Here is an idea: As human kind we discovered many laws, formulas, events, etc. in our surrounding environment.here is a way to discover the unknowns :

Now if we simulate a virtual world with our knowledge in computers we expect it to work, because we guess that we know all of it. If a bug occurs in the simulation or something acts differently from the expected in the real world, that may tell us that there is something new; a thing (law, formula, ...) that we didn't know, which causes that bug or that anomaly.

For examples imagine that we created a virtual simulation for motion dynamics and we didn't add the friction to our simulation. Everything in the simulation will work fine but in a different manner. So if we hadn't discovered friction yet, would it help us to think about if there is something we didn't add? Or if there is something we don't know?

But can we simulate the whole universe, everything we know, in one simulation? I know it will require a deadly high computational power. Something like what they did for DNA; two years of computation to analyse their pattern. can we simulate everything we know to create a virtual universe to compare it to our real universe?


closed as unclear what you're asking by John Rennie, Norbert Schuch, user36790, user108787, David Z Sep 9 '16 at 10:26

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 2
    $\begingroup$ It's called simulation and is widely used. $\endgroup$ – Jannick Sep 9 '16 at 8:03
  • $\begingroup$ i know there are CADs and other apps like them but i mean using this in some thins that never done before, for example simulating the evolution proccess or quantum dynamics, or dark matter simulation, or even simulating the whole universe to see if it act like ours or not @jannick $\endgroup$ – Mhmdrz_A Sep 9 '16 at 8:09
  • 1
    $\begingroup$ Why all that trouble? $\endgroup$ – anna v Sep 9 '16 at 10:16
  • 1
    $\begingroup$ I think that simulating "the whole universe" would be problematic because the simulation itself is part of the universe. I think that would lead to infinite recursion (and infinite computing resources needed). $\endgroup$ – James Sep 9 '16 at 13:18
  • 1
    $\begingroup$ There's two issues you will find. The first is the challenge of getting good measurements. In chaotic systems, that can be impossible, so we may never be able to prove that our simulation is right. More interestingly, that computer will not be able to simulate itself, if it uses pure logic, because you run into some really tricky mathematics that Godel explored in the early 1930s $\endgroup$ – Cort Ammon Sep 9 '16 at 14:26

If we have some theory that we think explains the universe then we test our theory by comparing it to exeriment. In many cases we can solve the equations our theory gives us by hand, however in other cases the theory is too hard and we have to use a simulation. Then we test the output from that simulation against observation to see how well the theory works.

So the answer to your question is that yes this is a useful way to test theories and indeed it is widely used. The problem with simulation is that if the predictions the simulation gives us don't match up with reality is isn't always obvious where the difference comes from i.e. what exactly is wrong with our theory. I suspect most physicists would prefer to solve the theory analytically if at all possible.

  • $\begingroup$ very well , but what you said is classical method to analyze a theory ,yep i got what do you mean , but i meant simulating all the theories we know , in 1 simulation to create a virtual universe , not 1-2 theories to see if they are true or not, absolutely this is OK way to check them , but i thinking about a deadly large scale , so if it even possible or not? $\endgroup$ – Mhmdrz_A Sep 9 '16 at 8:35
  • $\begingroup$ @Mhmdrz_A: it sounds as if you're thinking of genetic algorithms or something similar. As far as I know this isn't an idea that has been used much in physics. $\endgroup$ – John Rennie Sep 9 '16 at 9:12
  • $\begingroup$ is it possible to simulate a computer(machine) in virtual world and check if it is acting like real computers or not? $\endgroup$ – Mhmdrz_A Sep 9 '16 at 10:00
  • $\begingroup$ post had been edited $\endgroup$ – Mhmdrz_A Sep 9 '16 at 13:26

I don't know how my answer will be accepted by our community. But here is my try. First i must say that i agree with the answer given by John Rennie. Moreover i want to add a little more.

Think about the notion of 'machine learning. Where computers get the ability to learn without being explicitly programmed.

Machine learning explores the study and construction of algorithms that can learn from and make predictions on data. Such algorithms operate by building a model from example inputs in order to make data-driven predictions or decisions, rather than following strictly static program instructions.

Real world is a messy place. Where most of the problems (both classical and quantum) related to it does not have analytical solutions. One way to achieve solutions is simulating the problem in some powerful computer. On prime example is Quantum Chorodynamics (QCD)- the quantum theory of strong interactions. Whose non-peturbative extension is still illusive . Lots of people believe that, if one day computer become powerful enough, we can simulate the entire QCD and solve the problem.

Now if it happen that we are simulating a problem based on our findings from experiments or from pure reasoning. On the other hand, while analyzing the data, computer itself is learning through the process of "machine learning". Then there is a possibility that the computer itself may reveal a new feature of the theory which was not the part of its initial data input.

But i believe that this entire idea is highly speculative at this moment.

  • $\begingroup$ post had been edited. $\endgroup$ – Mhmdrz_A Sep 9 '16 at 13:41

Models may be build to predict future outcome. For example typical climate models are fed with all possible data, we know of, and scientists then try to let the simulation run into the future and hope that we added as many parameters as necessary so that the result is trustworthy - but it doesn't have to be anyways. As the climate models are examples of, uncertainty when doing predictions is huge.

if we cant predict something it tell us , that we don't know all of it yet, so maybe climate is related to other parameters that they are not added to the simulation , so we should expect new discoveries , right?

Not necessarily. Yes, maybe we are missing a fundamental correlation or formulae in our simulation. But maybe not; maybe we just haven't adjusted the parameters that are put in correctly.

If you do simulations on Newton's laws, you are quite sure that you know the laws and connections, but you might not get correct predictions anyways. Maybe you simply don't know enough about the values, even though you do know the correlations.

Simulations like climate models are usually made to fit to past data. Then as time passes and new data comes in, this is tested in the simulation. If it doesn't predict it correctly, it is adjusted until it does. Then this is repeated. In the end, you never know if you predict the future correctly, because you can never gather all climate data and parameters that have influence. But you can maybe estimate the maximum uncertainty, and then it is still useful.

But don't confuse a missing law of nature, formulae or correlation with inaccurate data and parameters.

  • $\begingroup$ Post had been edited because it was unclear. $\endgroup$ – Mhmdrz_A Sep 9 '16 at 13:06

Not the answer you're looking for? Browse other questions tagged or ask your own question.