Radiative Corrections Using Deep

Contents and Links

1. Getting the code.
2. Building the code.
3. Running the code.
4. Input file.
5. Output during execution.
6. Output files.
7. Pitfalls.
8. Talk given at the CLAS meeting, June, 2004.
9. Description of the the inelasticity (v_cut).

Send comments, complaints, flames, etc. to gilfoyle@jlab.org

The measurements of the d(e,e' p)(n) reaction in the E5 running period are subject to radiative corrections. We are using a modified version of the program EXCLURAD written by Afanasev, et al to perform those calculations (Phys. Rev. D66, 074004, 2002). This code applies a more sophisticated method than the usual approach of Mo and Tsai or Schwinger and takes into account the exclusive nature of our measurements. We have not yet applied those corrections to our data, but we have revised EXCLURAD so that it can be applied to these data. The code was originally written for the p(e,e' pi+)X reaction and we have modified it for the d(e,e' p)(n) reaction. The response functions for the reaction are calculated with Deep, a code written by Wally van Orden and collaborators (Phys. Rev C66, 044003, 2002). Guidance for building and running the code are below.

A tarred, zipped version of the code is here (release date 3/26/04).

A more recent release is here (release date 5/18/04). It includes both proton (d(e,e'p)n) and neutron (d(e,e'n)p) versions of the E5 EXCLURAD and code for running batch jobs on the Richmond cluster.
Copy the most recent zipped, tar file to the the directory where you want to store the directory containing this version of exclurad. Change to that directory and unzip and untar the file.
```
cp exclurad_wvo.tar-04-may-18.gz top directory
cd top directory
gunzip exclurad_wvo-04-may-18.tar.gz
tar -xvf exclurad_wvo-04-may-18.tar
```
This will produce a new directory called exclurad that will have three subdirectories.
1. wvo_proton - contains the code for calculating the proton version (d(e,e'p)n).
2. wvo_neuron - code for the neutron version (d(e,e'n)p.
3. run - codes for using the Richmond cluster.
Now go into wvo_proton or wvo_neutron and build the program
```
cd exclurad/wvo_proton or cd exclurad/wvo_neutron
make
```
This step has been tested on RedHat 7.2 linux (gcc version 2.96 20000731 (Red Hat Linux 7.2 2.96-112.7.2)) at Richmond (pscm1.richmond.edu). The first release was tested on the same Richmond machine and at jlab (gcc version 2.96 20000731 (Red Hat Linux 7.2 2.96-108.1)) on url2.jlab.org and it worked. Jeff Lachniet at CMU was able to build and run the first release of the program on RedHat 9. He had to make a change in the file include.mk. He changed the environment variable OSNAME to OS_NAME and set OS_NAME to LinuxRH9. If you have any trouble, let me know at ggilfoyl@richmond.edu.
Once you have an executable, make any changes to 'input.dat' which is reasonably well commented and run it.
```
exclurad_deep_Linux
```

The input file is called 'input.dat' and looks like the following.


4       !  1: AO 2: maid98  3: maid2000 4: deep
0       !  0: Full, 1: Factorizable and Leading log
5.75    !  bmom - lepton momentum
0.0     !  tmom - momentum per nucleon
1       !  lepton - 1 electron, 2 muon
2       !  ivec - detected hadron (2) proton always
0.310   !  vcut - cut on v. (0.) if no cut, negative -- v
5
2.218       2.218       2.218       2.218       2.218
1.4         1.4         1.4         1.4         1.4
0.25        0.05        0.05        -0.5        -0.95
0.0         0.0         0.0         0.0         0.0

A description of the contents of each line is given below. The number in the left-hand column corresponds to the line of the input file above.

Line	Line content and explanation.
A.	4 ! 1: AO 2: maid98 3: maid2000 4: deep Picks the source of the response functions. Always 4 for d(e,e'p)n.
B.	0 ! 0: Full, 1: Factorizable and Leading log Chooses which level of approximation to make. Factorizable (1) is fast while the Full (0) calculation can take from one to several minutes per point.
C.	5.75 ! bmom - lepton momentum Sets the electron beam energy.
D.	0.0 ! tmom - momentum per nucleon Not used here.
E.	1 ! lepton - 1 electron, 2 muon Pick electron or muon beam. Always electron here.
F.	2 ! ivec - detected hadron (2) proton always Parameter used to handle situations where the angles definitions are reversed. Always 2 here.
G.	0.310 ! vcut - cut on v. (0.) if no cut, negative -- v This is used to set the inelasticity cut which determines how far to integrate to recover the lost flux due to radiation. Units are GeV. Equivalent to the deltaE parameter in more traditional Schwinger or Mo and Tsai calculations. A full description in in the appendix to CLAS Approved Analysis Out-of-Plane Measurements of the Structure Functions of the Deuteron or in the short paper here.
H.	5 Number of cases to run. Each case is defined by W, Q2, costheta and phi.
I.	2.218 2.218 2.218 2.218 2.218 Values of W in GeV for each of the cases. The value of W in exclurad is calculated assuming the mass of the target/struck object is a deuteron. Pick out quasi-elastic scattering by require Bjorken x=1 so x = 1 = Q²/(2M_nnu) where M_n is the mass of a nucleon. This restricts the value of nu (the energy loss) if you have already picked Q² and vice versa. Then W² = M_d² + 2M_dnu - Q² where M_d is the mass of the deuteron.
J.	1.4 1.4 1.4 1.4 1.4 Values of Q2 in GeV2 for each of the cases.
K.	0.25 0.05 0.05 -0.5 -0.95 Values of cos(theta_pq) for each of the cases.
L.	0.0 0.0 0.0 0.0 0.0 Values of phi_pq in degrees for each of the cases.

When you run exclurad, it dumps lots of information to the screen. As example is shown below.


 npoi= 5 data points
 Warning!! Very small LTp structure function:  8.26599695E-28
 tai:  1  1   0.9625E-02  0.0021      231  0
 tai:  2  1   0.5937      0.0066     1617  0
 tai:  3  1   0.5071E-01  0.0001      165  0
 tai:  4  1   0.5317      0.0065     1617  0
 tai:  5  1   0.7875E-02  0.0023      231  0
 tai:  1  2   0.6748E-01  0.0098     1089  0
 tai:  2  2   0.1628      0.0084      627  0
 tai:  3  2   0.1976      0.0071      759  0
 tai:  4  2   0.1462      0.0076      627  0
 tai:  5  2   0.5198E-01  0.0047     1155  0
 tai:  1  3   0.9625E-02  0.0021      231  0
 tai:  2  3   0.5937      0.0068     1617  0
 tai:  3  3   0.5071E-01  0.0001      165  0
 tai:  4  3   0.5316      0.0068     1617  0
 tai:  5  3   0.7875E-02  0.0023      231  0
 tai:  -0.2085E-14   0.2317E-02  231    0
 test :   0.922E-13 -0.632E-17 -0.146E+05
   2.218   1.400   0.967   0.250   0.000   0.811 -.15E+05   0.821

The first line lists the number of calculations requested. The second is a warning about extremely small response functions that are at the limit of precision. The lines that start with `tai:' show intermediate results of the full calculation. The important ones are in columns five and six out of the seven. Column five is the result of the test of an integral exclurad performs and should always be less than 0.01. Column six is the number iterations performed in calculating the same integral. The line labeled `test' list the results of some tests on the calculation. The very last line has the results of the calculation in the following order for the example above.


W       Q2      epsilon cos(theta_pq)  phi_pq  full result   test     leading log result
2.218   1.400   0.967   0.250          0.000   0.811       -.15E+05   0.821

This pattern is repeated for each of the calculations requested.

Output files.


radcor.dat - contains the primary results of the calculations.
all.dat - contains the monitoring information that was printed out during execution.
radtot.dat - actually appears to contain less information than radcor.dat.

Pitfalls . After doing many of these calculations we noticed that occasionally the calculation will 'fail'. The symptom is that the radiative correction for say one angle out of a complete angular distribution will be noticeably inconsistent with the other values in that angular distribution. This is obvious if one plots the RC surface as a function of theta_pq and phi_pq for a given kinematics (Q² and W). More details can be found in the presentation made to the Nuclear Physics Working Group at the CLAS Collaboration meeting in June, 2004. In response we have developed code that will enable the user to perform many parallel calculations on the Richmond cluster and visualize the results with the symbolic math package Mathematica which is available at JLab and many universities. Once the user has identified a problem lowering the convergence criteria and/or increasing the number of throws in the Monte Carlo integration algorithm usually fixes the problem. A more detailed description of the problem and the solution are available in the E5 electronic logbook here and here. The username and password are the usual ones.

Physics Department | Arts & Sciences | My Home Page