Meeting Update 04/30/2013 - Eddie
- Finished the drift analysis for my 1D radiative shock models. You can see that the drift is indeed quite small. ehansen04262013
- Submitted introduction slide for the Star Formation Jamboree.
- Completed hydro run for 3D pulsed jet. The MHD runs quit due to divergence protections. I'm going to try initializing the field in a different way to see if I can fix this issue.
The run here is low resolution, a fixed grid of 45 x 450 x 45 which is 4 times less than the effective resolution of the original 2.5D runs.
Meeting update
- Flux limited diffusion is working in static case… need to still add explicit source terms. movie
- Working with Baowei and optimizing AstroBEAR's hydro update code.
- Confirmed correct growth of RT instability (yet again...)
Meeting Update Apr.30
- Star Formation: had a two hour discussion with Eric last Friday. There were some good new ideas and at the end we agreed that the phenomenon observed can be possible but need to be further verified. One idea we discussed is that the explosion happened on both sides of the bowshock (both expansions have a halo like feature, but the expansion on the righthand side is delayed) can be triggered by depressuring the two different regions. The lefthand side expansion can be triggered by the runaway outer halo (the spherical expansion is possible based on approximate calculation that the incoming flow from right to left resulted from colliding flows missing the star has similar Ram pressure compared to the incoming wind but much higher thermal pressure), the righthand side expansion can be triggered by the region not centered at the star but actually displaced to the right, where compression is highest (converging flow coming from the back of the clump, the incoming colliding flow attracted by the core but miss the star). I am currently making new plots according to our discussion.
- Krumholz accretion: fixed the way how the particle is initialized (basically, the original Krumholz accretion forms zero mass particles and the accretion rate is depend on the particle mass). A new restart run with fixed Krumholz accretion Mach = 1.5 triggered star formation is submitted.
- Resistive Paper: finished By Rm = 10 run, one run left (Bx Rm = 10). The paper will feature Rm = infinite, 300, 30, 10 runs for both Bx and By.
The week in review
- Will be meeting to shoot the footage for astrobear PR movie this Friday, going to do some discussion of multiphysics apparently
- Made a slide for the Star Formation Jamboree — should I include the BE plots I did? Should I include other names on this slide?https://astrobear.pas.rochester.edu/trac/astrobear/blog/erica04272013
- Finished the God. code. It was not exactly as trivial as I thought it would be. I practiced utilizing some different pieces of code, namely the Exact Riemann solver I already wrote for Euler's equations, using the module function of fortran for practice. To compile these separate files, I did 'gfortan file1.f90 file2.f90 -o program_name'. This command takes care of both compiling and linking the files into the executable program_name. Care must be taken that files are compiled in order such that the first ones are called by the later ones. The code I wrote is accessible from my God. page, here: https://astrobear.pas.rochester.edu/trac/astrobear/wiki/u/erica/GudonovMethodEuler Things compiled, but I am getting a Nan somewhere early on in the program. I am now debugging the code, which is good practice for tracking down Nans in astrobear. Will like to update that page with some notes on the code, and figures of the results, once I correct the bug.
Meeting Update 04/30/2013 -- Baowei
- Users
- Met with Arijit of LLE. Got visit run. Set up the 3DRT problem on LLE machines. Walked him through Jonathan's blog "Finding the RT Instability growth rate " https://astrobear.pas.rochester.edu/trac/astrobear/blog/johannjc04252013
- Tickets
- New #285 (Optimize AstroBEAR 2.0 on Blue gene/Q)
- Worked on
- Registered CIRC poster session
- Scaling test and optimization on Blue Streak(Ticket #285)
- Strong scaling of current revision code (done up to 2048 cores)
- Enabled parallelization of program code by turning on qsmp. Compiled, In queue for testing. (#285)
- Working on replacing vectorized FORALL loops with parallelized DO loops and corresponding modifications like swapping indices etc. in sweep_scheme.f90.
Drift of the 1D radiative shocks
I looked at the drift of the shock front for one computational time for all 18 models. One computational time is different for each model since the length scale is different for each model, so in order to compare apples to apples I decided that vdrift / vshock was the appropriate diagnostic.
The drift was not measurable for vs = 60 km/s. Even in the worst case, the drift velocity was only 0.065% of the shock velocity. Therefore, I think this effect is negligible. Obviously, if we ran the simulations for many computational times, the shock would eventually drift off the grid but that kind of simulation will take too long and will never be necessary. Still, there must be a very small error somewhere. I suspect it is somewhere in the chemistry or scaling, but again, the error is so small that we don't need to worry about it.
Findng the RT Instability growth rate
- First load the database into visit
- create a vector expression called gradrho which is 'gradient(rho)'
- create a scalar expression called gradrhoy which is 'gradrho[1]'
- make a pseudocolor plot for gradrhoy
- make slice(s) along y axis. (So for a 3D data set this would be 2 slice operators: orthogonal to x with 50 percent first — yz plane, orthogonal to z with 50 percent then — xy plane, or vice versa. Also make sure unmark 'Project to 2D' )
- Then perform a 'max' query using the 'actual data' (not the 'original data')
- This will spit out the maximum value of the gradient as well as the position.
- You could advance the time slider and repeat the query - but this is tedious so instead you can
- clear the query results
- Open the command dialog (from Control menu)
- copy and paste the following script
for state in range(TimeSliderGetNStates()): SetTimeSliderState(state) Query('max', use_actual_data=1)
- execute the script
- At this point the query dialog should have the location of the peak density gradient for each frame. Copy all the printed lines like this one:
VisIt: Message - gradrhoy -- Max = 21.989 (zone 4584 in patch level3,patch7 at coord <0.151042, 0.5, 0.1>)
Save this to a file and name the file visit_grepped.txt
- Now you can clean up the data using the following bash one-liner
sed 's/</ /g' visit_grepped.txt | sed 's/>)/ /g' | sed 's/,/ /g' | awk '{print .1*(NR-1), $15}' > data.out
Where we have used the awk expression
.1*(NR-1)
to calculate the time for each frame assuming each frame is .1 computational time units apart.
- Then in gnuplot you can load the data and fit it to any function you please… However you may want to also trim the data file so that you only have points during the linear growth phase.
- Then in gnuplot
g(x)=a+b*exp(c*x) fit g(x) './data.out' via a,b,c plot g(x), 'data.out'
Meeting update
- I've been studying the BlueGene Q architecture and the general tuning of codes to run on BlueGene. We should be able to get much better performance with a minor amount of effort. See this wiki page
- Hopefully with a little work, AstroBEAR can run more efficiently (6x) and I can resume the magnetized colliding flows run.
- FLD is on back burner until colliding flow run has resumed.
Meeting Update 04/23/2013 - Eddie
- Did all 18 1D radiative shock models ehansen04222013
- Trying to get 3D pulsed jet runs working. Had login, compile, and submission issues on Kraken and BlueStreak. Those are resolved now, but I am still having some runtime issues. Hopefully, they get resolved today.
Update
Finished reading through chapter 6 of Toro, here is the page I am creating for the Godunov method for Euler equations.. I know all the steps now in writing the scheme. I will next have to work on writing the program that will utilize both the conservation form of Godunov method, with the exact RP solver I used before. Once this is done, I will compare to Toro's output.
https://astrobear.pas.rochester.edu/trac/astrobear/wiki/u/erica/GudonovMethodEuler
Erini's Update -4/23/13
- still working on polytrope module
- almost done writing up generalized polytropes for wiki
1D Radiative Shock Models
Here is a table of the 18 models that I ran with plots of the hydrodynamic variables and the emission lines.
vs (km/s) | log(n) (cm-3) | beta | hydro | emiss |
---|---|---|---|---|
40 | 2 | 5 | ![]() | ![]() |
40 | 2 | 1 | ![]() | ![]() |
40 | 2 | 0.5 | ![]() | ![]() |
40 | 3 | 5 | ![]() | ![]() |
40 | 3 | 1 | ![]() | ![]() |
40 | 3 | 0.5 | ![]() | ![]() |
50 | 2 | 5 | ![]() | ![]() |
50 | 2 | 1 | ![]() | ![]() |
50 | 2 | 0.5 | ![]() | ![]() |
50 | 3 | 5 | ![]() | ![]() |
50 | 3 | 1 | ![]() | ![]() |
50 | 3 | 0.5 | ![]() | ![]() |
60 | 2 | 5 | ![]() | ![]() |
60 | 2 | 1 | ![]() | ![]() |
60 | 2 | 0.5 | ![]() | ![]() |
60 | 3 | 5 | ![]() | ![]() |
60 | 3 | 1 | ![]() | ![]() |
60 | 3 | 0.5 | ![]() | ![]() |
vs (km/s) | log(n) (cm-3) | beta | hydro | emiss |
---|---|---|---|---|
40 | 2 | 5 | hydro_v40_n2_b5.png | emiss_v40_n2_b5.png |
40 | 2 | 1 | hydro_v40_n2_b1.png | emiss_v40_n2_b1.png |
40 | 2 | 0.5 | hydro_v40_n2_b0.5.png | emiss_v40_n2_b0.5.png |
40 | 3 | 5 | hydro_v40_n3_b5.png | emiss_v40_n3_b5.png |
40 | 3 | 1 | hydro_v40_n3_b1.png | emiss_v40_n3_b1.png |
40 | 3 | 0.5 | hydro_v40_n3_b0.5.png | emiss_v40_n3_b0.5.png |
50 | 2 | 5 | hydro_v50_n2_b5.png | emiss_v50_n2_b5.png |
50 | 2 | 1 | hydro_v50_n2_b1.png | emiss_v50_n2_b1.png |
50 | 2 | 0.5 | hydro_v50_n2_b0.5.png | emiss_v50_n2_b0.5.png |
50 | 3 | 5 | hydro_v50_n3_b5.png | emiss_v50_n3_b5.png |
50 | 3 | 1 | hydro_v50_n3_b1.png | emiss_v50_n3_b1.png |
50 | 3 | 0.5 | hydro_v50_n3_b0.5.png | emiss_v50_n3_b0.5.png |
60 | 2 | 5 | hydro_v60_n2_b5.png | emiss_v60_n2_b5.png |
60 | 2 | 1 | hydro_v60_n2_b1.png | emiss_v60_n2_b1.png |
60 | 2 | 0.5 | hydro_v60_n2_b0.5.png | emiss_v60_n2_b0.5.png |
60 | 3 | 5 | hydro_v60_n3_b5.png | emiss_v60_n3_b5.png |
60 | 3 | 1 | hydro_v60_n3_b1.png | emiss_v60_n3_b1.png |
60 | 3 | 0.5 | hydro_v60_n3_b0.5.png | emiss_v60_n3_b0.5.png |
Meeting Update 04/22/2013 -- Baowei
- Tickets
- CIRC poster session
- Registration: http://www.circ.rochester.edu/poster_submission.html
- deadline, April 26th, this Friday
- Users:
- Next meeting time with LLE: training of visit and next step.
- Worked
- Optimization on stampede: https://astrobear.pas.rochester.edu/trac/astrobear/blog/bliu04042013
- New strong scaling test on stampede (O3):
- In summary:
- with level 3 optimization, stampede is about twice faster than kraken on less than 1000 cores, which is fair — 1 SUs on stampede = 2 SUs on kraken
- Kraken has better strong scaling up to 5000 cores. (The maximum for stampede normal queue is 4096 cores)
- Current usage: stampede (40% | 169,343 SUs left), kraken (99% | 1,134,191 SUs left)
- Working on
- optimization and scaling on blue hive and blue streak
- AstroBEAR poster for CIRC symposium
meeting update
Hi everyone,
I was out of town last week until Wednesday, and have been focusing on some science outreach stuff in the days since. I will be giving a talk at MCC tomorrow to 3 different classes of science students at noon. This talk is going to describe astronomy, star formation, computational techniques, and astrobear. If anyone would like to attend, it is in room 8-100
I have signed up for the protostars and planets conference, but am still curious about lodging and flights. Should we go ahead and book these? Any conditions we should be looking for?
Plan to have my new coding stuff up on the web tomorrow or Tuesday.
Best wishes
Erini's Update 4/16/2013
- finished LE_module
- made a project page for it
Meeting Update 0416
High Res TSF Run:
Movie:
http://www.pas.rochester.edu/~shuleli/0331/set1cut.gif
Meeting Update 04/16
- MHD Mixed Results
- Radiation Transfer coded, ready for testing.
Meeting Update 04/16/2013 - Eddie
- My pulsed jet module is working in 3D. Now, I need to move to a larger machine and do some runs. I will try both BlueStreak and Kraken.
- I am working on generating the data and plots for the 18 models of 1D radiative shocks. There will be a blog post before the end of the week containing my version of the plots. If everything looks good, the raw data will be sent to Pat so that he can create his own plots for comparison.
- There is still a small amount of drift in the 1D shock simulations. This is not a problem for generating the data and plots since those will come from the initialization frame. However, it would be nice to quantify this drift a bit more, so I plan on making some sort of plot to characterize this. Plotting vdrift/vshock vs. vshock for a few different shock velocities makes the most sense with multiple lines for different pre-shock densities, or betas.
Meeting Update 04/16/2013 -- Baowei
- Teragrid proposal submitted
- Users
- Christine: needs Stream Object and the 2.5D,asks the progress in implementing a radiative transfer feature in astroBEAR, will come back in May.
- New one asking for the code from VSSC (Vikram Sarabhai Space Centre?)
- Tickets
- New: none
- Closed: (none)
MHD mixed results
The MHD run ran out to frame 99 = 12 Myr before beginning to throw out Nan's in the reconstruction… So I copied the 5 GB file and loaded it into Visit - and looked at the results…
First the Good
—- Here are density isocontours
![]() |
movie |
—- And here is a close up of a single core with Field lines
![]() |
movie |
—- And here is the same but with shifted velocity stream lines
![]() |
movie |
And the bad
—- The velocity shows a red/black pattern (that is brought out by the color table
![]() |
—- But the density also shows red/black fluctuations of order 1
![]() |
Glimmer of hope
However given that the field lines look ok
![]() |
—- and the velocity stream lines show that this material is not reaching the collision region
![]() |
—- And that it only seems to be on the root level
![]() |
Perhaps there is a way to salvage the run… Obviously the boundary conditions need to be modified - but perhaps some diffusion/filter could be used to smooth out the 'ambient' to allow the run to continue?
After more analysis I believe what happened was that early on, outward velocities were achieved in the ambient around the incoming flow. Since the boundary conditions were 'outflow only', the boundary conditions were extrapolated for these cells - I don't know why gravity did not reverse this effect, or why the neighboring cells were able to maintain their 'inflow'… But given the red-black pattern - i suspect hypre truncation errors were somehow amplified by the boundary conditions… However, I don't know why this was not seen in the hydro runs..
![]() |
movie |
—-
![]() |
movie |
—-
So I did three fixed grid simulations to see what was responsible
![]() |
movie |
And it looks like magnetic fields make the difference - although self-gravity seems to exacerbate the problem.
So I also tried a reflecting wall boundary condition to see if that would prevent the striations - and that will finish soon. But in the meantime…
Fixing the striations
It is high frequency noise that could potentially be filtered? Though we don't want to introduce any divergence… Another option is to add local diffusion… ie \(q_{i,j,k} = q_{i,j,k}+ G_{i,j-½,k}-G_{i,j+½,k} +H_{i,j,k-½} - H_{i,j,k+½} \) where \(G_{i,j+½,k} = \alpha_y \left ( q_{i,j,k} -q_{i,j+1,k} \right ) \) and so on…
This would give a new value of \(q_{i,j,k}=q_{i,j,k} \times \left (1 - 2\alpha_y-2\alpha_x-2\alpha_z \right ) + \alpha_x \left (q_{i-1,j,k}+q_{i+1,j,k} \right ) + \alpha_y \left (q_{i,j-1,k}+q_{i,j+1,k} \right ) + \alpha_z \left (q_{i,j,k-1}+q_{i,j,k+1} \right ) \)
This can be repeated as needed, but setting \(\alpha_x = \alpha_y = \alpha_z = 1/12\) should give reasonable results…
Now to keep the B-field divergence free, we need to calculate an 'emf' to update the B-field with… \(\mathbf{B}=\mathbf{B}+\nabla \times \mathbf{E}\). Now we want to diffuse the B-field using \(\mathbf{B}=\mathbf{B}+\alpha \nabla2 \mathbf{B} = \mathbf{B}-\alpha \nabla \times \left ( \nabla \times \mathbf{B} \right ) \) so we can identify \(E=-\alpha \nabla \times \mathbf{B} \) This can be discretized as \(Ez_{i+½,j+½,k} = -\alpha \left (By_{i+1,j+½,k}-By_{i,j+½,k} + Bx_{i+½,j,k}-Bx_{i+½,j+1,k} \right ) \) and so on…
This would give
Now if we are only worried about Bx, and smoothing in y and z we can simplify this to
This should keep the ratio of \(\rho / B_x\) constant… consistent with flux freezing.
We can also reduce the diffusion where the mixing parameter is high - or where the grid is refined… if we use a diffusion factor \(f = 1 - e{1-4 \left ( \frac{\rho_L+\rho_R}{\rho}\right )2 } \) and then set \( f = 0\) if a cell is refined. In addition we can force diffusion near the inflow boundaries by scaling \(f = \max(f, \left (\frac{r-.75}{.25} \right ) 2 ) \) where \(r \) is a parameter that goes from 0 at the midplane - to 1 at the boundaries.
Bayes theorem...
So, Bayes theorem says that
so as an example, if we knew that a jar had equal odds of having either 3 or 4 numbered jelly beans, and we drew one at random that was numbered 1, we could calculate the odds of there being 3 as opposed to 4 jelly beans.
In this case
andNow the probability of drawing the number 1 jelly bean IF there are 3 jelly beans is
and the odds of there being 3 jelly beans (as opposed to 4) are but what is ? Well, we have to add all the ways we could draw a '1'. where . So then we have.
Now
was drawn from the discrete set of events , but we could imagine a continuous set of events (\x$)]]. Then Bayes theorem reduces toSo let's say someone placed you randomly along a line that could be anywhere from 0 to L meters long with equal probability… And you found yourself at 1 meter from the start. What does the probability distribution for the length of the line look like now?. Well initially
. Now, for and otherwise. So we have
The following image shows the prior probability distribution for line lengths of 10 and 20 as well as the new probability distributions after finding oneself at 1 meter. Also shown are the 50th percentile and the average expected length. For example, if the initial distribution is between 0 and 10 meters, then the modified distribution has an average expected length of 3.90 and the length will be less than 3.1 50% of the time.
The expected value and the 50th percentile are dependent on our initial guess as to the maximum line length. For example if we assume the length could be from 0 to 20 meters instead, then the 50th percentile is at 4.5 and the expected value is at 6.3. In general the expected value will go as
and the 50th percentile will go asThe above analysis is for a distribution in which there are equal probabilities for being found at any point along the line.
, and an equal likelihood for the line being any length less than L , however we can also consider a system in which the probability of being at any given point grows exponentially with distance (or time). while still keepingFor example if you chose a random human in history, the odds of finding a human living at time t, assuming the human civilization grows exponentially for x years before ending, should be something like
In this case we have
and . So we haveThis gives an expected value of
and an . In this case the expected value is fairly insensitive to and quickly asymptotes to 2. So in an exponentially growing system, a random selection will be heavily skewed towards the end and would give the expectation that the system would end within one more timescale.So if civilizations continue to grow indefinitely, then we are likely near the end of ours… however if civilizations cap out at a fixed number of people, then our expected lifetime becomes dependent on the maximum expected civilization lifetime…
So perhaps this is an argument against colonizing other planets? Or curbing the population growth?
Meeting Update 0408
Triggered Star Formation:
Particle plane vertical slice movie
3 slice movie
clump tracer movie
wind tracer movie
Krumholz Accretion:
wrote the code to turn on Krumholz accretion when particle is formed. Mach 3.5 and 2.5 jobs resubmitted to Stampede. Results this week.
Star Formation Jamboree:
Registered and submitted the abstract. The talk (if there is one) will be on the recent triggered star formation stuff.
Meeting update - 4/8/13
- Registered for star formation jamboree and hotel — staying Sunday and Monday night — have to submit an abstract still and 1 pg introduction slide
- finished paper (new abstract, introduction, discussion sections)
- Potential BE sphere plot from previous post — on comparing new modified BE sphere from the sims
- Made some progress in reviewing next chapter in Toro, will begin coding this week..
Meeting Update 04/08/2013 - Eddie
- Did cooling length analysis on one of my 2.5D jet simulations: ehansen04042013
- Made movie of 2.5D lab jet. Shown is the total density, the aluminum (jet), the argon (ambient, and the mesh.
- This week, I want to focus on getting some 3D runs done.
Meeting Update 04/08/2013 -- Baowei
- Users
- New user from Indonesia National Institute of Aeronautics and Space: Modelling solar phenomena, especially coronal mass ejection
- Jonathan and Baowei met with the CS graduate student working on optimizing MUSCL with openMP on Friday.
- XSEDE proposal
- scaling test on kraken & stampede: Kraken is much faster with AstroBEAR (https://astrobear.pas.rochester.edu/trac/astrobear/blog/bliu04042013)
- still working on the computing resources applying: 80% on kraken?
- proposal — merge the feedback ladder proposal and the colliding flows paper together: http://www.pas.rochester.edu/~bliu/Proposals/XSEDE_proposal.pdf
- A revision runs on Kraken: http://www.pas.rochester.edu/~bliu/AstroBEAR/krakenbear.tar.gz
Thinking of a stability analysis for modified BE sphere phase of Matched case...
I am thinking about some analysis Di Francesco did and how we may adopt a similar, but potentially more correct way of doing it…
In assessing the stability of the 'modified' BE sphere in my simulation, i.e. the post-compression wave phase, where the BE sphere has seemingly re-equilibrated in a smaller volume at higher densities, I am curious if there is a better way to evaluate xi for this modified BE sphere and if it exceeds xi_crit = 6.5.
My idea was as follows… I could gather from the plots I already made the new central density of the modified sphere, and the radial extent (i.e. from flat inner core region to flat envelope region). Then, initialize a new BE sphere with this central density, and radial extent, but with different xi's. The goal would be to see if we could 'fit' the profiles. Only if the profiles were within some small margin of error, would I feel comfortable using the definition of xi and xi crit in evaluating stability as Di Francesco did…
The idea is based on this plot:
Here we see how different xi's (given same rho_c and r) are in HSE of varying slope… Steeper slopes are in unstable HSE as the top curve is the critical value of xi. We would like to try to fit one of these unstable curves to the center, over dense modified BE sphere if possible…
In attempting to do this, I am still trying to work out scaling correctly, but here is a first shot:
Obviously the pink is the data from my matched run, from the chombo that began to resemble a supercritical BE sphere… Note there is a slight bump in it still, perhaps a remnant from the compression wave phase…
Pulsed Jet Simulations and Cooling Length
I'm setting up a diagnostic variable, so that I can see what the cooling length is everywhere on the grid in my pulsed jet simulations.
I have… (cooling length) = (velocity)*(internal energy) / (cooling strength)
Where cooling strength is in units of energy/time.
This should give the value to me in computational units. My simulation has 30 cells per cu at the highest level. Say I want at least 10 cells per cooling length, so if my cooling length drops below 0.33 then I'm under-resolved.
The only thing I"m not sure about is if it's correct to use the velocity, because it will be 0 in the ambient. So the cooling length would be zero everywhere, but maybe that's ok and we just ignore the ambient?
Below is an early snapshot of the hydro run with the Raga set up. The hydro case should have the strongest cooling since magnetic fields tend to puff up cooling regions. You can see that the cooling length starts at 0 because it's in the ambient. Using this criteria the region directly behind the bow shock, and behind that first internal shock appear to be under-resolved.
Strong scaling Test -- Kraken & Stampede
- From last run on stampede(https://astrobear.pas.rochester.edu/trac/astrobear/blog/bliu03272013),Hypre doesn't affect the scaling.
- This run, use a short final time and consider the profiling time and IO time.
- stampede
Runtime | ![]() |
Non-ghost zone portion | ![]() |
- kraken
Runtime | ![]() |
Non-ghost zone portion | ![]() |
- Data
- stampede Non optimization
Cores | Wall Time | Non-ghost zone portion |
---|---|---|
128 | 549.82 | 57% |
256 | 360.35 | 48% |
512 | 235.72 | 41% |
stampede O3
Cores Wall Time Non-ghost zone portion 128 54.0 57% 256 33.5 48% 512 20.4 41% 1024 13.31 35% 2048 10.63 29% 4096 8.66 23%
- Kraken
Cores | Wall Time | Non-ghost zone portion |
---|---|---|
120 | 118.18 | 56% |
240 | 69.00 | 50% |
480 | 40.99 | 42% |
1008 | 28.24 | 35% |
2016 | 16.88 | 29% |
4996 | 11.25 | 22% |
- Configuration of Kraken & Stampede
Kraken | Stampede | |
---|---|---|
Computing Nodes | 9408 | 6400 |
Core per Node | 12 | 16 |
Processor | 2.6 GHz AMD Opteron | 2.7GHz Xeon E5-2680 (Coprocessors Xeon Phi SE10P 1.1 GHz) |
Memory per Node | 16 GB | 32 GB |
- Standard output
1.128 cores
Total Runtime = 550.3116700649261475 seconds. Info allocations = ------ 280.6 mb message allocations = ------ 36.2 mb sweep allocations = ------ 49.8 mb filling fractions = 0.012 0.644 0.900 0.000 Current efficiency = 82% 16% 98% Cell updates/second = 973 1721 57% Wall Time Remaining = ------ AMR Speed-Up Factor = 0.1039E+04
- 256 cores
Total Runtime = 360.3508758544921875 seconds. Info allocations = ------ 200.6 mb message allocations = ------ 32.4 mb sweep allocations = ------ 59.7 mb filling fractions = 0.012 0.665 0.898 0.000 Current efficiency = 77% 21% 98% Cell updates/second = 735 1525 48% Wall Time Remaining = ------ AMR Speed-Up Factor = 0.7988E+03
- 512 cores
Total Runtime = 549.8217809200286865 seconds. Info allocations = ------ 280.6 mb message allocations = ------ 36.2 mb sweep allocations = ------ 49.8 mb filling fractions = 0.012 0.644 0.900 0.000 Current efficiency = 82% 16% 98% Cell updates/second = 974 1722 57% Wall Time Remaining = ------ AMR Speed-Up Factor = 0.1040E+04
- Standard output on Kraken
- 120 cores
Total Runtime = 118.1762299537658691 seconds. Info allocations = ------ 257.6 mb message allocations = ------ 41.2 mb sweep allocations = ------ 63.1 mb filling fractions = 0.012 0.668 0.895 0.000 Current efficiency = 75% Cell updates/second = 4837 8572 56% Wall Time Remaining = ------ AMR Speed-Up Factor = 0.8210E+03
- 120 cores
- 240 cores
Total Runtime = 68.9995868206024170 seconds. Info allocations = ------ 164.1 mb message allocations = ------ 32.8 mb sweep allocations = ------ 58.4 mb filling fractions = 0.012 0.654 0.897 0.000 Current efficiency = 70% Cell updates/second = 4099 8226 50% Wall Time Remaining = ------ AMR Speed-Up Factor = 0.7070E+03
- 480
Total Runtime = 40.9853310585021973 seconds. Info allocations = ------ 122.0 mb message allocations = ------ 24.2 mb sweep allocations = ------ 30.1 mb filling fractions = 0.011 0.706 0.901 0.000 Current efficiency = 68% Cell updates/second = 3414 8082 42% Wall Time Remaining = ------ AMR Speed-Up Factor = 0.5956E+03
Meeting Update Apr.1
TSF TSF with Mach 1.5 shock:
Movie
TSF with Mach 2.5 shock:
Movie
TSF with Mach 3.6 shock:
Movie
Particle Mass Evolution:
Mach 5 shock cannot form a star.
Some questions not answered in Boss: 1. how does the timing of star formation depends on the incoming shock speed (Boss is focused on whether formation is possible under certain conditions? from what we see in the above 3 simulations, there seems to be a inverse linear correlation between shock speed and the collapse time. 2. what is the end mass of the star after a long time? There seems to be a saturation for the end speed when the incoming shock is slow (Mach 1.5 and 2.5). The end mass seems to be again inversely correlated to the speed of the incoming shock, when shock speed is not saturated. The behavior can be better characterized with more simulations in the parameter space (Mach 3 and 4). I think we can write a paper by simply characterizing these behavior and maybe device an approximate model…
Weird Visit behavior?
So when I tried to plot the contours and the 3-slice, I got strange things… In the image output, the contour does not show transparency correctly, the 3-slice outputs a yellow colored image… The plots look fine in the viewing window. Tried all versions installed on the local machine, get the same thing.
Conferences Just got the email from MU4 on the proceedings, I started revising the proceeding according to their requirements. Star formation local meeting: I'd like to do a talk on recent stuff.
Meeting Update: Monday April 1, 2013
Updates:
Possibly Useful New Tool from Rice University: HPCToolkit (CGO'11, ISPASS'13, hpctoolkit.org, installed at rice, UMich, etc.)˘
- Can break cache miss latency down to each variable.
- Latency is read at register.
- Every 64,000th instruction (randomly - on average), latency is read.
A single instruction records what happens to it (hit or miss).
With interest, I may be able to install this on one or more CRC or physics machines.
Should be able to continue working on RubyBEAR next week after midterm and paper resubmit deadline.
Meeting Update 04/01/2013 - Eddie
- Got the new 1D radiative shock module running now, but the shock is not steady. It moves to the left, which is in the direction of the preshock gas. It's not a very fast movement, so it looks like something is just a little off. I'm trying different things to see if I can get it to stay put, but the initial profile looks good. I think it's safe to run some 1D models for Pat now.
- For the 1D models, I will basically do a small subset of the original data set that Pat had sent earlier this year. The idea is to avoid doing 100+ runs and just give Pat a handful of runs to compare astrobear to the raymond shock code. So my grid will be vs = 40, 50, 60 km/s, log(n) = 2, 3, beta = 5, 1, 0.5, x = 0.01. That's a total of 18 runs, but if Pat wants to look at preionization as well, then I will add 18 more runs with x = 0.1.
- I started implementing 3D capability into my pulsed jet module. I'll be running and testing that out this week. I also need to put some more thought into the specific runs that I want to do. I see the main focus of my paper being on the effect of magnetic fields on emission in a pulsed jet. Emission maps from a 3D simulation like this have not been presented in the literature to the best of my knowledge.
- I did a 2.5D run of the lab jet for Francisco. I implemented tracers for the aluminum jet and the argon ambient, and the code uses a separate cooling table for each. The tables come from an old paper, but this is what Francisco was using to estimate the cooling strength of the experiments. I will generate some images and movies, and send Francisco an update. I also think it might be worthwhile to see if I can find better tables that span the temperature range of the simulation.
Meeting Update
Some comparison plots for Whitworth — (these are a fixed radius, over time, rho)
Includes 2 radii, r = 0.5 and r = 0.3:
Meeting Update 04/01/2013 -- Baowei
- Teragrid Proposal
- Scaling on Stampede: https://astrobear.pas.rochester.edu/trac/astrobear/blog/bliu03272013
- Scaling on Kraken hasn't finished yet.
- Current computing resources
- Stampede: 61% | 256,868 SUs left
- Kraken: 99% | 1,138,233 SUs left
- Tickets
- New: none
- Closed: none
- Working on testing box.net for storage.
Erini's Update 4/1/13
- i have made a test polytrope code, it is located in my test directory if anyone is interested
-i have tried to upload a couple of images, but embarrassingly enough I cannot get it to work on the wiki
- my dimensionless total radius matches to at least 4 decimal places to Chandrasekhar's table and current tables