1. How do multi streaming regions form and evolve?
Alex Pang, Uliana Popov – UCSC
Katrin Heitmann, Salman Habib, and Jim Ahrens - LANL

2. Why is it interesting?
Multi-streaming events are associated with large scale structures such as:
sheets,
filaments,
clusters, and
voids.
The identification and characterization of these regions will extend our understanding of cosmological evolution. 

3. Large structures
The structures range in size from 1 to 100 Mpc, ~ 3 to 300 million light years.
Filaments –
largest known structures
in the universe
Voids –
empty spaces between
filaments

4. What is multi streaming?
Multi-streaming events happen, when
- at some location particles arrive from different locations and
- at different velocities
Places where multi-streaming happens are singularities in coordinate space.

5. Previous Work Halo finders seek to identify - clusters or
- high density regions in space, based on particle position
It is believed that large scale structures start out as - sheets
→ filaments (where sheets intersect)
→ clusters (where filaments intersect)l
Problem:
Looking for halo or high density regions will miss these aspects of structure formation.

6. Characteristics Multi streaming can be characterized by
different descriptive properties
- high shear in flow
- transition to non linearity
- different velocities
- irrotational
- divergence

7. Local Methods Divergence Dot Product Linearity Max Shear Stress
measures a degree to which a vector field is a source or a sink at a given location
Dot Product
measures similarity of directions of two vectors
Linearity
measures how linear the field is
Max Shear Stress
shear stress is force that acts parallel on a surface area
Variance
measures variability of velocity using both direction and magnitude
Vorticity
measures the tendency of vector field to spin. The direction is the rotation axis and the magnitude is the angular speed

8. Power Spectrum

9. Different Scales
← coarse
miss fine features
← fine

10. Different Scales ← coarse ← (too) fine
missed most of the features, because they are bigger than the size of a grid cell

11. Tracking multi streaming regions
In order to see more detailed, step-by-step, formation of the structures.
There are 2 main tracking strategies:
- region tracking
- particle tracking

12. Tracking multi streaming regions
Region tracking - “high level” tracking, less precise, captures the whole feature

13. Tracking multi streaming regions
Particle tracking – more precise, but may miss the feature

14. Data Data set: 512*512*512 particles. Each frame is 4GB.
Simulation: 500 frames.
We need the ability to handle very large data sets.
ISSDM clusters

15. Hypothesis
There is more than one type of multi streaming

16. Tracking Results
Frame Frame Frame 21
Frame 20, zoom-in

17. Tracking Results
Frame Frame Frame 21
Frame Frame Frame 21

18. Summary
Goals:
- find multi streaming regions using velocity field information
- characterize these multi streaming regions
We have:
- local methods based on velocity info;
- preliminary results based on particle tracking.
- observation, that there seems to be different types of behavior for multi streaming regions.
Future goal:
- identify type of behavior based on properties from local methods.