Kaon_LT Experiment

		Mar 04/22 Group Meeting Notes

                -----------------------------

Participants: Garth, Stephen, Vijay, Ali, Nathan, Love

Vijay:

- good discussion on how to deal with pion leakthrough in (e,e'K+) analysis

Three methods come to mind:

1. Histogram subtraction

   i.e. make a clean pion sample, normalize the pi+n MM peak to the leakthrough

   to determine the scale factor, and subtract the clean pion sample from the

   K+ sample to get a clean(er) K+ sample

   Advantages:

   - accurately models the pi+Delta0 shape underneath the Lambda and Sigma

     peaks

   - once the MM scale factor is determined, in principle this can also be used

     to clean up focal plane and other observables, for comparison with MC.

   * Cautions/Suggestions for Future Work*

   - this method only works if the pion leakthrough fraction is same for pi+n

     and pi+Delta0 MM regions

   * Need to investigate the stability of the pi+n/pi+Delta0 ratio

     vs. CoinTime, RF cuts

2. Polynomial subtraction

   i.e. fit a polynomial to the clean pion MM sample, and use that polynomial

   (with adjustment) to get a clean K+ sample

   Advantages:

   - Vijay shows that this method seems to work for at least one setting

   *Cautions/Suggestions for Future Work*

   - need to be sure that the Lambda, Sigma radiative tails are not

     over-subtracted, as they are an important part of the reaction yield

   - if this method is used, the best way would be to get the Lambda, Sigma

     peak shapes from SIMC and use that, in combination with the polynomial, in

     the fit

3. Clean K+ cuts

   i.e. make a clean (e,e'K+) spectrum by applying tight cuts to eliminate most

   if not all pion leakthrough

   *Cautions/Suggestions for Future Work*

   - almost certainly, this gives a lower K+Lambda yield than methods 1,2

   - the lost K+Lambda events would have to be corrected for via a "cut

     efficiency".  It is not immediately obvious on how this efficiency factor

     would be determined, maybe by comparing with methods 1,2, but in that

     case, why not just use them for the analysis?

   - however, a quick comparison of the lower yield from method 3 compared to

     methods 1,2 could be useful for systematic error determination, for

     example by setting an upper bound on how large the cut-event-loss effect

   is

Vijay should keep track of this discussion, for use in his further studies