- Sebastian

We have a complete database of all "non-crashed" electron beam runs taken during EG1 (since mid-September) correlated with additional information like target polarization etc.

1. Enter actual (best-guess) beam polarization data from Tony's analysis of Møller runs (after all corrections are known)
2. Finish re-analysis of target polarization
3. Check hand-written logbooks for additional important information, especially on trigger configurations (and changes), unusual detector parameters and status, etc.
4. Add information on raster settings if they can be obtained, and possibly target magnetic field direction (and whether it's on or off)

All "chefs" and other analysis workers are supposed to add information to the offline database/logbook on each run they work with.

Beam polarization history (Workers: Tony Forest with help from Arne and Brian)

This job is mostly finished by now; the results need to be corrected (very small correction) for proper analyzing power (from MC simulation done at FIU) and entered into the database.

Target polarization history (Workers: Raffaela Devita and Renee Hutchins)

This requires an offline analysis of all polarization measurements done, again yielding a database of target polarization (including sign) vs. time. Ultimately, this involves analysis of optimum background and pulse shape, and in the case of the deuteron, a comparison of the ratio method and the direct NMR area determination. For additional checks one needs to compare with the "online" values as written down and the runsummaries#.txt files. Once the "target computer times" have been reconciled/aligned with the "DAQ time", we can again extract the average target polarization for each run. Note that a more precise estimate of the target polarization would require to use all available information on the raster size, the charge deposited on each target and the relevant decay times to extract the polarization of the fraction of the target actually seen by the beam.

This job is done to a large part; the biggest part missing are the deuteron polarization numbers. We will probably never get a very accurate target polarization from the direct measurements, especially not for the runs where we kept burning a small raster pattern into the NH3 target. Our best bet will be to attempt an absolute determination for the early runs and then use known elastic asymmetries to get info on the remaining runs.

Lists of runs

Runs can be moved from one category to another later on, as information is added to the data base. For instance, runs during which the charge asymmetry of the beam exceeded 10^-3 might be declared junk runs.

All detector elements (DC's, TOF, Cerenkov, EC/LAC) needed to get their calibration constants (gains, offsets, start times,…) updated. These updated calibration constants were moved into the parameter maps. We might still need to check that target field and type, polarization, and torus polarity, are also entered correctly in those maps.

The following sequence (with preliminary names attached) was created by Stepan Stepanyan in preparation for the calibration. It may be out of date by now.

Code/algorithm improvements/modifications

Each event in the run is associated with -

· Strong magnetic field near the production vertex

· Beam helicity

· Beam position on the target (raster)

all of which require modification of analyses algorithm. (We can't really include raster information yet; the strong magnetic target field has been implemented but can also be circumvented by using "pseudo vertex" information).

Code modifications:

· Modification of fitting procedure

· Reconstruction of the vertex using raster information (NOT DONE)

· Reconstruction of three-vector at the vertex

· Store helicity information in the high DST banks

· New run control parameters

· Design the shell of the code to include all DST banks

· Monitoring histograms and database

· Scripts for farm processing and monitoring

Pre-calibration preparations

Set the initial calibration constants -

· Time-of-flight counters (Duke U. group, Volker Burkert, Hovanes Egiyan)

· DC timing (done according to Liming Qin)

· RF TDC

· Define set of the runs to be processed for calibration (see previous Section " Lists of runs")

· Get the "tagged" version of the analyses code (RECSIS) with first order modifications;

· Get the monitoring scripts/code (and the data bases?) ready to test - possible connection with the eg1 off-line logbook.

Proposed Ending time - February 12.

Processing for calibration (10%)

· Group the selected runs (by the beam energy and the target);

· Assign people (grad.stud.) to each group;

· Process all selected files using tagged code and scripts;

· Monitoring processing the data;

· Store all necessary BOS banks for the calibration;

Proposed Starting time - February 12.

Ending time - February 23.

Calibration.

Assign people to the following calibration tasks -

· DC - time, geometry/field (Rob Feuerbach), time-to-distance function and parameters (Junho Yun - only job done so far), vertex reconstruction (Franz Klein, Stepan Stepanyan, Tony Forest) "Roads" for all 5 different torus settings used during EG1 (Franz Klein - seems o.k.);

· SC - time calibration of TOF strips, determination of RF offsets, changes due to interchanged cables, time walk corrections energy losses (Duke/W&M/UVa/Elton Smith/Rustam Niyazov)

· CC - efficiency (Alex Vlassov), timing and amplitude;

· EC - energy (Cole Smith, Stepan Stepanyan), timing (Mehmet Bektasoglu, ODU)

· LAC - time, energy (Genova/INFN group);

· PID (Stepan Stepanyan, others);

· Helicity monitors (Tony Forest, Alex Skabelin, Raffaella Devita);

· Beam position (rastering) and current (Tony Forest and Sebastian Kuhn);

Proposed Starting time - February 12.

Ending time - March 21.

Re-processing of calibrated runs.

Experts should check the quality of the calibration.

· Detector specific characteristics;

· PID;

· Physics distributions;

· Specific analyses correlated to the beam and the target polarizations;

If requires, new processing for additional calibration should be done.

Proposed Starting time - March 23.

Ending time - March 31.

Under this headings fall all analysis tasks related to accumulate statistics for all event types that we are interested in.

Cooking - EG1 Pass 1

Cooking entails a first pass analysis of all production runs using a standard package like A1 or RECSIS. The goal is to extract from each run all good events with electron tracks, including pattern recognition, matching of tracks with hits in the TOF, Cerenkov and EC/LAC detectors, and identifying electron tracks and particle ID for all other tracks. The resulting banks can be used for either a second pass analysis (if there are major improvements in the analysis code and parameters) or, more to the point, for final analysis of publishable data.

Use the same scheme as for calibration -

· Group the selected runs (by the beam energy and the target) into 3 separate groups

· Assign 2 people (Renee and Alex, Junho and Angela, Mehmet and Raffaella) to each group;

· Store BOS banks (PART, EVNT and other high-level banks, plus some fraction of raw data).

· Bookkeeping, monitoring (look at histograms, run "tester");

ACTUAL Starting time - Beginning of May.

List of DST banks -

HEAD 0 8 1 8 I

TGBI 0 1 1 1 B32

CALL 0 3 3 5 B16

EVNT 0 18 5 90 I,F,F,I,F,F,F,F

DCPB 0 13 4 52 I,F,F,F,F,F,F,F

ECPB 0 16 4 64 I,F,F,F,F,F,F,F

SCPB 0 6 4 24 I,F,F,F,F,I

CCPB 0 6 1 6 I,F,F,F,F,I

HEVT 0 16 1 16 I,I,I,I,I,I,F,F

TBID 1 34 5 170 I,I,F,F,I,I,F,F

PART 1 13 5 65 I,F,F,F,F,F,F,F

and all scaler banks.

Data volume will be - ~ 1 Kb/event (with e-).

In the current "cooked" files in addition -

CC 3 3 1 2 B16

DC0 1 2 56 56 B16

EC 1 3 2 3 B16

EC1 1 5 5 13 B16

SC 4 5 1 3 B16

TRKS 1 10 1 10 I,F,F,F,F,F,I,I

TDPL 5 8 10 80 I,F,F,F,F,F,F,F

HBTR 0 9 5 45 F,F,F,F,F,F,F,F

TBTR 0 10 4 40 F,F,F,F,F,F,F,F

HBER 0 23 5 115 F,F,F,F,F,F,F,F

TBER 0 23 4 92 F,F,F,F,F,F,F,F

Data volume will be - ~ 6.5 Kb/event (with e-).

For calibration data volume will be - ~ 8.5 Kb/event (with e-).

Extract polarization information (Workers: Tony Forest, Alex Skabelin, Raffaella Devita)

We need to set up a consistent scheme to correlate beam events (scattered electrons) and their helicity information with the scaler events that record the integrated current and other parameters every second, also keyed with helicity. We need to check that all helicity signals are consistent and use cuts to exclude helicity pairs where this correlation is "confused", or where any relevant beam parameters (intensity, position,…) fluctuate to strongly, or where unusual backgrounds are present. Ultimately, we need a mask to extract only those events that pass these tests and sort them properly according to beam helicity. It is important to study whether any of these cuts might introduce false asymmetries (and vice versa, which cuts we need to minimize existing false asymmetries). We need to collect the charge accumulated for each beam helicity in a run, excluding scaler events where data where not taken (no beam) or didn't pass our cuts at all. One problem is that we cannot trust the "fast" nA-BPM charge measurements, so we have to first develop a reliable normalization scheme. The best way seems to be to use the Faraday Cup information (maybe cross-check with Beam scintillation counter information). We need to cross-check our normalization with guaranteed "unpolarized" events (e.g., 2-proton emission or W < 0.8 GeV events). We also need to very carefully analyze deadtime/Poisson effects. A final cross check will be the comparison between opposite target polarizations.

Optimize analysis code

Before starting a Pass 2 reanalysis of EG1 (most likely only after and in conjunction with EG2000), one would want to improve the analysis code. Possible areas of improvement are:

Cuts

In addition to the helicity-dependent cuts (see above), we also need to impose additional cuts. Most importantly, we will use a fiducial acceptance cut on the reconstructed "pseudo-vertex" to avoid regions where CLAS has a large (and hard to pin down) inefficiency (e.g., center strip in the Cerenkovs).

We need to study the polarization dependence as well as efficiencies of all cuts. Some cuts are "predetermined", especially by the trigger thresholds and by the cuts used in the particle

ID (SEB) code. Again, we need to gather as much information on the efficiency of these cuts.

Sorting

Events that pass all cuts need to be sorted into histograms or ntuples. For the inclusive data, we need for each run a 2D histogram of Q2 vs. W for each helicity separately, normalized to beam current integral. The binning must be determined by considerations such as statistics, resolution and Physics interest. We need to study deadtime corrections Poisson statistics and truncation.

We need as accurate as possible simulations of the spectra for both protons and deuterons alone as well as for targets with different nuclear species (Nitrogen or Carbon, Helium, Aluminum,…). This information is required to extract the dilution factor (see below) and/or the acceptance of CLAS as well as its resolution.

One potential problem is that the simulated tracking efficiency may be different from the actual one, especially dependent on the general direction of the track (whether it goes through noisy regions of drift chambers etc.) and on beam intensity. One might try "peppering" the simulated evens with real noise events to check these effects.

The dilution factor can be gotten by making a 3-parameter fit of the simulated H or D data plus the measured Empty and Carbon run data to the observed NH3/ND3 spectra. This fit should get the relative normalization for the Empty and Carbon data and will be mostly constrained in the sub-elastic region (where only non-hydrogen data contribute, e.g. W<0.8 GeV) and the elastic region (where the shape of the peak from hydrogen should be very well simulated). The form of the fit would be

The dilution factor could then be determined as

Extract proton/deuteron data

There are two possible approaches to extract Physics data from the inclusive data:

One takes the raw asymmetry and divides it by the dilution factor determined as described above. In the (quasi-)elastic region, one can use the well(-enough)-known nucleon form factors to predict the asymmetry. Comparing with the result from the data, one can get a reliable number for the product of beam polarization and target polarization (cross check with expected results from Møller and NMR measurements). After normalizing with this product, one is left with the radiated asymmetry A||. For final results, one has to radiatively correct this asymmetry (see below) and then extract quantities like A1 (assuming we can either determine A2 from different energy data or use a plausible ansatz for it). Note that A2 is limited by

|A2| ² [R^.(A1+1)/2]^1/2

where R is the ratio of longitudinal to transverse cross section (which must ALSO be known to extract A1 from A||). On the other hand, one can obtain g1 from A|| by using an existing fit to the world data on W1 and W2/R in the resonance region.

One takes the acceptance from the GSIM simulation and corrects (i.e., divides) the measured countrate difference N⁺ - N^- (for opposite beam helicities) with it. One then integrates over the well-separated (?!?) (quasi-)elastic peak (using a fiducial range in W and Q2) and normalizes all of the data with the result. This should "automatically" take out the product of hydrogen density times beam intensity times beam polarization times target polarization, since this product enters both the (quasi-)elastic peak and all the other data in the same way. Any overall constant error in the acceptance also will be cancelled. One can calculate the theoretical cross section difference in the (quasi-)elastic peak (integrated over the same kinematic range), and by multiplying the normalized data with this number one obtains the cross section difference for the whole measured range.

From the cross section difference, it is (more) straightforward to extract the spin structure function g1 which is less "contaminated" by our lack of knowledge about A2 or g2. On the other hand, by dividing with a fit to existing absolute cross section data, one can obtain A|| again.

Both methods have their advantages and disadvantages, and both should probably be pursued in parallel. The "asymmetries" method may be best suited to extract asymmetries for specific resonances, while the "cross section difference" method might be a quicker route to integrals (of the GDH/Ellis-Jaffe/Bjorken type).

Radiative, resolution and nuclear corrections

The final step to get Physics results requires 3 corrections (that I can think of right now) for the inclusive data, for either of the two analysis methods outlined above.

1. We need to correct for the (small) contamination from the polarized proton in 15N and possibly for residual 14N contributions and/or H contributions in the ND3 case. This correction should be small (even if 15N were 100% polarized, which it isn't - its more like in the 10% ballpark), the contribution would be only 1/9 of the H signal (since there is only 1 bound proton for every 3 free ones, and according to the shell model it's polarization is only 1/3 carried by its spin and 2/3 by its orbital angular momentum). This correction can be done with a simple Fermi-gas type model of 15N; the other contributions require some more work (if we think they could be significant).
2. We need to account for the finite resolution of CLAS. This can be treated together with 3).
3. We need to correct for radiative effects. For this purpose, we need semi-reliable models for cross sections and asymmetries, including adjustable parameters to fit our data. These models can be used to predict both Born cross sections for H/D(e,e') and, after running them through a radiative code (I propose RCSLACPOL which we are maintaining here at ODU) and smearing them (using the resolution parametrization from GSIM, see above), be compared to our actual data. There is a standard method how to use the results to extract radiative corrections in terms of a additive and a multiplicative term (see upcoming radiative corrections workshop where Frank Wesselmann will explain this method). I am presently working on setting this machinery up for EG1. We will also need a reasonable target model for external radiative corrections (Raffaella will look into that).

The end result of these corrections will be to extract either A||(Born) or g1(Born).

Final results

Resonance amplitudes/asymmetries, integrals, semi-exclusive channels, …

Write theses, publications, go to beautiful, exotic places to give talks,…