[Halld-physics] meeting this morning

[Richard Jones]

Blake,

Was unable to make the physics meeting this week (vacation) but read 
your slides -- some nice work.  Others may have responded to your 
question of, "Why is Prob(4gamma) = Prob(pi0)*Prob(eta) not flat?" 
during the talk, but anyway here is yet another (hopefully consistent) 
answer.

The product you formed of the two confidence levels is not flat, but 
roughly triangular-shaped, as your histogram shows.  The formal 
algebraic form of this function is

            / log(z/a^2)   : a^2 < z < a
   f(z)  =  | -log(z)      : a < z < 1
            \ 0            : otherwise

where z is the product of the two original 2gamma likelihoods that were 
flat on the chosen interval [a,1].  This function f(z) is not very 
useful, however.  It is better to sum the chi2 values and the degrees of 
freedom from the two original fits and then form a new confidence level 
out of the merged chi2 and dof.  That confidence level should be flat, 
or close to it.

for what it's worth,
-Richard Jones



Ryan Mitchell wrote:
> Hi All,
>
> Just a last minute reminder about this morning's physics working group  
> meeting.  Blake will present his work on the photoproduction of eta  
> pi0.  Slides and connection information are posted here:
>
> http://www.jlab.org/Hall-D/software/wiki/index.php/August_17%2C_2009_%2811AM%29%2C_Physics_Working_Group
>
> ESNet and EVO will both be available (but ESNet is preferred).
>
> Best,
> Ryan
>
>   
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3626 bytes
Desc: S/MIME Cryptographic Signature
Url : https://mailman.jlab.org/pipermail/halld-physics/attachments/20090818/c1b0d0a9/attachment.bin