Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Simulating the universe

  1. Mar 5, 2007 #1
    It is conjectured that one day it will be possible to produce a simulation of the universe with sentient beings. Obviously many problems will arise.

    One that occurs to me is that such beings would need to be able to 'see'. So how would you simulate light and sight?

    Any thoughts
    Last edited: Mar 5, 2007
  2. jcsd
  3. Mar 5, 2007 #2
  4. Mar 5, 2007 #3
    Sorry about the question mark.

    Ok so how you simulate 'sight' depends on how you simulate a 'thing' that becomes the 'image'.

    Given a world with many 'things' and many 'viewers' then how does a 'viewer 'know that a 'thing' is present.

    Does the 'viewer' see the 'thing' by an external (to the viewer) subroutine that calculates for all things and all viewers their relative position and what is viewable? OR

    Does the viewer see via objects (in the oop sense) internal to the viewer? If so how is information about the 'thing' related to the viewer? Would there be light particles (objects in the oop sense, that 'fly' around the universe from light sources)?
    Last edited: Mar 5, 2007
  5. Mar 5, 2007 #4
    there's that thing (actually a lot of them) called "ray tracer". is that what you are talking about? I'm sure many of them are in c++ and have some kind of oop architecture, so... look it up.
  6. Mar 5, 2007 #5


    User Avatar
    Science Advisor
    Homework Helper

    There are plenty of sentient beings on earth that can't 'see' (including blind humans). They seem to get along just fine. What's the problem?
  7. Mar 5, 2007 #6
    I am talking about a completely simulated universe.

    Try the question the other way round.

    Suppose our reality is just a computer simulation. How do you think they programmed light?
  8. Mar 5, 2007 #7
  9. Mar 5, 2007 #8
    Maxwell's equations describe the interrelationship between electric fields, magnetic fields, electric charge, and electric current.

    Ok so you can program the equations but what are you applying them to in the simulation. How do you program fundemental particles, or what are there methods and properties going to be?
  10. Mar 5, 2007 #9
    Now you're changing the question again. Well. People are working on it.
  11. Mar 5, 2007 #10
    Not really changing the question, just trying to get people to understand what I was actually asking in the first question
  12. Mar 5, 2007 #11
    universe simulations would either be direct simulations of a local quantum system by qubits- or a classical simulation of qubits which simulates a quantum system- such a simulation would simply perform the Schrödinger equation in a simple logic gate network as with a cellular automaton- it is thought by many that the processes which generate light/ gravity/ and all the other physical dynamics of our world would naturally emerge at higher scales from the operation of this quantum cellular automaton- [because that is how physics in our universe emerges]

    in other words you can do an accurate simulation but still not understand how the physics of the simulated universe works exactly- but a God's-eye-view would help you learn more-
    Last edited: Mar 5, 2007
  13. Mar 5, 2007 #12
    setAI- thank you for answering the question clearly. My underlying thoughts were whether in creating such a simulation how much insight could be gained into the creation of our universe. Now you have told me about qubits I have been able to read about them on the net and gained insight.
  14. Mar 5, 2007 #13


    User Avatar
    Science Advisor

    If you're just trying to reproduce our current perception of the universe you might not need to simulate as far as individual particles or quantum phenomena, certainly not all the time. The uncertainty principle could mean that the modeling system would be able approximate particle states from the information fed to the "sentient beings" only when necessary, while keeping things coherent and avoiding unnecessary computation.

    If possible we want to avoid having the CPU waste cycles in things that aren't that relevant such as what a tiny particle in front of me is doing, or what is happening on the dark side of the moon right now, until these things become necessary, and when they do just have the system guess an unimpossible state.
    Last edited: Mar 5, 2007
  15. Mar 6, 2007 #14
    Job - Would that imply that at the macro level of this simulated universe some events would only occur when observations take place. For example the position of stars would only be calculated and re-positioned when observations were taking place?

    Or because of the necessity of two observers agreeing on an observable event such as the position of the stars and the predicted position of thses stars in their future is it more efficient to have them continually calculated rather than obtaining the same position through calculations when they are observed?
  16. Mar 9, 2007 #15
    As I read somewhere tiny particles can change and shift through simple observation by a person. Therefore every molecule is impossible to specifically calculate because of its constant shifting. Especially on last observation.

    As I see it...
    If you are thinking video game wise like I am in this sentient being viewing sight scenario then.. There would be a projected geometric shape such as a cone of sight widening a person's sight draw range to infinity (or the ability of such a being with programmed distance limits). Everything outside of this cone of sight would be darkness and completely invisible to the being. It still exists but has no effect on the being except yatta yatta heat ray.:surprised . I'm guessing this would be a rudimentary comp simulation at first with AI. Anything further is useless to contemplate because we will never see its implementation at the current rate of technological increase in our sludgy world. (We're slow folks.):yuck:
  17. Mar 10, 2007 #16
    ultimately all one would have to do to simulate any possible human brain state [and thus any perceivable reality/history/self of any human-level observer who could ever exist in any possible world] is simply a universal simulator of the human brain: an electrochemical simulation of the brain's neural infrastructure that is able to process roughly 10^16 operations per second on 10^15 bits- given the implications of quantum mechanics [especialy the Superposition Principle] the amount of raw information processed by an observer is the limit of coherent information about any physical system that is 'real' to an observer anyway- [and the complexity of the observer's world- the amount of information that is 'real'- is limited by the beckenstein bound of the observer's local environment] so obvioulsy huge simulations of the entire 10^90 bits of the observable universe are not required-
    Last edited: Mar 10, 2007
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook