Wednesday, April 18, 2007

Falsifying non-local realism

Preprint for this blog entry: An experimental test of non-local realism
Off-topic: Seed magazine offers a tool (PDF) for living in the 21st century, namely a basic introduction to string theory. Clifford Johnson was apparently an adviser - so he told them that a Hydrogen atom is a deuterium ion with four electrons, as Wolfgang has pointed out. ;-)
Scientific American describes a new experiment of Anton Zeilinger et al. - laser light split into entangled pairs whose polarizations are then measured in different angles - whose authors claim that it falsifies "nonlocal realism".

Figure 1: A Zeilinger. The picture is not 100% relevant but I want to avoid accusations that most physicists discussed on this blog are male. :-)

The title of the SciAm article, "Quantum Mechanics Fails Reality Checks", is unfortunate because it creates the impression that they have shown something wrong either with quantum mechanics or with the Czechs. Quite on the contrary: they have shown that quantum mechanics holds, just like the Czechs say, even in cases when some critics would like to create doubts. ;-)

The normal Bell inequalities, combined with the known EPR experiments that agree with quantum mechanics, falsify "local realism", i.e. the combination of assumptions that some values of physical quantities objectively exist before they're measured - in contradiction with the postulates of quantum mechanics - and moreover they evolve according to local laws.

They are able to falsify this combined assumption because the (wrong) assumption implies, through Bell's proof, that the measured correlations must belong to a certain interval. But quantum mechanics predicts and experiments confirm that the actual correlations are often outside this interval. Quantum mechanics allows you to get much stronger correlations or anti-correlations than any hypothetical underlying local classical theory with hidden variables.

The sane conclusion is, of course, that we must finally do what all fully sane friends of Max Born did in 1926, take quantum mechanics seriously, and abandon "realism" (I don't mean political realism which is good but quantum realism which is bad!): only probabilities may be predicted and it makes no sense to talk about the "real" values of observables of a quantum system before these values are measured. Only results of experiments have a physical meaning.

Nevertheless, some people - for reasons that look completely irrational to me - still insist that it is plausible that "realism" holds and locality is what is violated. I find these statements completely bizarre because they contradict not only the spirit of quantum mechanics but they also contradict special relativity - and not just a little bit but in a strong way.

Relativity requires that the fundamental degrees of freedom - such as quantum fields - must evolve according to local and causal laws. This statement must be true with huge accuracy: it's not only beautiful but it has been experimentally validated. Getting this result approximately and "randomly" from a starting point that fundamentally disagrees with such a final outcome is equivalent to assuming that Jesus Christ's mother was a virgin: I apologize to my Christian readers for this metaphor.

Such an assumption is equivalent to a belief in huge miracles.

On the other hand, Zeilinger et al. now argue that they have falsified a large class of "nonlocal realist" theories, too, because the measured correlations are higher even than what "nonlocal realist" theories allow. I don't quite know how they can achieve such a goal. It is clearly a theoretical goal.

I think that every sane quantum physicist can predict the result of all these experiments and there can't be any new surprises here: quantum mechanics works and physics behind all these experiments is controlled by the same simple laws that give clear predictions to every setup. On the other hand, they must be using some "improved" version of Bell's theorem that also applies to some "nonlocal realist" theories, not only "local realist" theories as the original Bell's theorem. I don't know what this hypothetical improved version of the theorem is.

Nevertheless, I happily endorse their position that the results of all these experiments make any attempt to preserve "realism" - i.e. to deny the probabilistic nature of quantum mechanics - highly contrived. The more you understand how these experiments work, the more you agree with us.

And that's the memo.


  1. Dear Luboš,

    let me note that the beautiful lady is NOT Anton Zeilinger. Instead, she is Melanie Zeilinger, my former colleague at ETH Zurich.

    Ondřej Rohlík

  2. Hi kilhor, if you read the caption, you would know that I was fully aware of it, of course. I just didn't know that she had a colleague with a Czech name meaning Andrew Croissant. ;-)

  3. This is really funny coinsidence as I happen to be your colleague too - University of West Bohemia in Pilsen.