cocos – building a large scale cosmological simulation database_nowinski
TRANSCRIPT
COCOS
Cosmological simulation database
Aleksander NowiskiICM, University of Warsaw
Open Research Data, Warsaw 2015
In collaboration with: Wojtek Hellwing (simulation) and Wojtek Hury (software development)
OCEAN data analysis center
OCEAN is a new scientific data center built for ICM
O is for Open
Computers, databases, storage and people
COCOS is developed
as a part of OCEAN
Cosmological simulations
Simulations of the universe in a large scale
Does it work as we think?
Really large scale but also resolution: the more detail, the better answerMaybe we are part of somebody's simulation?
A simulation of the multiple particles within a large box
Afterwards we try to match it with real-world observation
Existing databases
There are already existing databases with results of the simulationMillenium simulation http://gavo.mpa-garching.mpg.de/Millennium/
MultiDark http://www.multidark.org/MultiDark/
SQL based interfaces (client types in query)
Dedicated extensions to databases, especially to solve coordinate based search
Limited but active audienceA will to share the results within the community
The Copernicus Complexio: COCO simulation
GADGET 2 code
2 simulations, for different physics ('warm' and 'cold' dark matter), run in ICM:2.5 mln. cpu hour Power 7 (cold)
1.8 mln. cpu hour Xeon (warm)
Wojciech A. Hellwing,
ICC, University of Durham,
ICM University of Warsawhttp://arxiv.org/abs/1505.06436
What is different about this database?
Just two datasets (actually 2+epsilon)
Each dataset is extremely largeraw data - 70TB (particle trajectories)
full data approx. 100TB
afterwards in post-processing complex objects are identified
database is created to allow individuals explore the results by analyzing identified phenomena (galaxies, dark matter halos etc.)
Challenge
Scale1010 particles, 169 frames,
Within results 104-106 objects identified
Objects evolve collide and aggregate
For analysis data from all levels is important grid (space properties), object and particle
Spatial queries for both objects and particles
Queries involving space and object evolution
Functionality
SQL interface for the users
Web interface for simple queries to support people too lazy to SQL
Ability to fetch also particle trajectories
API conforming selected IVOA standards
Technology
Java application
PostgreSQL databasebig challenge for the database
Custom solution for extracting particles
When?
This is ongoing work now
To be finished and open as a result of the OCEAN project December 2015
Stay tuned
Thank You!
Have you got any questions?