Concept visualizations for Geodynamo simulations
Topics:
visit
xsede
Conceptual renderings on 3x3 Grid for Br field, see Sequence and Video tabs below.
The table describes information about each tile in visualization
Southern half: Equatorial section looking towards South pole. | Top view: Shows north pole from top | Northern half: Equatorial section looking towards North pole. |
Volume rendering: Same view as center | Quad cut: Shows section with a two quatrants clipped off | Isosurface: Eight isosurfaces with same view as center |
Back half: Longitudinal section, showing back half from front | Bottom view: Shows south pole from bottom | Front view: Shows view from front |
- Here we can replace any tile with another view / section / variable / different kind of plot.
- We can also create another configuration perhaps show different variable in each row. e.g. 3 rows and 5 columns
- Feel free to add comments here, or send feedback via email.
- Stampede issue: VisIt's Index Select operator which we use, has a bug in parallel mode, but works in serial mode. I am talking to VisIt software developers for a fix.
Tabs
Credits:
Simulation and Data: Ashley Willis3 , Magie Avery1, Chris Davies2, Cathy Constable1, David Gubbins2
Science Advisors: and David Gubbins2 and Cathy Constable1
Visualization: Amit Chourasia1
1 UC San Diego; 2 University of Leeds, UK; 3 University of Sheffield, UK
Comments
A few comments from Cathy and Maggie
Submitted by Catherine Constable on
(1) In general we think the sections look great.
(2) I have a question about the equatorial slices looking poleward on the outside left and right of the top row. They seem somehow backward from what I expect based on the other views. My impression is that the inner core ought to be yellow and all the action should be in the outer shell. But perhaps I am misunderstanding what I'm looking at.
(3) I believe a remaining question for you is whether your visualizations now take account of the fact that the radial spacing in our state files in physical space correspond to the zero crossings of the radial Chebyshev polynomials used in the simulations. Maggie thinks perhaps they do because VISIT can interpolate an irregular spacing, but I don't know if the state files you have been given correspond to
r, theta, phi, u_r, u_theta, u_phi, b_r, b_theta, b_phi
or if the r, theta, phi information is considered implicit. It might be that we need to supply additional information there.
(4) One addition Maggie and I discussed is somehow putting $B_r$ and $u_r$ simultaneously on one plot. We're not sure how yet.
2) Take a look at this image
Submitted by Amit Chourasia on
2) Take a look at this image https://www.seedme.org/node/26236
For top-row left image: Imagine cutting the top hemisphere off in the bottom-row left image and then look downward from above. You can notice not much happing along equatorial plane in middle-row.
3a) VisIt does interpolation based on nodal grid data, which I think is the case here, but we should discuss this further.
3b) Yes I think the state files had r, theta, phi, u_r, u_theta, u_phi, b_r, b_theta, b_phi fields in the data.
4) Certainly we can think couple B_r and U_r fields. One idea might be to plot B_r shell and deform it based on U_r or vice versa
Turns out deformation is only
Submitted by Amit Chourasia on
Turns out deformation is only supported for 2d plots, wonder if deforming the projections be any good.
Another idea to exploit symmetry in the simulation by plotting B_r vs U_r in adjacent two quadrants.