CONVERTING A NEUTRINO-IRON EXPERIMENT TO LEPTON-PROTON 29-Sep-97 Bill Seligman [The following is an internal CCFR collaboration memo written to discuss the conversion of the CCFR F2 neutrino data to a "proton" form. Some minor revisions have been made to make the file suitable for public distribution.] A couple of months ago, Dieter Haidt at CERN asked Mike Shaevitz if our CCFR neutrino-iron F2 data could be converted into the electron-proton equivalent. The idea was to compare the converted data against NMC's proton data in the hopes of resolving the CCFR-NMC discrepancy. Mike asked me to perform this conversion, which I performed using the procedure described in the Appendix. You can find the converted CCFR data in http://www.nevis.columbia.edu/~seligman/ccfr/results/ccfrp/ccfrp-for-plots-only.txt (all the files mentioned in the rest of this memo can be found in this same directory). The name includes "for-plots-only" because the systematic errors on the data have all been added in quadrature, which makes the numbers useless for a phenomenological analysis. I plotted the CCFR "lepton-proton" data together with the NMC proton data; the results are in ccfrp-vs-nmc-1.eps and ccfrp-vs-nmc-2.eps. As you can see, the CCFR-NMC discrepancy is the same in the proton data as it is in the deuterium data: we are higher than NMC up until x~.1. Mike did not have the opportunity to review these plots, so I held on to them until he had more time. Now let's fast-foward to about two weeks ago: Alan Caldwell wrote a messsage to Mike, asking him for the CCFR data converted to lepton-proton so that it could be compared to HERA data. Mike forwarded the message to me, and I in turn sent the the converted data to Alan. I asked him to use the data only for plots, but to keep the file to himself. Alan then presented a plot (see ccfrp-vs-hera.ps) at a HERA review meeting. As you can see on his plot, the converted CCFR data appear to agree better with the ZEUS data than the NMC or E665 data do. Before we interpret this as proof that the ZEUS data is correct :-), note that the agreement/disagreement of CCFR/NMC vs. ZEUS is mostly visual. The impression is based on the lowest-Q2 point of the ZEUS data, which in fact is in agreement (within errors) with both the CCFR and NMC data. Alan's plot has sparked renewed interest in the CCFR data, but the requests I've received are for the "proton" data and not for our original neutrino data. This raises two issues: 1) What are the potential errors in my neutrino-iron -> lepton-proton conversion? I address this issue in the Appendix. 2) Should we make the converted data public? The latter issue is one of potential mis-use of the data. The structure-function analysis is complex and its results are spread out over many files. It's much easier to think of ccfrp-for-plots-only.txt as our "final result", hit the download button on your web browser, and throw that one file into your plots and your analysis. Already someone at Durham has asked if we can put the CCFR "proton" data on the HEPDATA database, and Robin Devenish wants to include the "proton" data in his review of deep-ielastic scattering. My worry is that the phenomenologists will treat the contents of ccfr-for-plots-only.txt as our measurement. But, of course, these "eP" data are _not_ our measurement. Aside from the simplification of adding our systematic errors in quadrature, the corrections to the CCFR data are partially correlated with the NMC and ZEUS data with which it is being compared. On the other hand, as you will note from the Appendix, anyone can do what I did. The conversions are not complicated. In fact, in a way these corrections are what (we hope) the phenomenologists do with our data when they perform the global fits. Perhaps the name of the file and the warning at the top will be sufficient. I look forward to your comments on both these issues. ------------------------------------------------------------------------- APPENDIX To understand the corrections that I made to the CCFR data to convert them to lepton-proton, Mike Shaevitz and Frank Sciulli asked me to prepare the plot in ccfrp-corrections.eps. This plot displays the four corrections I make to the CCFR nu-Fe data in the x=0.0175 bin. This is the x-bin that Alan Caldwell included in his plot. The four corrections are: 1) F2(l) / F2(nu) -- the correction from a neutrino probe to a charged lepton probe. This is the standard "5/18ths rule": F2(l)/F2(nu) = (5/18) * (1 - 3/5*(s+sbar-c-cbar)/(q+qbar)) I used the same correction in my thesis and the PRL to convert the NMC data to compare with the CCFR data. The parton distributions that I used for s, sbar, c, cbar, q, and qbar come from CTEQ4D, which parameterizes the parton distributions in the DIS scheme. This is important because the 5/18ths rule is valid to all orders in QCD only in the DIS scheme. To calculate the error shown in the plot, I re-computed this correction using GRV94. As you can see, although this correction is the largest of the four, the error in the correction is negligible. 2) F2(D) / F2(Ca) -- the heavy-target correction. Once again, this is the same correction as I used my thesis. This comes from a parameterization of NMC and E665 Ca/D data, and SLAC Fe/D data. You can see a plot of the data and my parameterization at http://www.columbia.edu/~seligman/ccfr/thesis/eps/fig8-5.eps (or look at Figure 8.5 on page 251 of my thesis). The error associated with this correction comes from parameterizing the errors on the available data, adding the parameterizations in quadrature, and including a Chi**2 expansion in the low-x region where the NMC and E665 data disagree significantly. (A "Chi**2 expension" means to increase the error by sqrt(Chi**2/(number of measurements - 1)).) This correction is the most controversial of the four. The E665 Ca/D data disagree with the NMC Ca/D data. If you use the E665 Ca/D correction by itself, the CCFR-NMC discrepancy goes away. However, the errors on the E665 data are larger than the NMC errors, which is why my parameterization tends to agree with NMC. Note that this correction is not Q2-dependent. The correction and error on the plot are exactly the same data and error at all Q2. 3) F2(N+P) / F2(D) -- the bound-nucleon correction. This corrects the structure functions for a bound neutron+proton state to the SF for a free neutron and a free proton. The data for this correction was sent to me by Arie Bodek, and I parameterized this correction and its error from that data. As you can see, this correction is small. Note that this correction is not Q2-dependent. The correction and error on the plot are exactly the same data and error at all Q2. 4) F2(P) / F2(N+P) -- the isoscalar-to-proton correction. Once again, I used CTEQ4D to compute this correction in the DIS scheme: F2(P) = (4*U(P) + D(P) + S(P) + 4*C(P))/9 F2(N) = (4*D(P) + U(P) + S(P) + 4*C(P))/9 F2(P) / F2(N+P) = 2/(1+ F2(N)/F2(P)) ...where U(P) = u + ubar in the proton, D(P) = d + dbar in the proton, and so on. If you thought this correction should be about 1/3rd, remember that we're looking at x=0.0175, where most of the quarks are sea, so F2(P) is about the same as F2(N). Again, the error in this correction comes from taking the difference from using the CTEQ4D and GRV94 distributions. Note: Arie Bodek requested that I use a revised parameterization of D/U from D0 W-asymmetry data to compute this correction. I did not do this, because a) his correction was expressed in terms of changes to the CTEQ3 and MRSA distributions, and I did not use those PDFs to compute my corrections; b) in preliminary tests I did using CTEQ3, the new values did not change my corrections significantly. All of the errors in the above corrections are added in quadrature to the statistical and systematic errors on the CCFR data, and this total error is what appears in ccfrp-for-plots-only.txt.