Download presentation

Presentation is loading. Please wait.

Published byLamar Pye Modified over 2 years ago

2
A Visualization Framework For Earth Materials Studies Bijaya Bahadur Karki Graduate Students: Dipesh Bhattarai and Gaurav Khanduja Department of Computer Science Department of Geology and Geophysics Louisiana State University, Baton Rouge, LA 70803 2nd Workshop: Minneapolis, August 5-10, 2007

3
Studying Materials Problems Simulation algorithms Compute- and data-intensive applications Visualization algorithms PWscf, VASP PCMD Parallel and distributed computing Mantle materials: Silicates and oxides Rheology Liquids Massive multivariate data: MDV STMR ReVis Tezpur (15.3 TFlops, 360 nodes 2 dual-core processor) Queen Bee (50.7 TFlops, 680 nodes 2 quad-core processors)

4
Visualization: Definition Process of making a computer image for gaining insight onto data/information –Transform abstract, physical data/information to a form that can be seen (i.e., visual representation) –Enhance cognitive process

5
Visualizing Materials Data Properties/processes of interest: –Microscopic Atomic structure, dynamics Electronic structure Data characteristics: –Three-dimensional, time-dependent –Multivariate –Massiveness, multiple sets –Computational, experimental origin –Macroscopic EoS, elasticity, thermodynamics

6
Application-Based Approach Numerous visualization systems exist –None of them may be good enough –Lack of desired functionality and flexibility How to meet domain-specific needs –Presentation and interactivity –On-the-fly data processing –Multiple sets of data –Visualization with database –Remote and collaborative visualization –Visualization/computational steering

7
Current Visualization Activities Multiple datasets visualization (MDV) –Electron density distribution Space-time multi-resolution (STMR) visualization –Atomic structure and dynamics Remote visualization –Elastic moduli and wave propagation

8
Multiple Datasets Visualization Simultaneous rendering of more than one set of data to examine cross-correlation among them Isosurface extraction GPU-based visualization Adaptive scalable approach

9
Example: Electronic Structure Mg 2- vacancy defect in MgSiO 3 post-perovskite PerfectDefectDefect - Perfect Difference in two images Initial configurations Final configurations (after relaxation)

10
Scalable Adaptive Isosurface Extraction Octree data structure High resolutionLow resolutionDual resolution Multiresolution approach Original cell Octree nodes

11
Performance Analysis Performance measurement on 64 sets of scalar volume data with size of 256 3 and 512 3 All-in-memory approach Scalable adaptive approach

12
GPU-Based Visualization Graphics hardware assisted 3D textures Interactive clipping Isosurface Khanduja and Karki WSCG 2005 GRAPP 2006 WSCG 2007

13
Example: Electronic Structure Mg 2- vacancy defect in MgSiO 3 post-perovskite PerfectDefectDefect - Perfect Difference in images Initial configurations Final configurations (after relaxation)

14
MDV Example 25 sets of the scalar volume data of 256 3 size in a planer clipped mode using 3D surface texture mapping Electron density in liquid MgO as a function of time Multi-scale color map: Blue: 0 to 0.05 Blue and green: 0.05 to 0.5 Red: above 0.5

15
Electron Density: Defects in MgSiO 3 ppv MgSiO Vacancies Migrating ions

16
Electron Density: Defects in MgSiO 3 ppv Spheres and lines Karki and Khanduja, EPSL, 2007 MgSiO Vacancies Migrating ions

17
Defects in MgSiO 3 ppv: Atomic Structure Mg: GreenSi: BlueO: RedVacancy site: Black MgSiO Vacancies Migrating ions

18
Space-Time Multiresolution (STMR) Atomistic Visualization Integration of visualization and complex analysis On-the-fly extraction and rendering of a variety of data Pair correlation, coordination and cluster structures Dynamical behavior

19
Atomistic Visualization Modules Approach –Spatial proximity –Temporal proximity –Spatio-temporal analysis Model –Complete data rendering –Local/extracted data rendering

20
Position-Time Series Data OHMgSi Atomic Species: Points: Complete data set Balls: Instantaneous configuration Data: {P(j t) | 0 ≤ j ≤ N} where P(t) = {p i (t) | 1 ≤ i ≤ n}

21
Coordination Environment Given atomic system: Hydrous MgSiO 3 liquid Atomic species: spheres 16 different pair correlation structures Cutoff distances from partial RDFs Si-O Coordination environment Coordination stability Coordination clusters Radial distribution functions

22
Pair Correlation Matrix OH Mg Si O H Mg Si 16 different types of nearest-neighbor pairs Diagonal: like atoms Off-diagonal: unlike atoms

23
Radial Distribution Function Spatial and temporal information on Si-O coordination

24
Coordination-Encoding 23456 Color map Three-, four- and five-fold coordination

25
Coordination Stability The lines (thickness encoding the bond stability) and center atoms (size encoding the coordination stability) are color- coded to represent, respectively, the length distribution and coordination states. The stability represents the fraction of the total simulation time over which a given bond or coordination state exists. Bhattarai and Karki, ACMSE 2007 23456 Color map

26
Stability of Different Coordination 3456 16 coordination states Four types exist 0123 4567 891011 12131415

27
Coordination Cluster The lines (thickness encoding the bond stability) and center atoms (size encoding the coordination stability) are color-coded to represent, respectively, the length distribution and coordination states. The stability represents the fraction of the total simulation time over which a given bond or coordination state exists. Spatial and temporal information on Si-O coordination Bhattarai and Karki, ACMSE 2007

28
Coordination Cluster Per Atom Spatial and temporal information on Si-O coordination

29
Coordination Visualization 16 different pair correlation structures Cutoff distances from partial RDFs Si-O Coordination environment Coordination stability Coordination clusters Radial distribution functions Atomic species: spheres Given atomic system: Hydrous MgSiO 3 liquid

30
Visualizing Dynamics Diffusion in 80-atoms liquid MgSiO 3 Spheres for atomic displacements Ellipsoids for covariance matrices Diffusion in 64-atoms liquid MgO Bhattarai and Karki, ACMSE 2007

31
Elasticity visualization Remote execution Visualization and database server Online data reposition

32
Elasticity Visualization - ElasViz Multivariate elastic moduli –Variation with pressure, temperature and composition Elastic wave propagation in an anisotropic medium –Velocity-direction surfaces –Anisotropic factors Karki and Chennamsetty, Vis. Geosci., 2004

33
ReadData CijPlot Modules of ElasViz GenerateDirectionGenerateVelocity AnPlot DrawVelocity Other ModulesDisplay

34
Global Visualization Mode

35
Selective Visualization Mode

36
Summary Visualization for gaining insight into a variety of datasets for important minerals properties and processes –Increasing amounts of data from simulations and other resources. Important visualization systems under development: –Elasticity, atomic and electronic data A lot needs to be done: –Adding more functionalities –Merging atomistic and electronic components –Extending for remote and distributed access –Adopting in virtual (immersive) environment. Support from NSF (EAR 0347204, ATM 0426601 and EAR 0409074).

Similar presentations

Presentation is loading. Please wait....

OK

Lecture #1 Introduction.

Lecture #1 Introduction.

© 2018 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Ppt on radio station Ppt on various means of transport and communication and their uses Ppt on real estate sector in india Ppt on world technology day Consumer behaviour ppt on luxury watch brands Ppt on mid point theorem for class 9 Slideshare download ppt on pollution Ppt on aircraft emergencies youtube Ppt on business cycle phases Ppt on water resources in hindi