Presentazione di PowerPoint

1
Cosmology Astrophysical Software
Technology (Theory, Simulations and VO-Data
Exploration)
Ugo Becciani, V. Antonuccio, A. Bonanno, A.
Costa, M. Comparato INAF - Osservatorio
Astrofisico di Catania
2
Tech 2004-2007 Main activities
High Performance Computing MPP-HPC systems _at_
INAF-OACT HPC Software development Virtual
Observatory and Data Exploration Visualisation
tool Theoretical Virtual Observatory
(ITVO) Knowledge Transfer GRID computing HPC on
GRID systems Knowledge Transfer
3
High Performance Computing
Project
Computational Astrophysics and High Education
Period 2002- 2004 CoPIs V. Antonuccio U.
Becciani
Funds
170 KEuro 140 KEuros HPC systems, 20 KEuros
contracts
Goals

• Science Studies on LSS of the Universe
• Technology MPP System, HPC codes and
  Visualisation
• Dissemination and High Education ( 2 schools
  for young researchers)

4
High Performance Computing
First MPP 24 Power 3 IBM Processors 48 Gbyte
RAM C, C, Fortran 90 Paradigms OpenMP, MPI,
LAPI Period 2000-2005
Second MPP System 8 Power 4 IBM Processors 16
Gbytes RAM Period 2002-2006
5
High Performance Computing

• Software Development
• FLY a tree N-Body code for cosmological
  simulations (U. Becciani, V. Antonuccio, M.
  Comparato)
• MARA parallelization a parallel structure for
  light curve analysis (U. Becciani, A.Lanza, M.
  Rodonò)
• Astrocomp (U. Becciani, A. Costa)

6
High Performance Computing
FLY Project Parallel tree N-Body code for LSSU
simulations. FLY is based on the Barnes- Hut
alghorithm and it is developed using Fortran 90
with the one-side communication paradigm
(MPI-2) each processor can access remote data
with GET, PUT and ACCUMULATE operations. The code
runs simulations and requires 250 MB Ram for 1
million particles. GOAL maximizing the number of
simulated particles without data replica and
without loss of overall performance Free
software downloaded by more than 60
reserachers. FLY is included in the software
library of Comp. Phys. Comm.
7
High Performance Computing
FLY challenges in the near future A NEW version
of FLY for the GRID paradigm is in progress and
will be ready in the next few months A New
approach will be applied for data distribution
and force computation maximization of local load
and only asynchronous SEND-RECEIVE messages will
be applied Challange Using the new GRID
infrastructure we will run a cosmological
simulation with more than 1 Billion particle one
of the largest simulation ever run having the
highest accuracy ever achieved
8
Virtual Observatory and Data Exploration
Project
VOTECH (EuroVO) 2005- 2008 PI A. Lawrence (IfoA
Edin. UK), Local PI U. Becciani
Funds
105 KEuro 90 KEuros contracts
Main Goals

• To assess new technologies and study the
  feasibility of their incorporation in EUROVO
• To create designs of new infrastructure
  components based on those new technologies
• To create designs of science user tools and
  datamining services

9
Virtual Observatory and Data Exploration
Virtual Observatory a great challenge The
concept of the Virtual Observatory (VObs) is that
all the world's data should feel like it sits on
the astronomer's desk top, made available through
a standard interface. Internationally this is set
to transform and re-structure the way astronomy
is done. EUROVO is a specifically European
implementation of this idea, and will produce a
world leading infrastructure providing a unified
virtual data resource and the ability to peform
complex data discovery and manipulation tasks
across the whole range of astronomy. The
International Virtual Observatory Alliance (IVOA)
coordinates for the development and deployment of
the tools, systems and organizational structures
necessary to enable the international utilization
of astronomical archives as an integrated and
interoperating virtual observatory."

• INAF - OACT Contribution
• Data Exploration and Visualisation (DS6 VOTECH
  Project IVOA Applications Working Group)
• Theoretical Virtual Observatory (DS4 VOTECH
  Project Int. Group IVOA, ITVO)

10
VisIVO (Interoperable Data Exploration tool for
VO data)
VisIVO is a data exploration tool developed in
collaboration between INAF (Catania Astrophysical
Observatory) and CINECA (the largest Italian
academic supercomputing center) with the specific
object of supporting visualisation and data
exploration of astrophysical data in the VO
framework

• Free C Software, Cross Platform (Windows and
  Linux)
• N-Dim Visualisation of Astrophysical data (LUT
  and GLYPHS gt scalar properties of each element)
• Data Analysis tool Point splatter, Power
  Spectrum, Correlation Function, Minkowsky
  Functionals
• Support for large data display (LOD)

11
Virtual Observatory and Data Exploration

• VO compliance and interoperability (PLASTIC
  with the contribution of INAF OACatania,
  Aladin, TopCat, etc.)
• VizieR (SDSS) data download, cone search
• Remote file load
• Points, Volumes and Vectors Visualisation
• Data Cluster extraction

12
Virtual Observatory and Data Exploration
VisIVO main challenges SDSS ? VisIVO will be
adopted for Data Exploration (Coll. Bob Nichol
Univ. Portsmouth) LOFAR ? VisIVO will be adopted
for Visualization VisIMED ? Technology Transfer
approved project (300 KEuros) Virtual
Observatory ? An important tool for data
exploration support for Huge data analysis (no
limits for data dimension). Interoperability ?
Use of Web Services, SNAP, data exploration and
selection (TVO)
13
Grid Computing
Projects
DRACO 2002- 2004 (PI F. Pasian, Local PI U.
Becciani) TriGridVL 2005 - 2008 (PI R. Barbera,
Local PI U. Becciani) PI2S2 2005 - 2008 (PI R.
Barbera, Local PI U. Becciani)
Funds
62 KEuros (Draco) 203 KEuros (Trigrid) 775
KEuro (PI2S2) ? 1,040 KEuros
Main Goals

• Infrastructure the larger HPC- EGEE compliant
  Grid for research
• Research support to approved research fields to
  solve complex problems (grid porting of
  applications)
• Knowledge Transfer

14
GRID and more...... we are building the future
The first EGEE National GRID HPC - oriented

• GRID sites
• INFN Catania
• INAF Catania
• University of Catania
• University of Palermo ( INAF Palermo)
• University of Messina

50 IBM BladeCenter H IBM LS21 blades 2,500
cores AMD Opteron 2218 rev. F 5 TByte of RAM
(2GB Ram per core) 275 TByte Storage 4.35
MSpecInt2000 5.43 MSpecFP2000 - more than 44
kSpec(Int/FP)Rate CISCO Topspin Infiniband-4X
15.000.000 in 3 year ! 250 man years ! (2/3
for new employees)
15
GRID and more...... we are building the future
INAF OACT supports main activities approved in
the PI2S2/Cometa project and other important
initiatives.
Large Scale Structure of the Universe and IGM (V.
Antonuccio, J. Silk , U. Becciani).
High-resolution simulations of the interaction
of Black Holes jet with the InterGalactic
Medium. Parallel codes MPI (FLASH) and MPI2
(FLY). Estimated not less than 160,000 CPU/hours
per year. Global Magneto-hydrodinamic
simulations in Galaxyes (A. Bonanno).
Understanding the non-linear evolution of the
Magneto Rotation Instability and Tayler
instability in the presence of Dark Matter.
Parallel codes MPI ZeusMP and Nirvana. Estimated
90,000 CPU/hours per year. Virtual Observatory
(U. Becciani, A. Costa). Development of a
Theoretical Virtual Observatory (ITVO) archive
integrated with grid-services (execution of jobs
in the grid). Serial and parallel codes.
Estimated 15,000 CPU/hours per year.
16
Other current main activities
COROT Mission (A.F. Lanza, A. Bonomo). Search for
extrasolar earth-like planets using the method of
the transits. . Estimated 200,000 CPU/hours per
year. GAIA Mission Sphere reconstruction task.
Status code revision. Support to GAIA-DPC PLANK
Mission Support for Plank/LFI simulations Model
of the current Stellar Perturbations on the Oort
Cloud (G. Leto et al.). Evaluation of the overall
effect of stellar perturbation on Oort Cloud
objects, for the development of a complete
evolution model of the Solar System including
gravitational perturbations. Estimated 250,000
jobs per year. Radio Emission from Stellar
Magnetosphere (P. Leto et al.). Study and
modeling of stellar magnetosphere from simulation
of their radio emission, in camparison with radio
measures with VLA. Estimated not less than
100,000 CPU/hours per year. Faint Galaxies in
Astronomical Images (R. Scaramella, S. Sabatini
INAF OARM). Faint galaxies galaxy formation
processes and studies of the baryons and of the
"dark matter" in the Universe. Estimated not
less than 100,000 CPU/hours per year.
Computational Cosmology (S. Borgani -
INAF-OATS). Detailed description of a number of
physical and astrophysical effects (star
formation, stellar and AGN feedback, metal
enrichment, thermal conduction, gas viscosity).
Estimated not less than 100,000 CPU/hours per
year. . and so on
17
GRID and more...... we are building the future

• GRID _at_ INAF OACT today
• HPC Oriented 352 CPU/cores, 700 GByte RAM, 30
  TByte Disk Space
• 4 contracts for young fellows
• 1 contract sys admin.

GRID _at_ INAF OACT tomorrow The larger National
grid CometaScope (Univ-NA) CosmolabCresco
(ENEA) New call from MIUR (EC Objective 1) March
2008 New calls from Regional Government New
CPUs, New contracts New scientific goals
18
Tech Publications

• Period 1996 - 2003
• 9 papers on Refereed Journals and 3 invited
  talks
• 27 papers on Conference Proceedings
• Period 2004 2007
• 5 papers on Refereed Journals and 1 invited talk
• 22 papers on Conference Proceedings

19
Final Consideration

• Since 1988 up to 2000 1 Technological Position
  (U. Becciani)
• 2000 Upgrade 1 Technolgical position ? 1 Tech.
  Researcher (U. Becciani)
• 2001 New Technological position (half time) (A.
  Costa)
• 2002 2007 ZERO NEW POSITIONS ? ???

• Tech. Contracts 8 young fellows (4 currently)

Without new staff positions (young researcher and
senior) in the next few years I think that this
technological activity WILL STOP!!!