It is an unresolved question whether the universe is discrete or continuous in its intricate quantum level structure.

See for example: Is the universe finite and discrete? How could spacetime become discretised at the Planck scale? Is time continuous or discrete?

It is often stated that it is beyond our reach to resolve this issue. See for example: Is time continuous or discrete?

Is this however really true? Consider a simple dynamical system, such as the Lorenz attractor. When you solve this system numerically it quickly becomes evident that the solutions found depend heavily on the numerical precision. The number of revolutions around one attractor point before the evolving curve moves to the other attractor point varies with numerical precision. At some point you can wonder if you are really studying general behavior rather than a near exact solution.

Would it be possible to set up an actual experiment with a highly non-linear system, exhibiting long term iteration, to show whether or not the real-world solution at some point deviates from high-precision numerical simulation?

  • 7
    $\begingroup$ IMO it is not a good idea to ask for an answer "drawing from credible and/or official sources" when the question itself represents a hazy idea that hasn't yet been developed fully. $\endgroup$
    – user4552
    Commented Oct 23, 2013 at 5:55
  • 2
    $\begingroup$ Although it can well be through the learning found in credible sources that one struggling with hazy ideas works these ideas into sounder ones. This is a deep and subtle question - almost the same as the question of "can one experimentally observe an infinity". Of course strictly speaking the answer has to be no by definition, but maybe the question of "are there any phenomena whose pithiest explanation is overwhelmingly that of continuous space rather than simply one with a very small discretisation length?" is better. You might find physics.stackexchange.com/a/9721/26076 enlightening. $\endgroup$ Commented Nov 22, 2013 at 5:42
  • $\begingroup$ in this entry physics.stackexchange.com/q/33273 which answers your question. an experiment is mentioned arxiv.org/abs/0908.1832 which pushes the limits of lorentz invariance violation, which would happen on a discrete local spacetime $\endgroup$
    – anna v
    Commented Jan 21, 2014 at 7:02
  • $\begingroup$ @BenCrowell, I don't care who or what answers this question. All I care about is soundness and completeness. Any convincing answer, whether affirmative or not, would be highly welcome. $\endgroup$ Commented Jul 24, 2014 at 18:25

2 Answers 2


I don't have an answer for the discrete/continuous question, but experiments like the one you propose won't resolve it.

What you are describing is what is often called 'sensitive dependence on initial conditions' or SDIC, and is very common, as you know. The problem with such systems is that they are really sensitive to initial conditions, and not just to the convenient subset of initial conditions that you'd like them to be sensitive to: the ones you might stand a chance of measuring. As the system evolves its state starts depending on where you are standing in the lab, then on what kind of shoes you are wearing, then on where everyone is standing, then on the position of the Mars rovers, then of the positions of all the dust in the Solar System, and so on. No simulation can take the initial conditions into account in any useful way.

This is not a joke: there is a lovely thought-experiment called 'the electron at the edge of the universe': in this experiment you are asked to consider a perfect billiard-ball classical gas in a perfect box which insulates everything but gravity. Everything is Newtonian physics. The universe outside the box is empty, except that there is a single electron 'at the edge' of it -- billions of light years away somewhere, but you don't know where it is. You know all the initial conditions apart from where this wretched electron is, and so you don't know its gravitational influence on the gas.

So the question is: how far can you predict the microstate of this system forward? About how many collisions does each particle undergo before the first particle leaves a collision 90° from where you predict? (After that point you clearly can't say anything useful.)

The answer is about 50.

  • 1
    $\begingroup$ Can you add a reference for the thought experiment you describe which would explain how the answer is obtained? $\endgroup$
    – Ruslan
    Commented Sep 2, 2016 at 16:15
  • $\begingroup$ @Ruslan Sadly, no. I heard about it in a talk on the BBC sometime in the 1980s: it made a deep impression on me at the time, but I have since failed to find any record of it, or whose experiment it was. I do search from time to time. However it is just a particularly striking example of the butterfly effect, and there are other well-known examples of this. $\endgroup$
    – user107153
    Commented Sep 2, 2016 at 23:12

Naturally, we are only operating in our current level of awareness about the universe. We cannot argue about what we don't know yet. Right now, it appears that indeed the universe is not continuous, that there is a quantum of energy, time, and so on. But as you also probably know, relativity does not agree very well with quantum mechanics, so we do not seem to know the full story. There are unification theories out there, but none of them is complete enough to be accepted as "the truth".

On the other hand, you are mixing math with physics in you argument about the Lorenz attractor. Solving such a differential equation as the Lorenz attractor is done mathematically with continuous functions. So of course, the mathematical language used in this problem does not give you any discrete effects equivalent to Planck's scale. You also talk about numerical precision: that is a computational problem that has nothing to do with the real, physical system, or the mathematical solution to it.

Finally, and going back to my first argument about our current scientific awareness of the universe, quantum mechanics tells us that the principle of uncertainty would render impossible the thought experiment you are describing. Such an experiment would require not only huge numerical precision (that is, in principle, achievable), but also near-perfect detection precision which is quantum-mechanically unfeasible.

  • 3
    $\begingroup$ Current modern physics, including QM, are completely based on underlying continuous mathematical models. You misunderstand my arguments with respect to the Lorenz system and numerical simulation. The point with the Lorenz system is that non-linear systems behave very differently depending on the precision of the underlining numerical calculation engine. If the universe is indeed performing an ongoing computation, in the local-reference frame, on a non-continuous structure, then inaccuracies could potentially be detected, even in non quantum scale systems. $\endgroup$ Commented Oct 23, 2013 at 4:41
  • 1
    $\begingroup$ You are right, pretty much all physics is explained by means of continuous mathematics. That does not mean that it is the proper thing to do at the subatomic level. It's like trying to use classical mechanics to explain the orbit of Mercury or the particles inside the LHC. I think that your argument is simply not feasible in practice due to the many complications of such an experiment as the one you are suggesting. I do not even think that one might even know what to look for in such a case. I do agree with you that it's an open question so I may be wrong, but so far that's my opinion. $\endgroup$
    – legrojan
    Commented Oct 23, 2013 at 20:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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