Towards an Environment for Multiscale Applications

tunisianbromidrosisInternet και Εφαρμογές Web

5 Φεβ 2013 (πριν από 4 χρόνια και 6 μήνες)

169 εμφανίσεις

1

The
Mapper

project receives funding from the EC's Seventh Framework
Programme

(FP7/2007
-
2013) under grant agreement n
°

RI
-
261507.

Towards Environment for
Multiscale Applications

Katarzyna Rycerz
,
Eryk Ciepiela, Joanna Kocot, Marcin Nowak,
Pawel Pierzcha
ł
a,

Marian Bubak

KUKDM
,
Zakopane,

9
-
11.03 2011

2

Overview


Multiscale simulations
-

overview


Tools for multiscale simulations


MUSCLE
-
based multiscale application in
GridSpace


Example demo with In
-
stent restenosis
application


Summary and future work


3

Multiscale Simulations


Consists of modules of
different scale


Examples


e.g. modelling:


virtual physiological
human initiative


reacting gas flows


capillary growth


colloidal dynamics


stellar systems


and many more ...



virtual physiological human
virtual physiological human
fusion
fusion
hydrology
hydrology
nano material science
nano material science
computational biology
computational biology
the reoccurrence of stenosis,

a

narrowing of a blood
vessel,

leading to
restricted blood flow

4

Tools for multiscale simulations


M
odel
C
ouling
T
oolkit



A
pplies a message passing

(MPI) style of communication between simulation models.


O
riented towards

domain data decomposition of the simulated problem


P
rovides a support

for advanced data transformations between different models


J. Larson, R. Jacob, E. Ong ”The Model Coupling Toolkit: A New Fortran90

Toolkit for Building
Multiphysics Parallel Coupled Models.” 2005: Int. J. High

Perf. Comp. App.,19(3), 277
-
292.


Astrophysical Multi
-
Scale Environment (
A
MUSE)


Scripting approach (Python) is used to couple models together., MPI used to distribute modules


Astrophysical models : stellar evolution, hydrodynamics, stellar dynamics and radiative transfer


S. Portegies Zwart, S. McMillan, at al. A Multiphysics and Multiscale Software Environment for
Modeling Astrophysical Systems, New Astronomy, volume 14, issue 4, year 2009, pp. 369
-

378


High Level Architecture components
:


Application components run concurrently and communicate using HLA mechanisms


Coponents are steerable from outside during runtime using script interface


Support for synchronisation between multiscale modules
-

time stamps, advanced time
management


K. Rycerz, M. Bubak, P. M. A. Sloot: HLA Component Based Environment For Distributed
Multiscale Simulations In: T. Priol and M. Vanneschi (Eds.), From Grids to Service and
Pervasive Computing, Springer, 2008, pp. 229
-
239





5


Goals


Support composition of simulation models.


scripting approach to programmatically access

simulation components to build multi
-
disciplinary and
multi
-
scale “in silico” experiments


Support execution of such experiments and achieve their
reusability


Integrate solutions designed for multiscale simulations’
development with possibilities given by
:


tools for multiscale simulations


environments

for application composition


computational
e
-
Infrastructures


6


Experiment workbench


Constructing experiment plans
from code snippets


Interactively run experiments


Experiment Execution Environment


Multiple interpreters


Access to libraries, programs and
services (gems)


Access to computing infrastructure
:
Cluster, grid, cloud


Experience


Virolab

project


PL
-
Grid

NGI


E. Ciepiela, D. Harezlak, J. Kocot, T. Bartynski,
M. Kasztelnik, P. Nowakowski, T. Gubała, M.
Malawski, M. Bubak; Exploratory Programming
in the Virtual Laboratory, in Proceedings of the
International Multiconference on Computer
Science and Information Technology pp. 621

628.


GridSpace

7

Multiscale Coupling Library and
Environment
(
MUSCLE
)



P
rovides a software framework to build simulations according to the complex automata

theory


I
ntroduces concept of kernels that communicate by unidirectional pipelines

dedicated to
pass a specific kind of data from/to a kernel

(asynchronous
communication
)


J. Hegewald, M. Krafczyk, J. Tlke, A. G. Hoekstra, and B. Chopard. An agent
-
based coupling platform for
complex automata. ICCS, volume 5102 of Lecture Notes in Computer Science, pages 227233. Springer,
2008.


# CxA configuration of sample application

# configure cxa properties

cxa = Cxa.LAST


cxa.env["max_timesteps"] = 2

cxa.env["cxa_path"] = File.dirname(__FILE__)


# declare kernels

cxa.add_kernel('w', 'examples.simplejava.Sender')

cxa.add_kernel('r', 'examples.simplejava.ConsoleWriter')


# configure connection scheme

cs = cxa.cs


cs.attach('w' => 'r') {


tie('data', 'data')

}


Sender

Module

Console


Module

Sample application

MUSCLE communication

8

GridSpace

for MUSCLE
application


Integrated environment for:


Configuring modules connections and
parameters in cxa file


Visualizing modules connections


Running application on a chosen

e
-
infrastructure


Interactive post
-
processing the output using
various tools (e.g. MATLAB)




9

MUSCLE application in

GridSpace


GridSpace Experiment host:



Interpreters and libraries accessing PBS



User files


GridSpace Experiment Workbench


QCG Grid Resource Management

System


BF

Module

SMC

Module

DD

Module

other

Modules

QCG infrastructure

MUSCLE communication

BF

Module

SMC

Module

DD

Module

other

Modules

Local DRMS (PBS)

MUSCLE communication

MUSCLE CxA Gra
ph
ical
Viewer

-

Models connections

Vari
o
us scripts editors

General (Python, Ruby, Perl ..)

Spec
i
fic (Matlab, Mathematica,
CxA interpreter)

Infrastructure access layer

User file management

-
simulation output view

cluster

cluster

10

Tool for automatic MUSCLE
application distribution


Main features


Accessible from GridSpace
level


Automatically distributes
MUSCLE applications in
GRID environment


Live stdout/stderr streaming


Based on Distributed Ruby
(DRb) and PBS



11

Demo



Instent restenosis in
GridSpace


http://www.youtube.com/watch?v=3S9
-
kljyXIw

12

Summary and Future
Work


GridSpace can be used as a high level tool for setting up and
running MUSCLE
-
based multiscale applications


We plan to extend our solution to a set of tools supporting
programming and execution of multiscale applications in
general


To control and test behaviour of such applications we plan to
support creation of their skeletons


parametrised „empty” multiscale application of the same structure
and requirements as the real one.


We plan support for various European


e
-
infrastructures and cloud resources


See: http://dice.cyfronet.pl