PDA

View Full Version : Discussion: Computer To Simulate Exploding Star



Fraser
2004-Apr-30, 07:44 PM
SUMMARY: Researchers from the University of Chicago are about to run the most complex simulation of a supernova ever attempted. The simulation will use 2.7 million hours of supercomputing time from the U.S. Department of Energy; computers which are used to simulate nuclear weapon explosions. Scientists think that a supernova explodes inside the core of a white dwarf star, and then expands towards the surface like an inflating balloon, but the exact stages are still unknown. The results from the simulation should be ready by summer.

What do you think about this story? Post your comments below.

isferno
2004-Apr-30, 10:01 PM
Are they trying to find out how much energy is released during the fusion into a neutron star at collapse time?

Makes me wonder how they are going to do it when a blackhole is created, fission due to share gravity? Or rebounce energy during collapse?

antoniseb
2004-Apr-30, 10:12 PM
Originally posted by isferno@Apr 30 2004, 10:01 PM
Makes me wonder how they are going to do it when a blackhole is created, fission due to share gravity? Or rebounce energy during collapse?
From the description in the article, they are only going to be modeling the type 1a supernova, in which a white dwarf accretes enough material from a companion to suddenly convert lots of Carbon and Oxygen into heavier elements very suddenly.

This one does not leave a black hole. I'm not sure it leaves much of anything except a cloud of Silicon, and a lonely former companion.

knealy
2004-Apr-30, 10:51 PM
OK, somebody explain to me what 2.7 million hours of supercomputer time means. That's 308 years. So obviously the experiment isn't going to run 308 years. How is this number calculated? Where's it come from?

isferno
2004-Apr-30, 11:04 PM
I'm sorry, but the blackhole remark was meant as future planned calculations.

As for white dwarfs, as I understood, from Iron, Neon and up, the whole process requires more and more energy input! just to have fusion and in the end might leave a neutron star or a pulsar.

With a gravitational collapse converting part of the star into a neutron star, is how I understand the sudden release of energy in a fraction of time, with probably generating enough energy output for an increased fusion of the heavier elements.
(I've tried to catch up on SN though I also guess quite a lot)

Hence my question about the energy output of the "new" neutron core.


As with blackholes, I assume that suns with a large mass passes through the stage where the core once or more times collapses, shedding material, before it finaly collapses into a blackhole.
As I can't understand the energy released from a blackhole during collapse, I "guessed" gravity might be large enough to tear atoms apart creating fission Energy instead of fussion Energy, (with fusion when the fission energy pushes its "random" way out)

om@umr.edu
2004-May-01, 03:52 AM
An interesting story.

I am reminded that two astrophysicists at the 1976 AGU meeting in Washington, DC told me they had done computer modeling of supernova explosions.

Their conclusion was that supernovae always explode isotropically, uniformily in all directions, and throughly mix the final nucleosynthesis products before hurling them outward.

The Hubble telescope, however, revealed a different picture of exploding stars.

The moral: The validity of computer output is no greater than the validity of computer input.

With kind regards,

Oliver
http://www.umr.edu/~om

isferno
2004-May-01, 08:32 AM
Originally posted by knealy@Apr 30 2004, 10:51 PM
OK, somebody explain to me what 2.7 million hours of supercomputer time means. That's 308 years. So obviously the experiment isn't going to run 308 years. How is this number calculated? Where's it come from?
the time given is measured in cpu (central processor unit) time.

so, with 308 cpu's it only takes one year, though as it concerns 4 projects with a total time of at least ~ 5,5 * 10^6 hours, the number of cpu will probably be much greater.

VanderL
2004-May-01, 09:16 AM
Their conclusion was that supernovae always explode isotropically, uniformily in all directions, and throughly mix the final nucleosynthesis products before hurling them outward.

The Hubble telescope, however, revealed a different picture of exploding stars.

The moral: The validity of computer output is no greater than the validity of computer input.

Hear, hear Oliver, give me observational science anytime.

There are many computer simulations of processes we don't know enough about. Every time I see an article showing the results of simulations of galaxy collisions or accretion disks forming planets there are so many unproven assumptions underneath the models that the results can give us the false idea that we are on the right track. Despite assurances that they must at least be partially right and give us some new constraints on the models.
Maybe it would help to plug in some ludicrous values at the start and see what comes out, it won't help, but at least it would be fun, although I don't want to think about the cost of this simulation.

Cheers.

om@umr.edu
2004-May-01, 11:54 AM
Originally posted by VanderL@May 1 2004, 09:16 AM

Their conclusion was that supernovae always explode isotropically, uniformily in all directions, and throughly mix the final nucleosynthesis products before hurling them outward.

The Hubble telescope, however, revealed a different picture of exploding stars.

The moral: The validity of computer output is no greater than the validity of computer input.

Hear, hear Oliver, give me observational science anytime.

There are many computer simulations of processes we don't know enough about. Every time I see an article showing the results of simulations of galaxy collisions or accretion disks forming planets there are so many unproven assumptions underneath the models that the results can give us the false idea that we are on the right track. Despite assurances that they must at least be partially right and give us some new constraints on the models.
Maybe it would help to plug in some ludicrous values at the start and see what comes out, it won't help, but at least it would be fun, although I don't want to think about the cost of this simulation.

Cheers.
Thanks, VanderL.

Yes, the greatest danger is that many young souls, and unfortunately even the scientists making the calculations, may mistakenly believe that "2.7 million hours of supercomputing time" must yield meaningful output.

The two astrophysicists at the 1976 AGU meeting had both fallen in this trap. They were both from Ivy League universities; one was a powerful member of many NASA committees.

With kind regards,

Oliver
http://www.umr.edu/~om

Higher Dimensions
2004-May-01, 04:21 PM
308 years of Dept. of Energy supercomputer time means that the Energy Department has thousands of supercomputers. If they have only 1000, then they have donated to this experiment 0.308 of a year, 24 hours per day. But DOE's own uses must get most of the computer time. Yet the experimenter says, I hope by summer well have all the simulations done." DOE also awarded 2.2 million hours (251 years) to two other experiments. That would total 0.308 + 0.251 = 0.559 of a year for 1000 supercomputers. My conclusion is that DOE has several thousand super-expensive supercomputers. DOE keeps the supercomputer industry alive.

om@umr.edu
2004-May-02, 01:37 PM
Yes, this represents a tremendous investment of our resources in the University of Chicago.

However, the output from 308 years of DOE-financed supercomputer time will depend on the input.

Will they assume, for example, that

1a. The parent star was spinning? or

1b. Assume angular momentum = 0 as was done in 1976?

2a. Ordinary atomic matter is compressed to nuclear dimensions during the explosion? or

2b. The explosion blows away ordinary matter revealing the neutron star already there?

3a. Interactions between neutrons are well described by textbooks on nuclear physics? or

3b. Interactions between neutrons are strongly repulsive, as recently reported in
"Neutron repulsion confirmed as energy source", Journal Fusion Energy 20 (2003) pp. 197-201?

http://www.umr.edu/~om/abstracts2003/jfe-n...-neutronrep.pdf (http://www.umr.edu/~om/abstracts2003/jfe-neutronrep.pdf)
http://www.umr.edu/~om/abstracts2003/jfe-neutronrep.ps
Etc.

Higher Dimensions is right, tax dollars flowing through DOE must be keeping the supercomputer industry alive.

With kind regards,

Oliver
http://www.umr.edu

antoniseb
2004-May-02, 02:57 PM
Originally posted by om@umr.edu@May 2 2004, 01:37 PM
"Neutron repulsion confirmed as energy source", Journal Fusion Energy 20 (2003) pp. 197-201
So I looked on the JFE website, and this paper was listed as 2001 not 2003 . BTW, guess who wrote this paper!

This paper makes an argument for Why the Iron Sun Shines. It includes a [i]revolutionary observation about simple nuclear chemistry that has so far eluded mainstream physicists.

To be fair to mainstream physicists, they have probably felt that the excess energy [reduced stability] in nuclei with odd numbers of neutrons had to do with the shell model of the nulceus, and that unpaired neutrons were freer to escape.

But, assuming the shell model of the nucleus is a load of crap, and extrapolating the behavior of the unpaired neutron in a heavy nucleus to all neutrons in a neutron star, suddenly a neutron star could provide a third to a half the energy required to make our sun shine, just by having neutrons escape into a layer of iron above the neutron star.

I wonder why Geminga [an isolated nearby neutron star] doesn't shine with a third the brightness of the sun?

om@umr.edu
2004-May-03, 01:49 PM
Thanks, antoniseb, for reading our paper

"Neutron Repulsion Confirmed as an Energy Source" Journal Fusion Energy 20 (2003) pp. 197-201
http://web.umr.edu/~om/abstracts2003/jfe-n...-neutronrep.pdf (http://web.umr.edu/~om/abstracts2003/jfe-neutronrep.pdf)

Yes, this reveals an energy source for the observed luminosity of Geminga (an isolated nearby neutron star) and the Iron Sun.

But nothing in our paper suggests "the shell model of the nucleus is a load of crap".

Please, antoniseb, stop posting such misrepresentations.

With kind regards,

Oliver
http://www.umr.edu/~om

antoniseb
2004-May-03, 03:08 PM
Originally posted by om@umr.edu@May 3 2004, 01:49 PM
Yes, this reveals an energy source for the observed luminosity of Geminga (an isolated nearby neutron star) and the Iron Sun.
This is a bit off-topic for the main thread here. Why don't you post your link to this paper in the Iron Sun discussion, and I'll repost these concerns there.

I have a few concerns about it, but would rather broach them in the IS thread.

om@umr.edu
2004-May-03, 03:35 PM
Thanks, antoniseb.

I will post the link to "Neutron repulsion confirmed as energy source", Journal Fusion Energy 20 (2003) pp. 197-201 in the Iron Sun thread when we get to the discussion of III. WHAT MAKES THE IRON SUN SHINE.

Meanwhile, please post a correction here or a defence of your suggestion that this paper implies "the shell model of the nucleus is a load of crap".

With kind regards,

Oliver
http://www.umr.edu/~om

antoniseb
2004-May-03, 03:58 PM
Originally posted by om@umr.edu@May 3 2004, 03:35 PM
Meanwhile, please post a correction here or a defence of your suggestion that this paper implies "the shell model of the nucleus is a load of crap".
Fair enough.

Your paper gives a simplistic view which entirely ignores the shell model of the nucleus, but doesn't actually say it is a load of crap in its presentation. In fact, anyone doing a search for the word 'crap' in this paper will find nothing. The paper claims to confirm that there is some neutron-neutron repulsive force that for some reason doesn't apply to neutrons that are somehow monogamously paired up.

Further, it implies that this force, which has never been seen in particle experiments, applies to unpaired neutrons in neutron stars as well.

om@umr.edu
2004-May-03, 05:38 PM
Thanks, antoniseb.

I find it hard to believe that well documented repulsive interactions between

1. electrons
2. protons, and
3. neutrons

must be "a load of crap" if electrons, protons, and neutrons pair off in atomic and nuclear orbitals - which they do.

These are not mutually exclusive.

With kind regards,

Oliver
http://www.umr.edu/~om

antoniseb
2004-May-03, 05:49 PM
Originally posted by om@umr.edu@May 3 2004, 05:38 PM
With kind regards
We should really move this over to the Iron Sun thread, as the last several posts have had nothing to do with the Computer Simulation in the story, and these comments will be lost to others who are following the Iron Sun discussion.

isferno
2004-May-03, 06:27 PM
Antoni, Oliver, Please,

I've grown up while everybody told me the world is a globe. I wouldn't know better if they had teached me that the world was flat.

You two are both discussing a topic which I have to continuesly search background info just to grasp what you two are talking about.

Whether the Iron sun with a neutron core is nonsense can be decided if you at least listen, and either implement new info and/or show where someone made a mistake.
Such discussions are never a waste of time even if it only is to remind you why things can or cannot be.

antoniseb
2004-May-03, 06:39 PM
Thanks isferno,

I'm sure Dr. Manuel and I both appreciate your support for the general thread.
Dr. Manuel is preparing a post about why the Iron Sun shines, which will go up sometime soon. I am preparing to make a more numerically oriented rebuttle, but won't be posting it until after his post. I expect that both of these will happen in the Iron Sun thread under Alternative Theories.

om@umr.edu
2004-May-03, 06:44 PM
Thanks isferno.

I agree it is time to move on.

With kind regards,

Oliver
http://www.umr.edu/~om

John
2004-May-03, 10:22 PM
IBM has been building supercomputers for the DOE for years now. They are huge multiprocessor machines with upward of 65,000 processors running in parallel.

The newest one in the works for the DOE will be able to perform 350 terraflops (that's not a typo) per second using thousands of Power4 and Power5 processors. That'll make it ten times faster than the current record holder in Japan. They switched from the Cray style super computer over 10 years ago and have been concentrating on grids and massively paralleled arrays for the last 15 years. I read an article that the next goal after the 350 terraflop machine is a 1000+ terraflop machine (quintaflop?).