Karen Andeen's Work, Problems leading up to my Prelim

Unfortunately, at the last minute here, I am having various problems. The main problem is that depending on how I make my plots in root, they turn out vastly different. For example, I have below two scatter plots of log10(K50) vs log10(S30). I expected them to be identical, however, they are not. The difference between them is simple. For the top plot I typed into root: 

h100->Draw("log10(K50[6]):log10(Fitid[0])","w"*(Sim_S30RC));

For the second plot, I typed this: 
 TH2F *K50_S30 = new
TH2F("K50_S30", "log10(K50) vs log10(S30)",10000,0.4,4,10000,-5.8,2);
K50_S30->SetMarkerColor(02);
h100->Draw("log10(K50[6]):log10(Fitid[0])>>K50_S30","w"*(Sim_S30RC))

The "w" is my weighting function, and Sim_S30RC is my S30, radius and cylinder size cuts. If you'll notice, the plots are identical until log10(S30) = 2.5. After that, the histogram for which I have defined bins and range and things cuts off entirely, while the top one keeps going. If the events are actually being cut out, there's a huge problem, but so far I've spent an entire day trying to figure out why it looks so terribly different with no success! If anyone understands this problem, let me know, I'd be happy to hear it.


Note that the reason I'm using 10000 binson each axis is that 1000, 100, or 10 just don't match at all. 100 bins is the only one that doesn't cut off on S30, but the distribution of events looks totally wrong. Here are examples of 1000, 100 and 10 respectively:


The scripts that I'm using are here:
Basic Loop
The Weighting Function
Radius Calculation
The Data Files (You may want to right click and save this one??? Or it's on the Madison machines at /data/kandeen/spase/data/sim_old/reco_Oct2005/mergedsim.root)


The next problem I'm having (which might be solved if I can solve the first problem, but I'm not sure of that) is that I have produced two possibilities for the log10(K50) vs log10(S30) plot and I'm not sure now which one's right. Again, I will let you see the differences for yourself. But no matter which plot is correct, I obviously have a lot of "stragglers" at low K50 but high S30, which is another problem. Additionally, if you look at the third plot (the one in red) it is immediately apparent that the same problem does not exist in the data. Thus there is some problem with the Monte Carlo. This problem concerns me greatly because this is the same Monte Carlo we've been using for a good long time and this problem did not seem to exist before. Thus, I've been making lots of plots to see if we can get to the root of it. I'll discuss those next. On the left below we have the plot from January, on the right is the one I made in February. Below is the experimental data.

Now, I've been making a bunch of plots to see what the problem could be (for results of some of these plots look at my January plots here). In summary, the only variable that has yet been found that has a very strong correlation to the low K50 at high S30 is Nchannel, and so I am keeping in mind an nchannel cut. However, I'm not sure where to make it, or whether this is a good idea. Additionally, due to the problem explained above, I now know that the numbers on the web page referenced just above the numbers are wrong in the chart. For more reasonable (but similarly untrustworthy) numbers and percentages of data that would be cut for every cut level on Nch, see my Early February page: here. I am currently exploring the possibility of other variables, such as track length and radius from the center of AMANDA (which I had neglected to check earlier). Those plots will be posted soon, along with recreations of all of the plots on the January page, likely with different numbers yet again! (Urgh!)