Intro: Numerical batch mode is useful when making a large number of repetitive calculations

Note: any information that I believe is helpful will be noted with *'s and will appear at the relavent steps.


*if there are still questions that this guide leaves unanswered it is always prudent to check the num_bath.pl script in your work directory. It contains a fairly straightforward explanation of numerical batch and its uses.

1. You will need to start out with some scattering process (i.e. plug in the beams, energies, and any restrictions on the diagrams you're using), and select the square diagrams option
*it is not stressed in any of the comphep documents, but it is essential that the following step comes before anything else, so don't forget.

2. Choose symbolic calculations, write your code, and then use the c-compiler
*num_batch.pl will tell you to cd to your results folder and run n_comphep after writing your code, but i've noticed it often causes problems, and have seen none if you use the GUI c-compiler instead. It still creates customized batch.dat file you're looking for in the end.

3. In the numerical calculation menu go ahead and set up the calculations for your first subrprocess (cuts, model parameters, distributions)
*in the vegas menu the options Itmx and ncall correspond to the accuracy of your integrations; Itmx = numer of integrations, and ncall = number of points used in each integration

4. Now that you've got your first process set up the way you normally would, quit n_comphep (hit F9 or kill the window)

5. Back in your work directory, run ./num_batch.pl, it should run through your processes and ultimately create a filed called batch.dat in your results directory.
*Now what you've just done is take the same parameters you set for your first subprocess and used them as a template for your other processes. If you'd like to calculate different things for different subprocesses you can use a handy little command that's mentioned in num_batch.pl or you can manually change the batch.dat file with your text editor of choice
**There are a couple of other fancy things like sending our results to a different directory, but again it's covered in num_batch.pl very clearly

6. You're ready to run batch calculations for your scattering process. The most simple way to do it is simply run ./num_batch.pl -run within your working directory.
*there are several options you can run, all of them are explained more in depth in the num_batch script (i told you it was handy). The important ones though are "./num_batch.pl -run vegas," "./num_batch.pl -run max," "./num_batch.pl -run evnt." These commands run ONLY the: vegas calculations, maxima calculations, and event generation, respectively. Keep in mind that to generate events you need to calculate maxima, and to calculate maxima you need to run vegas calculations.

7. A quick way to display your vegas results is to run "./num_batch.pl -lcs results/prt_*"
*However I find it more useful to work with the actual output files. All Vegas events are in files labeled prt_(session #) and all events are stored in events(session #).txt

Special notes: It is not made entirley clear in num_batch.pl, but to change the number of events generated one needs to directly edit the batch.dat file. Look for any row beginning with "Events" and change the LAST value.
Example (default value): The line #Events 500 1 0.200000 2.000000 10000 will generate 10000 events.

-- BruceThayre - 19 Jul 2006

Topic revision: r2 - 2006/07/19 - 20:45:48 - BruceThayre
 
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback