community, and, as far as I know, is available for anyone to download and build. Full disclosure - I did HEP experiment once upon a time but that was 20+ years ago. So if someone currently active corrects me on this point, I'll withdraw the answer - but I doubt if Geant4 would be useful to anyone without at least some training in HEP and detectors. Still, it would give you some sense of what you are up against.
One of the comments mentioned the Kaggle Higgs competition, as far as I know that's already heavily processed data, but in theory it's available to anyone.
I doubt if any collaboration would provide raw data from experiments publicly without some verification that the person requesting it knew what they were doing.
Cautionary note- HEP data analysis isn't like amateur astronomy - it takes a large team. Just look at the length of the author list on any typical HEP paper.
-
1$\begingroup$ Geant remains the core of the simulations, but the standard generators are insufficient. Atlas and CMS and the other LHC experiments have been hard at work on generators. Nor does raw geant give you a simulation of the experiments in questions. As usual you have to provide the framework with a lot of experiment specific context, geometry and readout specifications to get anything like an experimental data stream out of it. $\endgroup$ Sep 14, 2014 at 3:46
-
$\begingroup$ I suspected as much- if the OP were to look at raw Geant, even just the documentation, they might get a sense of the daunting nature of their question. I hate to be discouraging of a question that's good in principle, but in this case it's kind of like walking into an office building and asking if one can have a seat on the board of the corporation headquartered there. $\endgroup$– paisancoSep 14, 2014 at 3:49
$\begingroup$
$\endgroup$
If you are just looking to simulate the data produced at the LHC then the right program to use is Madgraph. It is by far the most popular Monte Carlo generator to use for simulations at the LHC and can produce events for any process you want. Madgraph will then also invoke Pythia and PGS to simulate hadronization and detector simulations if you wish. There are many built in models that you can use for your simulations (SM, MSSM, etc.) and you can also write your own, though that has a steeper learning curve. There are many tutorials online that you can follow to get it working on your machine.
If you are looking for actual data then I believe you have to go the papers published by CMS and ATLAS they often publish tables that you can get the data from.
$\begingroup$
$\endgroup$
Whenever searching for datasets, I always look at Kaggle, Google Dataset Search, and the Ultimate Dataset Aggregator. From doing a quick search, there doesn't seem to be a lot out there. However, recently there has been a resource uploaded to Google Dataset search (screenshot below) from the California Institute of Technology and CERN. This may be worth a look.
