This repository is a collection of notes I've written while working with the O2 software. Keep in mind that I am still working on this repository and it is definately not complete. If you have any suggestions feel free to message me on mattermost :)
If you are new to the ALICE experiment, there are a lot of acronyms used here. You can find some of them here: ALICE acronym list
In this repository you can find instructions on how to get started with the ALICE O2 Framework. Keep in mind that this framework is constantly evolving and many files in this repository can be outdated.
Here are some links you may find useful:
- ALICE O2 documentation - O2 documentation,
- @saganatt's Learning O2 repo ;),
- @lbariogl, @saganatt & @zchochul O2Physics Tutorial for CF.
Keep in mind most of us use Ubuntu 18.04 / 20.04 / 22.04 or CentOS. Firstly, build O2 with alibuild as explained in the ALICE O2 documentation. So, you need to follow instructions from Installation via alibuild page, but DO NOT follow instructions on "build packages" there! You also need to acquire GRID certificate, it's explained in detail here -> Get a Grid certificate.
If you plan to only do the analysis in the O2Physics you can init only O2Physics (O2 will be created to, but in the alice/sw/ directory). If you choose not to init it you will speed up a bit the updating process. You will not need to update the O2 directory.
Keep in mind that setting everything up for the first time may not be easy and can take quite a long time and that's completely normal.
WSL 2 error - `c++: fatal error: Killed signal terminated program cc1plus`
That means the compilation uses too much RAM of what is available in the WSL2, compile with flag -jX, where X should be the amount of cores you want to engage, for example: -j6. Keep in mind not to use more than what you have available on your machine ;) I would advice you to use this flag even if you are not using the WSL. It would extend building time, but it's safer if you want to use your computer for other things during building.
aliBuild -j6 build O2Physics --defaults o2ALICE O2 Framework is constantly evolving so remember to stay up-to-date with the official repo. It would be ideal to update it everyday, but try to do it at least once a week. To update it you need to do the following:
In alice/O2 type
git pull --rebaseIn alice/O2Physics type
git reset --hard
git pull --rebaseIn alice/alidist type
git pull --rebase upstream masterLast, but not least in alice/ type
aliBuild build O2Physics --defaults o2 --debugBut you can also use the update.sh script in the files of this repo. Paste it to the alice/ directory and don't forget to use the chmod command before ;).
Unfortunately, it takes a while and you have to do it quite often due to the constantly evolving system. You might want to use a bash script to update (look for aliBuild.sh in this repository). Sometimes there are some errors if you are doing an update from the different branch than master.
Sometimes it is crucial to remember version you are on. In all repositories (in alice/alidist/, alice/O2/, alice/O2Physics) type git log. You can see the last commit that you have in your version. To remember version you are on write down hash number.
Here is an example ;)
hash number in this case is: 291457fce1bf6da3c0ac5cf1afe53a12624a3a6c
To get AOD file go to -> Analysis train : Run3_Conversion and scroll down. Choose train number you're interested in and click on run number (for example 246). Now you have two options, you can download:
Smaller file:
Then you need to click on the Test Results (in the middle). Scroll down and find full train option and then click on output. Now you're supposed to see a lot of files, but you're only interested in AO2D.root. Then you need to only click on it and it will start to download.
Bigger file:
You need to find the table with {Status, Files, Processing, Merging, Statistics} rows. In the Processing row click on the processing progress. It will redirect you to the alimonitor.cern.ch/catalogue. Click on the output directory you are interested in. Then scroll down and find AOD directory, then enter another directory (I think it doesn't really matter which one) and then download the AO2D.root file. This may take a while, as the files are quite large. Please save the path your AOD is from (I have a .txt file with [names : paths]).
Information about trains (job details) can be found here -> alimonitor.cern.ch/job_details.
Quick version
Go to the Hyperloop page and look for the datasets linked to the dataset you are using. So now let's do the example ;) Go to the Hyperloop page and choose the desired dataset. I will be doing that for the LHC22o_pass2_test_small (DATA) (link). Under the name of the dataset there is a jira ticket so sometimes there are crucial information there. Look at the ** Productions** section. You have there a listo of runs, remember a run number from there and look for it in the MonALISA page that's linked above.
This is a hint where to look for the link!
When you follow the link look for the run number I've asked you to remember. CLick on the link of its output directory. It should send you to the MonALISA repository. Now choose one of the directories of the type o2_ctf_run527734_orbit and than go into whichever directory (so for example 001) and you should be able to find there a AO2D.root file.
Advanced version :)
Firstly look on the DPG page for the recommended datasets for each period (link to the DPG page). Than you can go to the MonALISA page (link to the MonALISA catalogue). Here you need to fo to alice/data and go to the year you are interested in (on the DPG page there were datasets LHCXYz_passN, so to find desired dataset go to the year 20XY). At the time of writing this, I know that the calibration of the TPC for the high interaction rate (IR) may not be fully correct. That for sure will be changed in the future ;)
QC summary data -> link to the spreadsheet.
example file I've been using for now: /alice/data/2022/LHC22o/528347/apass3/0130/o2_ctf_run00528347_orbit0002457600_tf0000000001_epn136/001.
If you can't even enter alimonitor it is possible that you haven't added your GRID certificate to your browser correctly. Follow instructions on the Get a Grid certificate page.
You can find useful examples on how to write your code in ~/alice/O2Physics/Tutorials/src/.
Then place your code in C++ somewhere, for example: ~/alice/O2Physics/Tutorials/src/. Remember to always change CMakeList in the parent folder (in this case ~/alice/O2Physics/Tutorials/) when you add a new file to your analysis. Without that step, you won't be able to build it. When you add a new file remember to use cmake . command before building (it will be explained in the How to build a task section).
Easiest and the most safe way to build your task is to simply do the alibuild (aliBuild build O2Physics --defaults o2 --debug in the ~/alice/). But you can also ask explicitly to build a given task using ninja.
To use ninja go to ~/alice/sw/BUILD/O2Physics-latest-master/O2Physics.
As soon as you enter a lot of text should appear and it should look like this:
In ~/alice/sw/BUILD/O2Physics-latest-master/O2Physics enter ninja and O2Physics environment (alienv load ninja/latest O2Physics/latest ). Then build your task using ninja stage/bin/<your-analysis-file>. If you don't know what you should type in <your-analysis-file> place, open CMakeList.txt and see how your analysis is called there.
Keep in mind that if you add a new file or modify CMakeList you need to use cmake .
Then, after building part, copy this builded file to directory with AOD file (cp stage/bin/<your-analysis-file>). You can skip this step (copying) when you use ninja install instead of ninja.
You don't see a lot of text after entering the directory
If upon entering the ~/alice/sw/BUILD/O2Physics-latest-master/O2Physics you don't see much text you probably don't have eval "$(direnv hook bash)" in your ~/.bashrc file. Type it there and open a new terminal.
In general, in ~/.bashrc you should have following lines:
export ALIBUILD_WORK_DIR="$HOME/alice/sw"
eval "`alienv shell-helper`"
eval "$(direnv hook bash)"
This can be written anywhere and it will work, but try writing it in one place, for example at the end of ~/.bashrc file
Couldn't find O2Physics/latest
So there are two options:
- When you built the O2 there was an error and it didn't compile completely. You have to correct the mistakes and try to build again.
- If it did in fact compile completely there is another error (I had it only once but it is possible xD). You need to change the
/latestpart for something you have in the~/alice/sw/ubuntu2004_x86-64/O2directory. But when I fully built again, the error disappeared.
In your directory with an AOD file type:
alienv enter O2Physics/latest and alien-token-init your-cern-nickname
Then to run your code you need to type:
./o2-analysistutorial-simple-analysis --aod-file <aod_file_name> -b
But sometimes you need to add more for it to work. When you see Exception caught: Couldn't get TTree O2[sth] you need to add a helper task that creates this [sth] table to your analysis.
Example
While performing PID analysis you need to type for example:
./o2-analysistutorial-simple-analysis --aod-file <aod_file_name> --aod-memory-rate-limit 100000000000000 -b | o2-analysis-pid-tpc -b | o2-analysis-pid-tof -b | o2-analysis-trackselection -b | o2-analysis-tracksextention -b
Those o2-analysis-pid-tpc and other tasks are called helper tasks and they are used to produce tables that will be later used in the analysis. But keep in mind not everything might be useful in your case. For more examples look at the files section of this repository.
The -b option stops GUI from showing. In --aod-memory-rate-limit 100000000000000 option tells you the largest shared memory size in bytes that is allowed to be filled with tables that are read from input files.
In the .json config file in the internal-dpl-aod-reader section you should type not the path to the file but attach the .txt file as follows:
Segmentation fault and returning to the command line
Make sure you use -b after every helper task. For example:
./o2-analysistutorial-simple-analysis --aod-file <aod_file_name> --aod-memory-rate-limit 100000000000000 -b | o2-analysis-pid-tpc -b | o2-analysis-pid-tof -b | o2-analysis-trackselection -b | o2-analysis-tracksextention -b FDD conversion
For converted data when using train number larger than 267 we did not include fdd converter helper task. We received an error type: Couldn't get TTree O2fdd.
Couldn't get TTree.
Sometimes you may get an error, for example: [ERROR] Exception caught: Couldn't get TTree "DF_2853960297589372650/O2v0dataext from <your AOD file>". It means that this v0dataext couldn't be found in your AOD file or it wasn't produced by any attached helper task, helper tasks are attached to your analysis like in the Example toggle above.
So now you know which table is missing but finding the right task to add is still very confusing. Don't worry, there are two paths you can choose ;)
- Easier one
Here is tutorial from ALICE O2 documentation- Tree not found and it's useful if you want to find more common tasks. So basically you need to enter ALICE O2 documentation - helper task tables and then you just search the table you are missing and you will find the task you need to attach. - More difficult one
So now, you need to enter directoryalice/O2Physics. Then imagine you are missingpidtofkatable. You will need to use grep command (if you didn't know it already check this page ;)). Here is an example to make it easier:
grep -rnw . -e "PIDTOFKA"
grep -rnw . -e "PIDTOFKa"
grep -rnw . -e "pidTOFKa"Now you should have a list of files where this phrase occurs. Next you need to find out the name of the file which produces this table. And then you need to go to CMakeLists.txt file and there you will find the name of the task you are looking for. If you are not familiar with CMakeLists don't worry! You can see that it has many sections but they all look alike, and here is an example:
o2physics_add_dpl_workflow(<task-name>
SOURCES <C++ file>
PUBLIC_LINK_LIBRARIES O2::Framework O2::DetectorsBase O2Physics::AnalysisCore
COMPONENT_NAME Analysis)So you need to find the task-name which will correspond to the C++ file you've found using grep ;)
Fatal errors with non-existing CCDB entries.
While trying to run various O2 tasks/tutorials you can run into fatal errors with non-existing CCDB entries for specific data.
Usually adding --isRun2MC to executables (likely o2-analysis-timestamp) might help. We also added --isMC to o2-analysis-event-selection.
For example: ./o2-analysis-mm-dndeta --aod-file AODmc2.root --aod-memory-rate-limit 100000000000000 -b | o2-analysis-pid-tpc -b | o2-analysis-pid-tof -b | o2-analysis-trackselection -b | o2-analysis-trackextension -b | o2-analysis-event-selection --isMC -b | o2-analysis-timestamp --isRun2MC -b
GRP magnetic field information.
- GRP with magnetic filed information for converted Run 2 data CCDB path:
GLO/GRp/GRP - GRP with magnetic filed information for Run 3 data CCDB path:
GLO/Config/GRPMagField(with the new GRP objects structure)
Running command ./o2-analysistutorial-simple-analysis --aod-file <aod_file_name> -b (and others of that type) creates .root file and .json file.
To enter AnalysisResults.root file enter O2Physics environment (enter O2Physics/latest) and type: root -l and then new TBrowser. (-l is optional but it runs root without additional information about it).
You can change histograms manually or you can write macros and run them on .root file. If you've never written any root macros consider this repository. You can find more on root and how to use it here -> root.cern/manual/.
There has been a series of tutorials led by Victor Luis Gonzalez. It's an amazing source of examples! Most of those sessions are recorded and all of them have slides attached to them, I've listed them below, feel free to check them ;)
Tutorials by Victor Gonzalez:
- Getting started: Configuring tasks, selecting collisions, filling outputs
- Basic error recognition, json config files
- Templates and process switch
- Templates continuation, how to not repeat code
- CCDB: Correcting for NUA, NUE, ...
Tutorials done for ALICE by David Dobrigkeit Chinellato -> link to the tutorial on indico
O2 Analysis Tutorial -> link to the tutorial schedule on indico
This step may seem obvious, but sometimes you might find grep command useful. Especially, when you want to find tutorials with functions you are interested in.
Sometimes you may encounter an error. Screenshot of a problem may not be enough to help you. You will need to send a debug file, which simply contains everything that appears to you on the screen after building or running something. So if the command shows you something on the screen simply pipe it to debug.txt like that:
command > debug.txt 2>&1You can find useful information on the ALICE O2 documentation.
Firstly, create a JIRA ticket for your analysis on the alice.its.cern.ch/jira. You have to log in. Click on the Create button on the top of the page. You need to fill the table. For example:
- Project ->
PWG-CF (PWGCF), so here select your PWG - Issue Type ->
Analysis, only those will be linked to the Hyperloop system, - Summary -> , it should be as broad as the analysis you are doing, but it should be informative
- Component/s -
PAG-CORRELATION-EBYEchoose your PAG - Run 3 analysis -
Yes, always - Analyzers - add analyzers from your team, remember to add yourself ;), those people will be able to work with you on this project,
- Data Set - you can write there anything,
- Public Result -
Yes
You do it once per analysis. Now you should be able to to see it on the AliHyperloop page. It can take some time to synchronize JIRA with Hyperloop.
When you click on your analysis you will be able to see all thge wagons that are created for this analysis (now it's empty click on the Add new wagon). In wagon name write the wagon name, then version of the O2Physics (choose the latest version of the repo, but you can choose a different one if different version produces errors for you). In workflow name choose the main task you want to run. And then you save.
You can later configure it. As you know tasks usually need a lot of the helper tasks. You need to add them in the dependencies. You also need to add Datasets. Choose if you want to work with Data or MC, then choose type of collision and a period you are interested in. You will get a list of datasets available for now. But this part is also constantly evolving too, so there are a lot of new datasets comming ;)
Now, the good information for the people previously working with the AliPhysics, the tests are done here automatically ;) When the test is done you should see a ✔️ ☀️ or ✔️ ❗. This means that the analysis went okay. But if you see a 💣, it means that something went wrong. Maybe there weren't enough helper tasks added or maybe the dataset is faulty. But if everything is okay you can click on the ☀️ and find test output with the AnalysisResults.root.
You have to create a request on the ALICE's O2 Hyperloop Operation mattermost channel simply by writing on the chat to the train operators. It is crucial that in order to turn on the long train beforehand to have an acccpeted test with small dataset linked to this long train. That's what I can see after this request:
When it's finished it will say Done. Then you need to click on the Train number and then on the Merge output and you need to download the AnalysisResults.root file and open it with root.
Example codes are available here: O2Physics/PWGCF/.
First of all you need to simplify your AOD. After typing alienv enter O2Physics/latest run a shell script called prod.sh I'm changing it quite frequently so it's just a concept, but I'll try keep it up to date. To run it you need to simply type:
./prod.sh Make sure you have [`prod-config.json`](https://github.com/zchochul/AliceO2/blob/main/prod-config.json) file, you can download it from this repo. You need to change the "aod-file" part to match with your .root file/files (keep in mind this file also evolves, so this file may not be the latest version).
I am working mostly in the FemtoUniverse on the track-Phi task and also track-track. To run track-Phi task use:
./post-phi.sh post-phi.jsonYou need to be in the O2Physics environment (type alienv enter O2Physics/latest). Files mentioned above can be found in the files of this repository. I try to update it frequently. If by any chance you have any errors while running that, please message me on mattermost.
You need to make sure your aod is properly produced for femto analysis. Using typical AOD files will cause errors (for example: Couldn't get TTree "DF_2853960030894995121/O2femtodreamcols" from <your-AOD-file>).
Another problem I've encountered was the problem of leaves in several trees not filling up, make sure that you've changed ConfCutPartOne and ConfCutPartTwo value to the one from cutculator in run command and in code itself.
This directory was first created during my bachelor studies (2022). As a base we've used the FemtoDream directory, but we are doing quite different things with it. During my master studies (2023) we moved to FemtoUniverse to upgrade FemtoWorld and do it in a more efficient way.
Special thanks for the help with developing this software goes to my supervisors dr hab. Łukasz Graczykowski, dr hab. Małgorzata Janik, dr hab. Georgy Kornakov and prof. Adam Kisiel. They introduced me to the beautiful world of particle physics, thus fulfilling my childhood dream. Thank you so much for your patience in answering the multitude of questions I have. I am also very grateful to Maja Kabus, who helped me and continues to help me understand this complex software and Wiola Rzęsa for all the help she gave me during the years with macros and physics questions. I would also like to thank every person I have not named who has helped me develop this software. I am very grateful for your comments and suggestions 😊
Version of FemtoDream FemtoWorld is based on is:
alice/alidist/commit hash: 8ebd919bb75e468cb1a994f1012abfa22235ff2falice/O2Physics/commit hash: c3dcdbca87dc9535de85f9837ad777de475ceee3alice/O2/commit hash: 291457fce1bf6da3c0ac5cf1afe53a12624a3a6c
static constexpr ID Pion = 2;
static constexpr ID Kaon = 3;
static constexpr ID Proton = 4;
static constexpr ID Deuteron = 5
Here I've listed some steps but if it doesn't work in your case please follow @saganatt's github instructions they are better ;)
According to this tutorial create fork of O2Physics repo. Then setup user configuration on your local computer:
git config --global user.name "<Firstname> <Lastname>"
git config --global user.email <your-email-address>
git config --global user.github <your-github-username>
Then (while at your O2Physics directory) point git to this fork with:
git remote add origin https://github.com/<your-github-username>/O2Physics
When you use git status you will be able to display the state of the working directory and the staging area. It lets you see which changes have been staged, which haven't, and which files aren't being tracked by Git. If you want to add files to your branch (for example femtouniverse) use:
git branch -vvv<-- it tells you on which branch you are ongit checkout <desired-branch><-- in this way you can switch to your desired branchgit fetch upstreamandgit rebase upstream/master<-- to update the master branch of your forked repositorygit status<-- everything not added should be colored redgit add <file><-- adding file to commit, now ingit statusthey will be colored greengit commit -am "<commit-name>"<-- submitting a commit, (a for all files and all changes)git push -f<-- pushing
The -f is actually required because of the rebase. Whenever you do a rebase you would need to do a force push because the remote branch cannot be fast-forwarded to your commit. You'd always want to make sure that you do a pull before pushing, but if you don't like to force push to master or dev for that matter, you can create a new branch to push to and then merge or make a PR. This action forces your remote repo to ignore differences in commits between your local and remote versions.
If you want to add your file not to your branch only but also to main repo you need to make pull request.
and then after clicking on 1 commits ahead this page will show up and on it you need to create pull request
In order to add changes to official repo, your code needs to pass clang formatting test.
What you need to do is quite easy, just enter clang environment by using alienv enter Clang/latest and then in the directory you have files you want to add, type find . -iname "*.h" -o -iname "*.cxx" | xargs clang-format -style=file -i. Keep in mind that this command changes only .h and .cxx, but you can add more or change to fit your files.
`git reset --hard upstream/master`.
Branch is not following the branch on remote
You need to use:
git fetch origin
git branch --set-upstream-to=origin/femtoworld