Super-K Code Example 3: Ntuples

This example is similar to the first two, except that you will now create an ntuple, and use the PAW program to view it. The PAW tutorial may help here, too.

Ntuples are quite a powerful feature of CERNlib HBOOK/PAW data analysis software: they allow you to look at the variables associated with a bunch of events, and plot any variable versus any other; you can also make selections using other variables. For instance, it is very easy to ask PAW to do something like: "plot a vs b for all events for which c is greater than 2".

This example uses a "quick and dirty" way to generate ntuples. The really proper way to do it is to create the ntuple file and write to it directly from HBOOK calls inside the Fortran source code. This is most efficient for processing a lot of data. However the method I find more convenient when you're just playing around with data, and developing analysis methods, is the following: I create a text file containing a list of events and the variables you are interested in. Next, I use a PAW kumac to create the ntuple and fill it from the text file, from inside PAW. This works fine for small amounts of data, and it's easy to make modifications.

Files for this example

Copy the following files from

/var/phy/project/hep/neutrino/work/schol/superk/examples

• example3.F: contains the source code for the program
• example3.sh: a shell script for running the program
• Recopy the Imakefile used for examples 1 and 2, which has had instructions for compiling example3 added to it.
• example3_setup.kumac: a list of PAW commands which will create the ntuple from the event data text file.
• example3.kumac: a list of PAW commands to do some plotting using the ntuple.

Take a look at the differences between example1.F and example3.F.

• Before the file loop, a text file, example3.out is opened using a Fortran open statement.
• For each event in the hit loop, the output is written to the text file ("logical unit number" 22) rather than dumped to the screen ("logical unit number" 6).
• At the end of the event and file loops, it closes the text output file.

Compile and run this program in the same way as for examples 1 and 2. You should get less output to the screen (because the output goes to the file instead of the screen), but you should get a file called example3.out created. This contains the list of events and their variables.

Creating and working with the ntuple

• Run the command "paw". You should get a command line interface.
• Enter the following line:

  exec example3
  

  This calls example3_setup.kumac to create the ntuple in memory, and reads the text file to fill the ntuple. It then makes a plot of charge vs time.

Stuff to try

After you've gotten this working try some of the following:

• Try nt/print 1 to see what's in the ntuple (it's ntuple no. 1)
• Try plotting other variables versus each other.
• Try putting "cuts" on the variables. For example

  nt/plot 1.q%t t>500&&t<1000

  plots charge vs time for all times within 500 and 1000 ns.

• Change the Fortran source code to add new variables; modify example3_setup.kumac accordingly, to add the new variables to the ntuple. For instance, add idtgsk to the variable list: this is the trigger type of the event.
• Try plotting distributions of single variables: for instance nt/plot 1.q will plot just a histogram of charge. Try putting cuts on these distributions, too.