Slashdot Mirror
Astrophysicists Build Realistic Virtual Universe
sciencehabit writes "In the most detailed effort yet, astrophysicists and cosmologists have modeled the evolution of the universe right down to the formation of individual galaxies. The results of the mammoth computer simulation neatly match multiple astronomical observations, ranging from the distribution of galaxies in massive galaxy clusters to the amounts of neutral hydrogen gas in galaxies large and small (abstract). The findings once again neatly confirm cosmologists' standard theory of the basic ingredients of the universe and how it evolved—a result that may disappoint researchers hoping for new puzzles to solve."
I wanted to get a dollhouse for the kiddies, but a universe is even better.
Where the extra matter went and how the universe expanded faster than the speed of light, temporarily?
Because something tells me TFA is missing that bit or exaggerating in their last line about puzzles.
the big question is are entities in the simulation able to detect it is a simulation.
Choosing parameters that best simulate a model does not mean that model is correct.
42
My ism, it's full of beliefs.
I thought the same thing at first. However, assuming that the simulation implements only the very fundamental building blocks of physics at it's core, it is interesting to see that they translate to match our observations on a macro scale. Given that, it's just a matter of what else we can glean from the simulation runs that we have yet to observe IRL. These new insights don't have to be taken as truth, but rather lead us to new observations.
Umm, it is how science works? You want to check that current theories are consistent with what you observe. If they had found a large disagreement, then they would know to try something else. If they find agreement, then they know they need to look closer from both theory and observation ends for any other sources of disagreement.
We are trying to understand those things we can observe. That doesn't preclude us from trying to observe more stuff.
As a matter of fact they do. Their highly accurate model is based on the premise that "You, your ship, and everything in it die/decay/degrade beyond functionality before you get 10% of the way to the nearest star". If any sentient life happens to be orbitting that star then in a few thousand years when your remains arrive they may end up as part of a museum exhibit or Black-Ops coverup.
Without FTL the only things we could possibly trade in are knowledge and culture. And with FTL... well if Einstein was right then we'll probably be too busy altering our own history to worry much about trading with other stars.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
A more interesting thing, I think, would be to start with the universe we observe, and then run the simulation backwards to find out what initial conditions are necessary to create it. After all the equations are all fully reversible.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
Finally I'll be able to make an apple pie from scratch!
I love the part about "new questions to answer". As if "Where did the super dense mass the universe came from, come from, wouldn't be a good question to answer...
and apparently also know (without sharing) why the observed mass of the Higgs boson is so tiny even though the max energy times the fermion/boson sum should be huge. wow they have it all figured out...or they "cooked the books"
If there are no parameters for a model that allow the model to simulate reality, then the model must be incorrect.
If there are parameters for a model that allow the model to simulate reality, then the model may be correct, but may still be incorrect.
This work moves us from the first state to the second, at least when it comes to simulating rather large scale structure.
Not a sentence!
After all the equations are all fully reversible.
Most likely not in this case. I don't have access to the original article at my current location, but many other hydrodynamic cosmology models are not reversible because they result in an increase of entropy and a smoothing process that does not allow that to work backwards, which is kind of how reality is unless you are modelling every particle of gas clouds.
Meanwhile, all you ever needed to do was read Genesis to understand what really happened.
Meh, Genesis were never the same since Peter Gabriel left...
People should not be afraid of their governments - Governments should be afraid of their people.
You mean my RC plane is not really governed by regression equations? Shit, there goes my Nobel!
Table-ized A.I.
A whole lotta begatting, good times!
Table-ized A.I.
Confirmation bias.
Any predictive observations will necessarily be limited by the actual applicability of the model. A model may suggest directions to look for interesting phenomena, but it is NEVER confirmation of such. Simulations will only get you as far as your inputs. GIGO.
are a feature of some other species' universe-simulation: high
So when are they releasing it as a game?
I am anarch of all I survey.
When the parameters match observed conditions it does.
The simulation is uncanny! I noticed that there is only 1 planet with life on it!
Even without FTL, it's possible to arrive at another stellar system within your life time, if you are able to constantly accelerate (and decelerate after half the distance). Thanks to time dilation, for you, about 5 years pass, until you arrive at the next stellar system (the exact number depends on your actual acceleration). The main problem: If you go back. on Earth, millions of years have passed on Earth until you arrive.
Hint to /. editors and submitters: when talking about physics and astronomy papers, it's really helpfu to remember the existence of the arxiv, where the actual professionals go to find the papers.
http://uk.arxiv.org/abs/1405.1418
(Also, hint to commenters on cosmology articles: saying things like "simulations are pointless because they're confirmation bias" and "but they don't understand dark matter LOLOLOLOLOL" just make you look woefully ill-educated in the area, even to the level that a cursory skim of Wikipedia would give.)
This is a valid statement for any theory, not just a theory expressed in a simulation. Just because the theory works for some chosen values, it doesn't mean the theory is correct. Basicly yours is a null-statement, it doesn't yield any relevant information.
Science says you should be able to make predictions, so that the theory is valid. Matching observations in a model based on theories built to explain those same observations is circular reasoning.
This does not mean they are not on the right path, of course.
---- MISSING MISCELLANEOUS DATA SEGMENT --- [sigdash] trolololol
First you need political will. Sadly, that exists nowhere on this Earth.
Their parameters aren't simply chosen, though: most of them come from a disparate range of experimental observations, and the remainder are constrained to reasonable values. Getting experiment out with experiment in, particularly when it's a range of different experiment types in each case, is strong evidence that a model is accurate.
"The free parameters of our model are set to physically plausible values and have been adjusted within the allowed range to roughly reproduce the relation between mean stellar mass and halo mass inferred from abundance matching analysis. The resulting parameter settings have been tested on smaller-scale simulations and high-resolution zoom-in simulations of individual Milky Way-like haloes."
No kidding!!! What do you say at this point?
The observations their model makes are different from the observations used to construct the model.
No kidding!!! What do you say at this point?
Well, yes and no.
Depending on how far "the next stellar system" is from start, and your acceleration, of course.
For 1G and 5 years, you'll go about 11 light years.
Beyond that distance, you'll add extra distance very quickly - 22 light years will take you 6.2 years. 100 light years will take 9 years.
And you can manage 11+ BILLION or so light years in only 45 years.
While this does theoretically allow traveel interstellar distances within a lifetime, for practical purposes (we don't really want to burn a significant fraction of a lifetime travelling - it would be nice to arrive young enough to enjoy the sights, at least), we're talking 10-50 lightyears as the upper limit we'll be travelling that way.
Of course, if we're actually really serious about becoming an interstellar species, we'll do most of the work at small fractions of c. 10% will be enough to get to nearby systems within a couple generations, allowing large colony ships to go there to settle. And further trips will start from those colony worlds to more distant places. Which should be sufficient to colonize the entire Milky Way within a couple million years....
"I do not agree with what you say, but I will defend to the death your right to say it"
Probably a better one is "Simulation from the Big Bang results in output that looks like our universe at the galactic scale"
To suggest that this equals "Astrophysicists Build Realistic Virtual Universe" more than a touch hyperbolic.
-Styopa
Have they solved the problem with quantum theory and the big bang being mutually exclusive (other than saying the laws of physics changed somehow)? If not, there is still a really big problem to solve.
I think it would be quite fun to make a complete simulation of the events in Genesis. That'll make sure they were true!
Always seemed a bit like CS masturbation to me.
Nice troll, here's your cookie.
Next time you go to the airport think about the following. The skyscrapers you pass, the bridges you cross, the car you ride in, the multi-level car park you park in, the plane you board were all designed with CS masturbation. The fact that over the last 30yrs (about half of my life time) it's become virtually impossible to get finance for any engineering projects without first performing CS masturbation is testament to it's power and utility. Numerical integration is what these simulations are doing, and it's as valid as any other branch of mathematics for exploring the physical world. Simulation is how we find the results of solving the equations that comprise the so called "physical laws". Many (if not most) of the equations in those laws can only be solved through numerical integration since no analytical solution has been found. The laws themselves are just mathematical models that have been tested to a high level of confidence, they are no more or less "real" than the maths inside a well tested computer sim.
As an example there are no known analytical solutions to newtons laws of gravity when applied to a physical system such as the solar system (ie: the n-body problem). Since humans first started shooting rockets into space their trajectories have been planned using n-body simulations, such simulation are very accurate but not perfect, which is why we put small navigation rockets on space probes to correct it's course if it strays too far.
Isn't it a tautology that the simulations correctly reproduce the universe we would expect as they were generated from laws derived from observations of this universe?
No, simulations make predictions by solving the equations (laws) using numerical analysis. If the simulation does not match observations it could be a bug in the sim or a bug in the laws themselves (new knowledge), even better is when a sim predicts phenomena nobody has seen but is confirmed by later observations (new knowledge)
If the results that were reproduced differed then wouldn't the simulation be a poor simulation?
More often than not, yes. Things don't get interesting until the sim turns out to be correct when new observations are performed. For example the much maligned climate sims have discovered dozens of unknown phenomena that were later confirmed by observations. "polar amplification" and "stratospheric cooling" are two well documented examples.
It's not like these simulations are a true experiment.
As in they don't have a "control" and a "subject"? - Of course they don't, where do you get a "control" for a unique system such as the universe, the climate, the biosphere, the flow of molten metal in an engine block casting, a skyscraper that has yet to be built? Most practical problems in both nature and engineering have neither a control subject, or an analytical solution to the mathematical laws that (we think) govern them. In the modern world we attack those problems with a methodology known as systems science, we do that because it's track record says it works.. WP claims the field got started in the 50's but my CS degree covered the subject in depth, as such I think it can be more accurately traced back to the very first computer built during WW2, the first "real world" application of that computer was to run a numerical analysis of artillery fire to create artillery tables for use in the field, prior to this the tables were calculated by hand by thousands of people with adding machines and a note book. I very much doubt the WW2 generals and admirals would have used them if they had not provided a tangible military advantage.
What changed in the 50's was that well funded corporations and academics got to play
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
CrimsonAvenger did a great breakdown of time versus acceleration, so I'll just throw out a second factor: we can't actually pull off that kind of acceleration with any current (even speculative) fuel technology. Consider the size of the rockets used for orbital launches, just to accelerate a few thousand g at a few G for several minutes. Now increase that a few thousandfold to keep it up for a year. Except... it's even worse than that - far, far worse, because the fuel required for sustained acceleration increases geometrically (exponentially?) with the duration of acceleration - the fuel for the last minute only has to accelerate the rocket, but for the previous minute it has to also accelerate the fuel for the last minute, so it's slightly greater, and for the minute before that it needs to propel the fuel for the last *two* minutes, and so on and so forth - it snowballs *very* rapidly.
Now potentially a ramscoop + mass-energy converter might be able to maintain that acceleration after you get up to speed (I believe someone finally ran the numbers and showed that fusion wouldn't cut it - you couldn't produce enough energy to overcome the drag of your scoop - it's basically a massive parachute after all), but unless you're traveling through a pretty dense nebula that's probably going to mean you have to already be approaching lightspeed to pass through enough matter to sustain the propulsion, so you'll still have some mind-boggling initial fuel requirements.
All of which adds up to one thing: money. It might be justifiable to fund such a thing for exploration or colonization purposes, but for trade? What are you going to to trade in to justify the billions of dollars per pound shipping costs?
--- Most topics have many sides worth arguing, allow me to take one opposite you.
An excellent breakdown - just don't forget you also need fuel, though I suppose a ramscoop plus mass-energy converter would let you sustain accelerations once you got up to speed. The fuel for that initial acceleration and deceleration is still going to be pretty mind-boggling though.
I agree it's plausible for colonization or exploration purposes, but for trade? Trade needs to be cost-effective, and those shipping costs are going to be killer...
--- Most topics have many sides worth arguing, allow me to take one opposite you.
No, it's not worthwhile for trade, really. especially since no material good could possibly be worth shipping across interstellar distances. Trade, as such, would be trade in IDEAS, not things. And you can do that with a com laser.
As a completely off-topic aside: Is that dove EVER going to get laid? He's been chasing a dove hen around my backyard for over a week now....
"I do not agree with what you say, but I will defend to the death your right to say it"
And your point is?
No, we can't do that right now. Noone has suggested that we can.
Does the fact we can't do it now mean it will be forever impossible? No. Lot of things we do do now would have been described as "impossible" 100 years ago.
"I do not agree with what you say, but I will defend to the death your right to say it"
But it runs thousands of times faster than ours. Eventually they evolve intellignece, discover our universe, and break into it.
Yep, it took almost a century, several geniuses, the invention of telescopes and calculus, and years of painstaking observations, to refine the heliocentric model to a point where it outperformed the predictions of the ancient geocentric model. Now that we have space craft we can (finally?) determine which model is correct by observing the sun-earth system from an external viewpoint.
By definition there is no point in space outside the Universe, and we still can't even observe our home galaxy from the outside. For at least the last century the field of physics has been based on what Hawking has called model dependent reality.
Regardless of what Hawking thinks, you can always rely on Feynman to nail the root cause of the "problem" in layman's terms.
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
Observations use regression techniques, physical simulations do not. Two Nobel's in one day, that's gotta hurt!
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
They aren't observing, they're fudging things enough to make a simulation that matches reality. Dark matter is an example of a cheat code used to make broken equations resolve properly.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
One of the rare instance in music were a break up spawns 2 greater groups.
The Kruger Dunning explains most post on
no known material good could possibly be worth shipping across interstellar distances.
See: Dune for an example.
The Kruger Dunning explains most post on
Parameters are inputs, physical constants such as the gravitational constant. The model OUTPUTS are what you compare to observations
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
Do you just not understand what models are, how the work, and how to validate them? Is that why you just throw out that meaningless sentence?
The Kruger Dunning explains most post on
This does not mean they are not on the right path, of course
Indeed, when talking about the accuracy of any scientific theory (model) it should be noted that imperfect certainly does not imply useless.
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
I imagine that without FTL it would be cheaper to synthesize spice than ship it a few lightyears. Billions or trillions of dollars per pound? That's getting into the range where you could potentially fine-tune your best attempt at chemical synthesis by using little laser "tractor beams" to add/remove individual atoms, and still have a huge profit margin.
Now maybe if it was some new element with "miraculous" properties - like say element 543 that's the critical component of an alien antigravity drive... But I doubt we have anything physical worth shipping back.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
Gotta get that Alludium Phosdex. :)
That's the point. If the simulation differs, then something about actual natural law hasn't been properly translated into code. The interesting part is, if what we think we know _has_ been properly translated, then what we think we know may be wrong. The more common case, of course, is that the code has a glitch.
On what color pill they choose.
It's a cosmological model, the Higgs doesn't enter into it. It's gravity and hydrodynamics.
No kidding!!! What do you say at this point?
No, and you're not seeing a real problem either.
We have a lot of theories, some of which are hard to test, and which could conceivably interact with each other in unexpected ways. If we put them together in a simulation, and get results similar to reality, we have shown that the theories do in fact work together to get realistic results, and, further, no additional theories are needed to cover problems with the simulation (all of this being to the limits of the accuracy of the simulation and observations and agreement, of course). This is strong evidence that our theories of universe development are reasonably consistent and complete, and that's a useful result.
People used to check addition of long lists of numbers by using a primitive checksum called "casting out nines". If they could have been sure their addition was perfect, they wouldn't have needed to do that. Similarly, if the astrophysicists had been sure that the theories cover enough to create a universe, the simulation would have been useless.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes
No, but if you can't find parameters that make a model behave similarly to reality, that's evidence that the model is incorrect. This is science: all we can do is try to find evidence that shows that models are incorrect and try to make new models.
The success of the simulation shows that the theories are consistent and may be complete (in the sense that other theories are not required) to the limits of the simulation, observation, and agreement. This is evidence that the theories aren't incorrect and that additional theories on top are. It's nowhere near conclusive, but it's useful.
"When you have eliminated the unacceptable, whatever is left, however improbable, must be the truthiness" - Holmes