RT Update
I ran a higher resolution simulation with my module, all parameters the same as before. The grid is now 50 x 150 (half of the first run), but AMR is turned on with 3 levels. So the effective resolution is 4x the first run? RTmovie (Higher Resolution)
Putting together a library of summaries
As I collect and read through the literature, I thought it would be good for my later reference, as well as potentially for others, to put together a listing of what I learn. I thought it would have a good home on the wiki, so it can easily be made public to others. You can find this on my page, under the projects heading.
https://clover.pas.rochester.edu/trac/astrobear/wiki/u/EricasLibrary
Interesting and relevant paper: "Turbulence generation by a shock wave interacting with a random density inhomogeneity field" (2010)
Folks,
I just ran across what may be an interesting paper, titled above. It's by a guy in Spain, C. Huete Ruiz de Lira at the Instituto de Investigaciones Energéticas.
Here's a link to the abstract: http://iopscience.iop.org/1402-4896/2010/T142/014022/
This seems right up our alley, though to be honest with just glancing at it I'm not sure I have the mathematical background to appreciate everything he's doing.
Let me know if you can't get access to it and I'll see what I can do (it's from the journal Physica Scripta).
For completeness, here's the abstract:
"When a planar shock wave interacts with a random pattern of pre-shock density non-uniformities, it generates an anisotropic turbulent velocity/vorticity field. This turbulence plays an important role in the early stages of the mixing process in a compressed fluid. This situation emerges naturally in a shock interaction with weakly inhomogeneous deuterium-wicked foam targets in inertial confinement fusion and with density clumps/clouds in astrophysics. We present an exact small-amplitude linear theory describing such an interaction. It is based on the exact theory of time and space evolution of the perturbed quantities behind a corrugated shock front for a single-mode pre-shock non-uniformity. Appropriate mode averaging in two dimensions results in closed analytical expressions for the turbulent kinetic energy, degree of anisotropy of velocity and vorticity fields in the shocked fluid, shock amplification of the density non-uniformity and sonic energy flux radiated downstream. These explicit formulae are further simplified in the important asymptotic limits of weak/strong shocks and highly compressible fluids. A comparison with the related problem of a shock interacting with a pre-shock isotropic vorticity field is also presented."
Update for week of July 24, 2011
I've handed in some notes on the AstroBEAR paper draft. I've also updated the list of office procedures (including passwords).
This week is mostly going to be tying up loose ends. If Jason is around this week I'll read him in on any important remaining issues.
Paper figures
I made a few additional paper figures for the ScramblerPaper. Now just need to integrate them into the descriptions in the text.
.
Documented the ambient object and process.data file
Documentation on the ambient object is now available https://clover.pas.rochester.edu/trac/astrobear/wiki/AmbientObjects
I also added documentation about the optional process.data file https://clover.pas.rochester.edu/trac/astrobear/wiki/ProcessDataExplained
MultiClump Runs - finished - sort of...
Updated the MHD Clumps Page with information about the By_sparse run… There seem to be some issues with the PPM and magnetized cooling rarefactions that produce ripples…
BE update
Here is pressure http://www.pas.rochester.edu/~erica/Pressure_07182011.gif
Here is density http://www.pas.rochester.edu/~erica/Density_07182011.gif
TicketCharts is working again.
Rich found the problem that was keeping the TicketCharts macro from working—apparently a Python module was missing, and the error message was turning up in an odd place.
Anyway, the pie graphs in the debug meeting page are functional again.
meeting update
I've been working on the RT problem and the implementation of the resistive MHD this weekend. I've also updated the project wiki page.
Meeting update - Week of July 18, 2011
The Building AstroBEAR page has been updated to reflect VPN changes brought about by the Great Crash. There is also a new section with instructions for building AstroBEAR on alfalfa
.
Speaking of which, the RAID setup on alfalfa
is being reconfigured to allow us access to the full 5 TB (this gives us RAID 1).
Some preliminary edits have been made to the Debugging Meeting page to make it clearer which tickets are covered by which reports. I'm still working with Rich to get the TicketCharts macro up and running again (which will restore pie graphs to the debug meeting page).
I'm currently looking over the draft of the AstroBEAR 2.0 paper.
400 km/s Runs of CRL 618
Running 400 km/s clump and jet in Bluehive both using 32 processors. These are 96 x 32 x 32 + 4 amr simulations.
We set an ambient density of 1,000 cm-3 and flow density of 100,000 cm-3 with a 1/r2 density profile. The temperature is 10 K. For a 400 km/s shock, we expect a postshock temperature of ~3.6 x 106 K. This corresponds to cooling between 10-23 and 10-22.
Incomplete number density movies http://www.pas.rochester.edu/~blin/sum11/400dens.html
Weekly Update- Week of July 18th
I have been working on initiating collapse of the Bonner Ebert Sphere. Why does a BE sphere that has no change across its boundary in density, pressure, and temperature, collapse under self gravity, WHILE a BE sphere that has a 20x greater external pressure does not? Is astrobear decreasing the density by a similar factor in the ambient to maintain an isothermal state…
NO change to external medium from the inside of the sphere. (Pout = (rho_out/rscale)*(iso_t/tempscale), rhoOut=rho_out/rscale). This causes collapse, as we saw before, because the medium is more massive than the clump. We rectified this by dividing rhoOut by a weight that decreased the ambient density.
Now, would expect, by leaving the external density the same, and INCREASING the external pressure further, to also see collapse. I am unsure of why this is not happening. The external pressure (Pout) is now about 20x greater.
Module Tutorial Updated
I've finished updating the Wiki tutorial page on writing a problem module. The biggest changes were in the initializing a grid routine. I put in more info on the q-array, which USE statements are required, a heavily-commented sample module, and a brief bulleted list of what I felt was most important.
The q-array might have more details that I missed. Also, I did not add anything about the aux-array since I don't know much about it. For the same reason, I didn't touch the Additional Physics sections.
RT Instability appears to be working
I have a seemingly successful simulation of the RT instability. The simulation is in 2D with a fixed grid of 100 x 300. The next step is to crank up the resolution, turn AMR on and then calculate the growth rate to check if it is indeed working accurately. RTmovie
Updates to Building Astrobear tutorial
I've made some updates to the Building Astrobear tutorial to reflect the system changes brought on by the Great Crash:
- Bluehive and Bluegene now sport reminders to set up a VPN connection, as well as links to the University IT page that explains how to do so.
- Grass's build instructions have been updated with the new module list, as well as how to enable modules in your
bash
environment. - A new section has been added for
alfalfa
build instructions. This section looks a lot like grass, except that it also has a VPN reminder.
Agenda Items
I am hoping that folks will start to get their stuff on the Wiki by sunday night (post with general description of results for that week) so I can check them out and have an idea for setting agenda's.
Back on bluehive and bluegene
Apparently my account just had a lock on it - similar to other users following the attack on the crc clusters… but because I was out of town and didn't discover or report the issue until a week later - the dots weren't connected… (It didn't help that my ticket was re-assigned to someone on vacation)
Anyways, I should be able to get the multiclump runs back up and running on bluegene… (That and work on the paper)
Meeting update - week of July 11
I have created a page for the bonner ebert project. This can be found on my page https://clover.pas.rochester.edu/trac/astrobear/wiki/u/erica, or the projects page, https://clover.pas.rochester.edu/trac/astrobear/wiki/AstroBearProjects.
Here is a direct link:
https://clover.pas.rochester.edu/trac/astrobear/wiki/BonnerEbert
Meeting Update for week of July 10, 2011
Alfalfa, the new grass-style machine, is now <palpatine_voice>fully operational</palpatine_voice>. It can be reached via ssh:
ssh <username>@alfalfa.pas.rochester.edu
Users still need a VPN to connect to alfalfa from off-campus, but no VPN is needed to connect from another machine in the PAS domain.
All of the modules accessible on grass
and clover
are accessible on alfalfa
, as is ifort
. AstroBEAR's Makefile.inc.grass
seems to work for alfalfa
as well.
It turns out we lost email notifications when the wiki was brought back online, due to changes in the department's SMTP protocol configurations. This problem has since been fixed, and users are receiving emails about tickets again. We are still working to restore notifications about repository checkins.
I've created a page about scaling in AstroBEAR; it's still pretty rudimentary, but it's at least RFC-ready at this point. This is part of my effort to consolidate some of the tags in our wiki, and eliminating the 'attention-needed' category. Most of other 'attention-needed' pages were based on AstroBEAR 1.0 specs, so I moved them over to 'deprecated'.
Finally, an open question for the group: what should go on the visualization page? Most of the visualization tutorials are on the [Tutorials tutorials] page; should they be linked on the visualization page as well? Is there something else entirely that people would like to see there?
Visited UNC to work with Christina on setting up Colliding Flows
I had a good time in Chapel Hill working with Christina and Fabian on the Colliding Flows module. We were able to get 2D AMR Hydro and MHD working before the queuing system stopped working. We did manage to perform a 3D AMR Hydro 64+4 simulation on 128 processors for an effective resolution of 512. The code seemed to perform quite well tracking the NTSI interface and Christina should be posting some nice screenshots soon
NaN Errors with 2 AMRs
In Bluehive, Martin and I are running a 500 km/s cooling jet with number density of 10,000 cm-3 and temperature of 10 K. The first time running, we use 2amr + cfl=.5. This shows nan error in fluxes (see /scratch/mhuartee/astrobear2/bullet3/jet500kms/bullet3.767749.bhsn-int.bluehive.crc.private.out).
We then update to using 3 amr, but the simulations are taking a while.
How do we interpret the nan flux errors?
Magnetized Clump Project Page
I created a page for the magnetized cloud/clump project. I will be mainly talking about it in the meeting.
https://clover.pas.rochester.edu/trac/astrobear/wiki/AstroBearProjects/MagnetizedClumps
Meeting Update for week of July 3, 2011
You've probably already noticed it, but the wiki's front page has been reorganized a bit.
Also, references to /var/trac and /var/repositories have been replaced with /data/trac and /data/repositories, respectively. The two exceptions are:
- Documentation pertaining to AstroBEAR 1.0.
- Tickets created prior to the Great Crash of '11.
I'm not sure whether it's worth altering AstroBEAR 1.0 documentation, and I thought I'd get the group's opinion before I start tampering with comments on old tickets.
In other news, the data file tutorials page is now completely filled out. Tutorials now exist for all the major .data
files and most of the minor ones (problem.data
being the exception).
And in other other news, alfalfa
, our new grass
-style machine, is online. It doesn't have an Intel compiler installed yet, but it's another 2 TB of storage for people who need it. It is currently only accessible from on campus, unless a VPN is used. A subsequent blog post will have more details.