Agenda (12:00 CST, 12 April 2008)
1) Review Pipeline: what needs to be done?
- Initial Conditions Generator (done)
- Gadget2 (include lightcone hack?)
- External lightcone code?
- Group finder (FOF)
- merger tree code?
- .. other codes for Mock generation, etc ..
2) Status and timeline of running simulations
3) FOF Groupfinder: plans for now and future
- Parallel FOF: scaling problem (memory challenged)
- Serial FOF: how well is it working? no merger trees?
- if necessary: how important is merger tree output from first few sims?
- any help welcome for tweaks or tests!
- binary halo catalog format work?
Summary of Discussion
Participants: Andreas, Roman, Frank, Risa, Michael, Cameron
Pipeline reviewed
- lightcone calculation included in Gadget2 (Michael) and is being tested on Carmen (Roman)
- external lightcone generator from snapshots will be put on wiki (Roman)
- groupfinder issues, see below
- merger tree code: less important right now. save 1 to several "full" (100+ snapshots) realizations for merger trees.
- mock generation: several codes available, we should make them available to the collaboration via the wiki.
Group Finder
- serial FoF code works on Carmen for mocks and mass functions. too time consuming for merger tree output (100 snapshots per realization)
- serial FoF code may not work on available machines for Consuelo (Oriana and Esmeralda possible, but not yet tested)
machines with highmem capabilities? (Cameron & Andreas: check teragrid; Frank checking MPA)
- parallel FoF code needs to be tested: both in parallel and also in serial on high-mem machine (like Altix)
- another parallel FoF code (Frank is asking someone at Max Planck)
- SO algorithm might be included in Gadget2, won't happen until late summer (Michael)
- no other parallel SO codes seem to exist
- serial SO codes? (yes from Michael)
- halo output format: standardize group catalog format?
- outputs should include halo list (mass, pos, vel, etc. of haloes without member particles)
Current Simulation Status
- Roman has run 13 Carmen at NYU. Will start Oriana soon. Snapshots for merger trees stored at PSC.
- Michael / Risa have run 3 simulations, finishing up a Consuelo run within a week.
- Cameron / Andreas are starting Esmeralda (on bigben and soon Ranger). Plan is to run 1 of each luminosity bin and then more Esmeralda.
- Frank: working on access to resources at MPA
Plans and Timelines
- focus simulations on the standard 10 outputs and generating 1st generation mocks (possible SDSS use). Sooner is better.
save the 100 snapshots for AT LEAST one realization now, and AT LEAST TEN total (for merger trees).
- even 10 realizations for merger trees in each luminosity bin is better than currently available (Risa, Frank agreed)
- eventually, we want to publicly provide halo catalogs and multiple galaxy mock generating codes