CHICOS Maximum Likelihood Code

There are two versions of the maximum likelihood code:

• CTShower v3.01
  • Uses "primitive" probability code.
  • Generally reliable.
  • Now available as an option from the public chicos page.
• CTShower v3.02
  • Uses "integral" probability code.
  • Sometimes hangs - process needs to be killed manually.
  • Only available on the shower test page.

Write-up of the Max likelihood code.

I looked at a few sample showers to see how the max likelihood code behaved compared to v. 2.42.

• 12-16-06
  • V. 2.42 all hits, iteration 1, iteration 2, final.
  • V. 3.01 all hits, iteration 1, iteration 2, final.
  • V. 3.02 iteration 2, final.
• 12-12-06
  • V. 2.42 all hits, iteration 1, iteration 2, final.
  • V. 3.01 Didn't work.
  • V. 3.02 Didn't work.
• 11-29-06
  • V. 2.42 all hits, iteration 1, final.
  • V. 3.01 Didn't work.
  • V. 3.02 Didn't work.
• 11-03-06
  • V. 2.42 all hits, iteration 1, final.
  • V. 3.01 all hits, iteration 1, final.
  • V. 3.02 iteration 1.
• 10-27-06
  • V. 2.42 all hits, iteration 1, final.
  • V. 3.01 all hits, iteration 1, final.
  • V. 3.02 final.

The -log(likelihood) parameter does not seem to correlate very well with the goodness of fit. It sometimes goes to 0 on marginal fits. (See first example.)