ARE WE READY FOR THE FIRST LONG RUNS?

Advance notice required for offsite farms? How much cpu time is available this week/next few weeks?

Which CPU's to use?

MIT, UVA, OU, ODU

Purposes of first long runs

-to compare data to simulation with high statistics

-to generate first acceptances

What type of data should we generate?

- CELEG with elastic scattering and resonances turned on MIT, UVA/2

- Elastic scattering from CELEG - OU, ODU

- Pi-zero from AO - UVA/2

- Background events (later?)

We need a local coordinator for remote institutions?

MIT - David Rowntree, Jianguo Zhao

OU - Allena Opper

ODU - Andi Klein

UVA - Cole Smith

FSU - Larry Dennis

How big should the file sizes be?

Does it matter? depends on frequency of crashes, internet copies, etc.)?

How many events do we need?

In ideal case:

W bins of 20 MeV, 0.9 GeV -> beam energy

Q-squared bins 0.2 -> 1.1*beam energy, 10 bins logarithmically spaced

Hadronic theta bins - 10 degree bins from 8 to 140 degrees

Hadronic phi bins 20 degree bins from -180 to 180

At 1.6 GeV, number of bins is 35(W) x 10(Q-squared) x 13(theta) x 18(phi) = 81,900 bins

81,900 bins x 1000 events/bin (3% statistics) x 3 (electron fraction) = 246 million events

246 million events @ 1 second/event = 2844 cpu days

2844 cpu days @ 20 dedicated cpu's = 4.7 months (per energy, magnetic field setting, and event type)

For elastic scattering, need fewer events because of e-p correlations, one W bin, limited theta:

At 1.6 GeV, number of bins is 1(W) x 10(Q-squared) x 8(theta) x 18(phi) = 1440 bins

4.7 months x 1440/81,900 = 49 cpu days => 12 days @ 4 dedicated cpu's.

If aim for 10% statistics instead of 3%, above numbers become 2 weeks and 1.2 days.

================> Proposal <===============

use 10% of distributed farm for elastic scattering from CELEG - 1.6 GeV, 60% field
use 30% of distributed farm for pi-zero from AO - 1.6 GeV, 60% field
use 60% of distributed farm for CELEG 1.6 GeV, 60% field
accumulate a repository of files starting ASAP
plan to run for 3-4 weeks, starting ASAP
start analyzing the simulated data after week 1 , modify plan in response

Need to coordinate analysis of simulated data

Can analysis be done at JLAB?

Which energy to start with?

1.6 GeV

- high fraction of elastics

- smallest number of events needed

- pi-zero fraction ok

- faster to simulate

- smaller event size

Known problems

TOF geometry - when will it be ready (and tested)?

Systematic errors seen by Kyungseon and Burin

Are there known crashes on particular platforms?

What about the MCX bank or other banks?

Append them later, after the long runs

Statistical accuracy issues

CELEG uses different seed for each run (check that this is still true)

AO uses different seed or same?