Download ppt - RISC

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
WP A1-c „SEE-GRID“
(Virtual Eye Surgery)
Wolfgang Schreiner
Research Institute for Symbolic Computation (RISC)
[email protected]
SEE-GRID

Joint Project

Research Institute for Symbolic Computation (RISC)


Upper Austrian Research GmbH (UAR)


Wolfgang Schreiner, Rebhi Baraka*, Karoly Bosa*
Michael Buchberger, Thomas Kaltofen, Daniel Mitterdorfer*
SEE++ Software (UAR)

Biomechanical model of the human eye
for simulating surgical eye muscle operations.
* Students funded from the project in 2004.
RISC/UAR
http://www.risc.uni-linz.ac.at
http://www.uar.at
2
SEE++ Problem Domain

Basics



Eye model: set of parameters
describing the biomechanical
structure of an eye
Gaze pattern: 2D diagram of
reference points
Problems


RISC/UAR
Forward: compute gaze
pattern from eye model
Inverse: compute eye model
from gaze pattern
http://www.risc.uni-linz.ac.at
http://www.uar.at
3
SEE++ Architecture

SEE++ Clients



SEE++ Server


Standalone application
Java applet
Simulation component
SOAP interface

RISC/UAR
RMI bridge for applet
http://www.risc.uni-linz.ac.at
http://www.uar.at
4
SEE-GRID
Transform SEE++ to a grid service


RISC/UAR
Globus bridge (Client remains unchanged)
http://www.risc.uni-linz.ac.at
http://www.uar.at
5
SEE-GRID

Parallel/distributed computation of gaze patterns.

Speculative parallelism and data parallelism.
hessDiagramGrid(E1, E2, E, I):
T := lookupCache(E1, E2, E, I)
if T <> null then
C := gridWait(T)
clearCache(E1, E2, E, I)
return C
T1 := gridStart(hessDiagramGridCore, E1, E2, E, I)
T2 := gridStart(hessDiagramGridCore, E2, E1, E, I)
C := gridWait(T1)
storeCache(T2, E1, E2, E, I)
return C
RISC/UAR
http://www.risc.uni-linz.ac.at
http://www.uar.at
hessDiagramGridCore(E1, E2, E, I):
for each intended gaze position iI do
T[i] := gridStart(position, E1 ,E2 ,E, i)
for each intended gaze position iI do
C[i] := gridWait(T[i])
return C
6
SEE-GRID

Parallel/distributed computation of eye models


RISC/UAR
Combinatorial search problem (branch and bound algorithm)
Hierarchical master/worker scheme.
http://www.risc.uni-linz.ac.at
http://www.uar.at
7
SEE-GRID

Grid database of eye models and gaze patterns


Database search involves parallel simulations.
Source and destination of grid computations.
1 Enter data in local
database
4 Search for data in
grid database
2 Forward data to
grid database
3 Grid calculations that
use/generate data
RISC/UAR
http://www.risc.uni-linz.ac.at
http://www.uar.at
8
SEE-GRID

SEE-GRID middleware



Globus 3.2 WS (pre-WS only as a temporary solution)
Globus 4 when available and useable
Higher-level functionality?

Support for grid database.
http://www.risc.uni-linz.ac.at/parallel/projects/agrid
RISC/UAR
http://www.risc.uni-linz.ac.at
http://www.uar.at
9
Related documents