AMI and pyAMI

Last update: 18 Nov 2024 [History] [Edit]

AMI web interface is the dataset metadata catalogue. It is a “mediating” application, which means it correlates information from many sources.

pyAMI is the python client for AMI. It is in the ATLAS release and also available standalone on lxplus or as a tar file for download to your laptop. pyAMI documentation explains how to use pyAMI.

Making an account in AMI

To use AMI you must have a valid grid certificate and be registered with the ATLAS VO.

If this is your first time accessing AMI through the web interface you will need to set up an account. If you have not already done so, follow the instructions described in Registering with AMI.

Finding datasets

Exercise

Using the AMI web interface, find a simulated dataset.

Instructions

We will look for the ttbar sample: mc23_13p6TeV.601230.PhPy8EG_A14_ttbar_hdamp258p75_dil.deriv.DAOD_PHYSLITE.e8514_s4159_r15530_p6491

  • From the main AMI web page click on Datasets, select mc23 in Simulated data and then select some tags to find our dataset
  • For example, select Data type and use the wildcard %DAOD_PHYSLITE%
  • Then select dataset number and type the ttbar process we have been using %601230%
  • If you click View Selection you should see quite a few datasets
  • Find the one that has a version matching e8514_s4159_r15530_p6491 and there you go!
  • You can click on the arrows under details to see all the metadata associated with that sample.

tip The exact ttbar dataset listed above might have been deleted (according to ATLAS data deletion policy) by the time you follow this tutorial, so you might not find it. You can probably still find another similar dataset with the only difference being a higher p-tag number.

tip If you know exactly what dataset you want to find, try going to Datasets and then to Simple Search. You can again select e.g. mc23 and then provide the name of the sample you’re after, and you will find it. Again, wildcards can be included with %. In the above example, you can try finding all the same datasets with all p-tags (derivation versions) by looking for mc23_13p6TeV.601230.PhPy8EG_A14_ttbar_hdamp258p75_dil.deriv.DAOD_PHYSLITE.e8514_s4159_r15530_p%

tip Different tools in ATLAS deal with “scopes” differently. The scope of a dataset is the piece that sometimes precedes the dataset name with a colon after it, like mc23_13p6TeV:. AMI prefers that you not include the scope in the name. Rucio prefers that you do. Where possible, the tools will try to do the right thing.

tip These searches can also be done via the pyAMI command line tools. For example:

# Set up the pyAMI tools
lsetup pyami
# Get a grid proxy for authentication
voms-proxy-init -voms atlas:/atlas
# Search for your dataset
ami list datasets "mc23_13p6TeV.601230.PhPy8EG_A14_ttbar_hdamp258p75_dil.deriv.DAOD_PHYSLITE.e8514_s4159_r15530_p%"

Notice that the searches might not return the exact same results — can you think of some reasons why not?