Re: [BLAST_ANAWARE] new monte carlo formalism

From: Chi Zhang (zhangchi@MIT.EDU)
Date: Fri Sep 03 2004 - 13:10:34 EDT


Hi Simon,

The home-brewed variation of spline worked for many of the 41 f's, however
it is still not bullet proof. I tuned the code so it is well
controlled for the few f's that are largest in sized. But when I
later carefully check every 41 functions at every slice of theta_cms, I
can still clearly see riples in some of them.

We received data from Dr. Deuteron later for a denser grid and natural
spline works just as well as "cbrt" and is quite bit faster. So at this
moment, we are using natural spline for the steepest varying two
dimensions and linear over the rest 3. I believe the "ridges and dips"
in the omega-theta_e plane is still a tough thing to deal with. spline is
not safe and linear is not good.

Michael mentioned using our beam energy to back out the f's into functions
over omega, Q2, and theta_cms. I wonder if that would seperate the
electron and hadron side as expected and maybe the resulting shapes would
be easier to handle with the explicity "theta_e" kicked out. I have not
had time to do it yet.

Also about the use of computer resource. It is possible to reduce amount
of system recource required by a DGen/EEP session. Currently, DGen sets
up lookup table for all 6 helicity-target spin states combo, and
generate events that can be analyzed using almost exactly same code
used for actual data. One can simulate events in only one helicity-target
state, reduce the amount of memory need for lookup tables by a factor of
6. The saved memory can be used for duplicated tables for the other
degrees of freedoms. But this way, one has to run MC for each spin state
and write new codes to analyze them.

Chi

On Fri, 3 Sep 2004, Simon Sirca wrote:

> Here goes the atatchment, sorry.
>
> Simon
>
> --
> Simon Sirca
> Dept of Physics, University of Ljubljana Tel: +386 1 4766-574
> Jadranska 19 Fax: +386 1 2517-281
> 1000 Ljubljana, Slovenia
>



This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:31 EST