Overview and Introduction to Scientific Visualization

Transcription

1 Overview and Introduction to Scientific Visualization Texas Advanced Computing Center The University of Texas at Austin

2 Before We Begin Please: Open a browser Go to // g1yteg6fp878hqm/bz3nxq8km_ Download disk_out_ref.ex2 Download and unzip smallwrf.zip

3 Scientific Visualization The purpose of computing is insight not numbers. -- R. W. Hamming (1961)

4 Visualization Process Summary A picture is worth not just 1000 words, but potentially tera- or peta-bytes of data Larger datasets demand not just visualization, but advanced visualization resources and techniques Visualization system technology leverages lots of advancing technologies: GPUs, high-speed networks, web technology. Visualization software takes time to adapt

5 Today Introduction to Visualization Information Visualization Scientific Data Visualization With Paraview Parallel Visualization For Very Large Data

6 Visualization of Large Scale Turbulent Flow Kelly Gaither, Hank Childs, Greg Johnson, Karl Schulz, Cyrus Harrison, Diego Donzis, Texas A&M; P.K. Yeung, Georgia Tech Remote interactive visualization of 17 time-steps (34 TB) of the largest turbulent flow simulation computed to date ( ). First time this had been visualized interactively at this scale. Equal parts data mining and remote interactive visualization goal was to characterize flow behavior over time. Gaither, K., Childs, H., Schulz, K., Harrison, C., Barth, W., Donzis, D., and Yeung, P.K., Using Visualization and Data Analysis to Understand Critical Structures in Massive Time Varying Turbulent Flow Simulations, IEEE Computer Graphics and Applications, 32(4), Jul/Aug 2012.

7 Stellar Magnetism Greg Foss, TACC; Ben Brown, University of Wisconsin, Madison A Sun-like star undergoes magnetic cyclic reversal shown by field lines. Shifts in positive and negative polarity demonstrate largescale polarity changes in the star. Wreath-like areas in the magnetic field may be the source of Sun spots. Terabytes of data to mine through and visualize.

8 Coronary Artery Nano-particle Drug Delivery Visualization Ben Urick, Jo Wozniak, Karla Vega, TACC; Erik Zumalt, FIC; Shaolie Hossain, Tom Hughes, ICES. A computational tool-set was developed to support the design and analysis of a catheter-based local drug delivery system that uses nanoparticles as drug carriers to treat vulnerable plaques and diffuse atherosclerosis. The tool is now poised to be used in medical device industry to address important design questions such as, "given a particular desired drug-tissue concentration in a specific patient, what would be the optimum location, particle release mechanism, drug release rate, drug properties, and so forth, for maximum efficacy? The goal of this project is to create a visualization that explains the process of simulating local nanoparticulate drug delivery systems. The visualization makes use of 3DS Max, Maya, EnSight and ParaView.

9 Volume Visualization of Tera-Scale Global Seismic Wave Propagation Carsten Burstedde, Omar Ghattas, James Martin, Georg Stadler and Lucas Wilcox, ICES; Greg Abram, TACC Modeling propagation of seismic waves through the earth helps assess seismic hazard at regional scales and aids in interpretation of earth's interior structure at global scales. Discontinuous Galerkin method used to for numerical solution of the seismic wave propagation partial differential equations. Visualization corresponds to a simulation of global wave propagation from a simplified model of the 2011 Tohoku earthquake with a central source frequency of 1/85 Hz, using 93 million unknowns on TACC s Lonestar system.

10 H1N1 Flu Outbreak Simulation: Integrating Science and Art Greg Johnson, Brandt Westing, Karla Vega, Kelly Gaither, TACC; Ned Dimitrov, Lauren Meyers, UT Comp. Bio; Francesca Samsel, Austin, TX. Visualization of a swine flu epidemic spreading throughout North America. Epidemic begins in Mexico City. Visualization classifies individuals into three groups: susceptible (blue), infected (red), and recovered (green). Available antivirals are shown in purple. Cities and transportation links are highlighted in red to indicate large numbers of infected individuals and infectious travelers.

11 Texas Pandemic Flu Toolkit Greg Johnson, Adam Kubach, TACC; Lauren Meyers & group, UT Biology; David Morton & group, UT ORIE.

12 Visualizing Distributed Learning Weijia Xu, Karla Vega, Makoto Sadahiro Researchers are conducting studies to compare the effectiveness of virtual and real manipulatives in the support and success of learning This research explores the use of information and scientific visualization tools for data mining and feature detection

13

14

15 Visualization Allows Us to See the Science Raw Data Visualization Application

16 Getting from Data to Insight Data Source Data Representation Visualization Algorithms Rendering Display Computational Refinement Changing Changing Steering Techniques View And using insight to get more insight

17 I Think Of Two Kinds Of Data For Visualization Data for Scientific Visualization F(spatial dimensions[, time]) -> attributes E.g. Weather data: F(latitude, longitude, altitude) -> temperature, wind velocity, direction humidity Data for Information Visualization List of facts, which have multiple attributes E.g. A list of movies: Title, year, director, length, gate, male/female leads..

18 Scientific Data altitude longitude

19 Info Data Gene Expression Data

20 Any Number of Packages Do Viz R (GenomeGraphs) Python (matplotlib)

21 Probably The Most Common Viz Tool

22 Geosciences

23 SciVis Data Data mapped onto a computational domain Heat distribution in a block of material F(x,y,z) -> temperature for (x,y,z) a point in the block of material Multiple variables (or properties) Weather F(x,y,z) -> pressure, temperature, wind-velocity for (x,y,z) a point in the atmosphere

24 SciVis Data Dimensionality Domain is generally 1, 2, 3 or more dimensions Directly interpreted geometrically (see heat distribution) Indirectly interpreted geometrically F(lat, lon) -> temperature X = (earth radius)*cos(lat)*cos(lon) Y = (earth radius)*cos(lat)*sin(lon) Z = (earth radius)*sin(lat)

25 SciVis Data and Time Time may vary also Heat transfer in a block of material F(x,y,z,t) -> temperature for (x,y,z) a point in the block of material and t a point in time

26 Higher Dimensional SciVis Content(l state(t) = = 00,(l, l 01 l, 1, l 12 l, 2, l 23 l, 3, l 34 l, 4, l 45 l, 5, l 56 l, 6 )) l 6 = ) - A 7D l 0 space + l 1 + lcurve 2 + l 3 + representing l 4 + l 5 + l 6 - the A 7D amount function of representing oil stored at a the given total point amount in time of oil stored at the point (l 0, l 1, l 2, l 3, l 4, l 5, l 6 ) - A contour surface at C represents all the ways C barrels of oil can be contained in 7 tanks An Oil Tank Farm

27 SciVis Data: Discrete vs. Continuous Data Discrete data is known at a finite set of points in the domain F(x,y,z) -> temperature for (x,y,z) from a finite set of points in the domain, unknown otherwise Continuous data is known throughout the domain F(x,y,z) -> temperature for all points (x,y,z) in domain

28 The Grid Points in the domain can be regular - specified by origin, delta vectors and counts, or explicitly listed For interpolated grids: Topology: how the points connected (implicit or explicitly listed) Interpolation Model: How data values at an arbitrary point are derived from nearby points

29 Types of data at a point/cell Scalar Vector Tensor/matrix Labels, identifiers Other tuples

30 Example

31 Example Points

32 Example Connectivity

33 Example Data Visualized

34 Example Varying in Time

35 Types of Input Data Point scattered values with no defined structure

36 Types of Input Data Grid regular structure, all voxels (cells) are the same size and shape

37 Types of Input Data Curvilinear regularly grided mesh shaping function applied

38 Types of Input Data Unstructured grid irregular mesh typically composed of tetrahedra, prisms, pyramids, or hexahedra.

39 Types of Input Data Non-mesh connected point data (molecular)

40 Visualization Operations Surface Shading (Pseudocolor) Isosufacing (Contours) Volume Rendering Clipping Planes Streamlines

41 Surface Shading (Pseudocolor) Given a scalar value at a point on the surface and a color map, find the corresponding color (and opacity) and apply it to the surface point. Most common operation, often combined with other ops

42 Isosurfaces (Contours) Surface that represents points of constant value with a volume Plot the surface for a given scalar value. Good for showing known values of interest Good for sampling through a data range

43 Clipping / Slicing Planes Extract a plane from the data to show features Hide part of dataset to expose features

44 Particle Traces (Streamlines) Given a vector field, extract a trace that follows that trajectory defined by the vector. P new = P current + V P t Streamlines trace in space Pathlines trace in time

45 Visualization Techniques Surface Rendering is an indirect geometry based technique Direct Volume Rendering is a technique for the visualization of 3D scalar data sets without a conversion to surface representations

46 Volume Rendering Expresses how light travels through a volume Color and opacity controlled by transfer function Smoother transitions than isosurfaces

47 Example Volume Rendered

48 Visualization Resources Personal machines Most accessible, least powerful Projection systems Seamless image, high purchase and maintenance costs Tiled-LCD displays Lowest per-pixel costs, bezels divide image Remote visualization Access to high-performance system, latency can affect user experience

49 TACC Support for Remote, Interactive Visualization Maverick Longhorn I don t have a picture Wrangler 2014(?) I don t have a picture Maverick 2014 Spur Stampede 2013 Lonestar 2011 Scientific Visualization Visual Analytics

50 XSEDE Visualization Resources Maverick (TACC) - HP core Ivy Bridge nodes 256GB system memory per node Nvidia Tesla K40 GPU FDR InfiniBand interconnect Designed for interactive visualization Stampede (TACC) - DELL core Sandy Bridge nodes 256 GB system memory per node 1TB) FDR Infiniband All with 62-core Xeon Phi 128 with NVIDIA K20 GPUs 205 TB Aggregate Memory

51 XSEDE Visualization Resources Maverick (TACC) - HP core Ivy Bridge nodes 256GB system memory per node Nvidia Tesla K40 GPU FDR InfiniBand interconnect Designed for interactive visualization Stampede (TACC) - DELL core Sandy Bridge nodes 256 GB system memory per node 1TB) FDR Infiniband All with 62-core Xeon Phi 128 with NVIDIA K20 GPUs 205 TB Aggregate Memory

52 Visualization Challenges

53 Visualization Allows Us to See the Science Raw Data Geometric Primitives Pixels Realization as Visual Entities Render

54 But what about large, distributed data? Gather the data? Realization as Visual Entities Render Gather

55 But what about large, distributed data? Gather Render Realization as Visual Entities Gather the geometry?

56 But what about large, distributed data? Depth-composite pixels Realization as Visual Entities Render Render full sized frames in place and depth-composite?

57 Or large, distributed displays?

58 Visualization Scaling Challenges Moving data to the visualization machine Most applications built for shared memory machines, not distributed clusters Image resolution limits in some software cannot capture feature details Displays cannot show entire highresolution images at their native resolution

59 Moving Data How long can you wait? File Size 10 Gbps 54 Mbps 1 GB 1 sec 2.5 min 1 TB ~17 min ~43 hours 1 PB ~12 days ~5 years

60 Old Model (No Remote Capability) HPC System Local Visualization Resource Data Archive Display Pixels Mouse Remote Site Wide-Area Network Local Site

61 New Model Remote Capability HPC System Large-Scale Visualization Resource Pixels Mouse Display Data Archive Remote Site Wide-Area Network Local Site

62 Analyzing Data Visualization programs only beginning to efficiently handle ultrascale data 650 GB dataset -> 3 TB memory footprint Allocate HPC nodes for RAM not cores N-1 idle processors per node! Stability across many distributed nodes Rendering clusters typically number N <= 64 Data must be dividable onto N cores Remember this when resampling!

63 Displaying Data Dell 30 flat-panel LCD 4 Megapixel display 2560 x 1600 resolution

64 Displaying Data Stallion world s 2 nd highest-resolution tiled display 328 Megapixels x 8000 pixel resolution Dell 30 LCD

65 What s the solution?

66 Summary Challenges at every stage of visualization when operating on large data Partial solutions exist, though not integrated Problem sizes continue to grow at every stage Vis software community must keep pace with hardware innovations