How to build the universe in just 20 days (news)

Area two-thirds of the universe analysed
Purpose was to predict how it will change
Study structures to old, distant to see



THE largest ever simulation of the birth and evolution of the universe has been run by the world's 26th most powerful computer.

Researchers at the Korea Institute for Advanced Study in Seoul used the Tachyon II supercomputer for the task, which has 157,392GB of disk space and over 26,232 processing cores - and it still needed 20 days to complete the task, the Daily Mail reported.
In the end the computer analysed an area around two thirds the size of the observable universe, made up of around 374 billion particles.
Researchers said the purpose of the study - called Horizon Run 3 - was to see if their virtual universe ended up with the same properties as the one we see around us, including galactic clusters and super-clusters, the Daily Mail reported.
By studying this virtual version, astronomers are better able to predict how the universe will change over time.
They are also able to study in detail structures that are too old and distant to see with telescopes and matter that only exists in theory.

news.com.au



A quick check and I didn't notice this posted elsewhere around here so I thought I would show it to you guys. To be honest it speaks for itself and I have nothing to say on the subject, other than it was interesting and thought I'd share

woah thats so cool!
I didn't even know they could do this. That technological singularity its coming closer and closer

Kind of daunting don't you think?
It's so FREAKIN huge, we have to quite using the word huge cuz huge just dosen't describe it anymore.

Great find,

If you go to arxiv.org and use their search function for Horizon Run 3, It will show you the full text link for the journal article. It is well worth checking out on the basis that it is an archive of a wealth of published journal articles in many different fields of study. If you already have knowledge of the archive, please forgive.

Thanks for the heads up, this is a fascinating study.

reply to post by decussation


Thanks for that, Ill give it a read

reply to post by Violence


I'm just curious of 2 things that seemed flawed here...

1. 2/3 of the observable universe? You mean 2/3 of an ever expanding universe...so it isn't really 2/3's at all...

2. Made up of 374 Billion particles. I'm confused by this statement, is this in reference to the amount of particles used in the simulation to estimate or base their calculations on 2/3 of the observable universe? If so they are WELL off the mark because there are 133,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 atoms on earth alone (that's 133 billion billion billion billion billion) so how can this experiment be scientific when the parameters aren't even similar to the universe?

I'd take any results from this simulation with a MASSIVE pinch of salt....

reply to post by jrmcleod


I had the same questions...
and the OP link is vague and only a few paragraphs...
gonna have to dig deeper on this

Read decussation's post.

Try that link. Top pdf of that page, hard to do everything from my phone , as my computer is being annoying at the moment :-P

Im only on the first page of the 18 page pdf.

reply to post by Violence


ahhh thanks friend!

I think these articles are on the this thread's topic...

"String Theorists Simulate the Big Bang"
www.space.com...

"String theory researchers simulate big-bang on supercomputer"
www.physorg.com...

I think the supercomputer only simulated the the first fraction of a second of the universe existance, up to the start of the inflation expansion - the first "10^-36 seconds".

I like that it requires 6 extra hidden dimensions to function. I think they may find those dimensions are also the explanation for how the CERN neutrinos travel faster than light.

Researchers said the purpose of the study - called Horizon Run 3 - was to see if their virtual universe ended up with the same properties as the one we see around us, including galactic clusters and super-clusters, the Daily Mail reported

Interesting post, but if they have run this already you would think there would be a little bit of information about how closely the properties of the simulation are to our actual universe. It would be nice to see a little data instead of them just saying they did it.

reply to post by Larryman



Thanks for the extra info, much appreciated.

Originally posted by decussation
If you go to arxiv.org and use their search function for Horizon Run 3, It will show you the full text link for the journal article. It is well worth checking out on the basis that it is an archive of a wealth of published journal articles in many different fields of study. If you already have knowledge of the archive, please forgive.

I'm not sure why you didn't just post the link, but here it is:

arxiv.org...

Originally posted by jrmcleod
I'm just curious of 2 things that seemed flawed here...

1. 2/3 of the observable universe? You mean 2/3 of an ever expanding universe...so it isn't really 2/3's at all...

This gets a little complicated. We can't see 2/3 of the universe as it is today, but we sort of can see it at various points in the past, when it was smaller. This has to do with light traveling at the speed of light and acting like a time machine of sorts...the universe wasn't that big 14 billion years ago.

2. Made up of 374 Billion particles. I'm confused by this statement, is this in reference to the amount of particles used in the simulation to estimate or base their calculations on 2/3 of the observable universe? If so they are WELL off the mark because there are 133,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 atoms on earth alone (that's 133 billion billion billion billion billion) so how can this experiment be scientific when the parameters aren't even similar to the universe?

You aren't familiar with modeling apparently, but it's possible to use models which assume far fewer particles than are in real life situations and still get highly accurate results. Here is a dramatic demonstration of such a model being tested in real life:

Boeing 777 wing test

Those wings have too many atoms to include in a model, so the model actually consists of small chunks of the wing, with each chunk consisting of many many atoms. As you can see, the model is still highly accurate in spite of this. While that shows one model was accurate, it of course doesn't show all models are accurate. However it DOES show that building accurate models is possible using far fewer parts than are in the system being modeled.

The authors seem pretty confident in their model but since you can't put a the universe in a test chamber to test it like the Boeing model, we can't be as sure it's right. But we can compare the model with observations we make of the universe, and they claim it matches as quoted below.

Originally posted by isyeye
Interesting post, but if they have run this already you would think there would be a little bit of information about how closely the properties of the simulation are to our actual universe. It would be nice to see a little data instead of them just saying they did it.

They are actually saving more analysis for future papers on the topic, but they did say this much in their current paper:

we have measured power spectra, correlation
functions, mass functions and basic halo properties
with percent level accuracy, and verified that they correctly
reproduce the "Lambda"-CDM theoretical expectations,
in excellent agreement with linear perturbation theory.
For example, by computing the halo multiplicity
functions we proved that our large-volume simulations
are able to resolve correctly about six decades in mass.
Much more scientific analyses can be carried out with
these simulations, and in the upcoming papers....

They they talk about what's coming up in upcoming papers. One apparently will look at the "great wall" and how well their simulation predicts a formation such as that. It should be an interesting paper since there's been some debate about that.

Anyway the bottom line is that they claim their model is good and they do present some limited data in the paper.