cse5559::visualizing the life and anatomy of cosmic particles
TRANSCRIPT
Target Task
● T2:Halo Identification and Visualization
– Visualize the evolution of halos overtime
– Visualize the evolution of a specific halo of interest(i.e halo with highest mass)
– Visualize the evolution of the internal structure of the halo of interest.
– Analyze the mass accrual pattern of the highest mass halo.
● T3:Diving Deep into Halo Substructures
– Provide a grid based representation scheme of the dark matter particles.
– Provide a Particle Based Volume Rendering framework to visualize the particle layout in the dataset.
Halo Mass Accrual History
● The mass of a halo is estimated using a theorem called “Virial Theorem” which is not a true estimator of mass[1]. Below is a plot showing how this virial mass of the largest halo accumulates over time. Also shown is a plot of the contribution of the dark matter particles in the mass of the halo.
● [1]: http://spiff.rit.edu/classes/phys440/lectures/gal_clus/gal_clus.html
Nearest Neighbor Particle Density Estimation
● Determining Grid Resolution: The number of grid per unit cell(cpud) is given by
● Particle Insertion: we create a 4D vector (i.e, one for every grid cell) and store all ID's of the points associated with a particular grid location.
● Interpolation: We perform an inverse-distance based interpolation at each grid vertex. Controlling parameters are the radius of the neighborhood and the maximum number of contributing particles.
● Use Case: Once we have the grid we can use it for isosurface extraction and direct volume rendering.
Particle Based Volume Rendering
● Particle Generation: Traditional approach generate particles per cell by sampling based on grid point. But here we already have a set of particle positions. So we only have to map the particles to the individual cells, achieved by indexing the particles to individual cells.
● Particle Projection: This involves projection from the object space to the image plane and then designing a proper transfer function
● Spatial Superimposing: Use z-buffer to decide the particle closest to the image plane and decide the color for a pixel. To add translucency we divide the pixel to sub-pixels and do a weighted average to find the final value.