Wednesday, 20 March 2013

Research lines that lead nowhere (II) to publishing in Nature: unnecessary experiments

Hi rats,

When I was a young muroid, my collaborators and I proved that, for the state estimation of pairs of coherent states of the form \ket{\alpha}\ket{\alpha*}, entangling measurements are more efficient than LOCC measurements, contrary to the case \ket{\alpha}\ket{\alpha}. The result was curious, and I was convinced that we would manage to publish it in a nice journal (at that time, PRA was a nice journal).

Then, something happened. My supervisor told me that he had contacted an experimental group which was willing to prepare pairs of coherent states and perform the optimal LOCC and entangling measurements.

I couldn’t believe my ears. An experimentalist, the superior species, wanted to test our result!! I was joyous and jubilant, because I was not fully convinced by our rigorous mathematical proof. My supervisor and I were so happy indeed, that we held hands and sang and danced together. Then Christopher Robin and Doraemon came with an apple pie, and we all had a nice meal under the shade of the Magic Oak Tree, in Sugarcandyland (Bromley South).

In the real world, however, I was confused and angry. We already knew what was going to happen. What did those experimentalists expect? That quantum mechanics was going to break in an experiment involving two coherent states, a beam splitter and homodyne measurements? That once the setup was complete, the skies would open and a deep voice would say: “thou shall not finish that experiment!”?

Well, the experiment was completed, and, behold, they measured what the theory predicted. Once more, quantum mechanics (and the world!) was saved.

This post is about futile experiments like mine which are perhaps a bit too subtle for an Ig-Nobel prize. You know what experiments I’m talking about: the kind which make you scream “for the glory of Cavendish!!” when you see them featured in the cover of renowned scientific journals. The kind which the theorists involved describe as: “…and then we performed the experiment. I’m sorry”.

Well, I’ve had enough. I won’t stay silent while promising QI theorists and experimentalists waste their talents in meaningless collaborations. Did you know that people in other fields (e.g.: organic chemistry) conduct experiments to actually advance the theory? It’s time to kick some asses.

Before starting my monthly rant, though, let me clarify what this post is not about.

In this post I’m going to discuss six experiments. I won’t criticize the theoretical results underlying these experiments (well, just one), or the technical ability and innovation of the experimentalists who carried them out. What I will argue here is rather the need to perform such experiments. So if I have happened to single out one of your papers and at any time you feel that I’m undermining your work, please come back to this paragraph and re-read it as many times as necessary.

And then be honest: do we really need more experiments like…?

1) Experimental demonstration of 2, 3, 5, 6, 8-photon entanglement.

Contrary to popular claims, we don’t have a use for generic entanglement, so most of these results have no practical application (what is the usefulness of an 8-party GHZ state!?). One could argue that entangling a vast number of particles may be theoretically impossible due to collapse theories, etc., and that it is interesting to see how far we can go. Even so, photons are a very bad candidate to look for violations of quantum mechanics; massive particles seem to me a better choice.

Where does this obsession to entangle photons come from? How many photons will have to get entangled before the topic dies out? For Christ’s sake, somebody write a paper showing how to entangle n+1 photons from n entangled photons, and stop this madness for good!

2) Experimental estimation of the dimension of classical and quantum systems.

The story begins an interesting theoretical study of the correlations generated by classical and quantum systems of dimension d in prepare-and-measure scenarios, followed by a complicated optics experiment where the authors certify dimension four. Unfortunately, certifying dimension four is not that difficult: I can do it with my balls an abacus, or a mango and a watermelon. And there's more! I can remember 9-digit phone numbers, so I can certify dimension 10^9. Don’t study quantum optics, study me!!*

*This sentence won the prestigious award Worst Pick-Up Line Ever 2004.

3) Environment-induced Sudden Death of Entanglement

This project was born dead. The authors present an experimental demonstration of entanglement sudden death for a two-qubit state subject to amplitude decay and phase damping channels. In case you’re not familiar with ESD, here’s the theory of the paper in three pictures:
The ellipsoid represents the set of separable states; the extremes of the stick, the initial and the final quantum state after repeated iteration of the quantum channel. (a) If the channel converges to a point in the boundary of the set of separable states, for certain initial states, the system will enter the ellipsoid in finite time (ESD). (b) For some others, it won't. (c) However, if the map converges to a point in the interior of the set, you will always observe ESD.
Fascinating. Let us now discuss the need for an experiment. The authors claim that “photons are a useful experimental tool for demonstrating [ESD] and, more generally, for investigating quantum channels like [the amplitude decay channel], as the decoherence mechanisms can be implemented in a controlled manner”.

Of course, this is all bullshit, because these two channels are defined mathematically, so one can perform a simple analytical study of the properties of the states which undergo such transformations (which the authors actually do). Implementing these channels in an optical scenario is not going to add any insight, just experimental errors. And as for ESD verification, an in-depth pub study of the different ways to touch an olive with a toothpick is equally revealing and much tastier.

4) Violation of Bell’s inequality in Josephson phase qubits

The CHSH inequality (there are so many Bell inequalities, why does everyone choose the same?) has been violated with photons, ions and cold atoms. So what? Violating CHSH with two yoghourt cans tied with a string is hardly surprising if you allow for locality or detection loopholes. The actual challenge is to violate local realism, i.e., to implement a loophole-free Bell test. If you’re an experimentalist with a genuine interest in nonlocality, don’t waste your time and ours with more non-conclusive games and go for the real thing once and for all.

And don’t make me speak of contextuality experiments; there the “loophole” turns into Madonna’s vagina.

5) Closed time curves via post-selection: theory and experimental demonstration

Synopsis: the authors come up with a model for quantum time machines, mathematically equivalent to teleportation with post-selection. Then, they decide to make an experiment to “test the predictions of the theory”.

OH-MY-GOD! A time travel experiment!! Our heroes travel back to 1955 in a modified DeLorean and accidentally seduce their own mothers in a thought-provoking adventure of self-discovery*.

*More concretely, the discovery that you’re inclined to practise incest.

Well… no. Rather, they perform a very expensive and time-consuming experiment of quantum teleportation with post-selection. Then they verify that, indeed, quantum teleportation gives the same predictions as their time-machine model, which by definition gives the same predictions as quantum teleportation. The paradox is therefore solved in a self-consistent way, Martin McFly’s right hand reappears and he can finally wank return to 1985.

I strongly recommend the authors to travel back in time and remove the experimental part from their letter. Not for me, or for you, but for the students. Think of them and their bleeding eyes when they read your paper!

6) An experimental test for non-local realism

Here an experiment to violate Leggett’s model of crypto-nonlocality is carried out. This experiment, as well as any other one trying to disprove Leggett’s model, is pointless: if a set of bipartite correlations p(a,b|x,y) is compatible with Leggett’s axioms, then it must correspond to the statistics generated by a two-qubit separable state (see arxiv:1303.5124). This implies that any experiment showing entanglement between two photons is a refutation of Leggett’s model. Since two-photon entanglement has been verified ad nauseam, Gröblacher et al.’s experiment was not necessary. This is a case where the theory was simply not advanced enough to embark on an experiment.

Enough blood for today. You have already seen several examples of unjustifiable waste of tax-payer’s money in pointless experiments. Yet many authors of these papers are respectable figures in QI. What is happening?

When the scientific community acts bizarrely, dig in and you will find an important journal at the bottom of the trash-bin.

Some years ago, an unhealthy paradigm of research in QI was established via journal feedback: the duty of theorists is to develop results which are experimentally testable with current technology, while experimentalists are expected to come up with ways to implement the protocols which the theorists devise.

Play by the rules of the game, and you will get rewarded: if you’re a theorist, you will get published in prestigious journals, like Nature or Science, where theoretical Physics hardly ever appears (and if it appears, it is usually in embarrassing forms). If you’re an experimentalist, you can claim that your technical achievements are actually interesting -read “practical”- for quantum information processing.

The negative side, of course, is that many theorists are limiting their theoretical research to subjects where “experimental investigations” can be carried out straightforwardly. Most worryingly, the paradigm has driven experimentalists to theorist hunting, a recent mass phenomenon that I invite you to contemplate at your next theoretical seminar: hordes of experimentalists, sitting on the back row, breathing anxiously, their claws ready to trap any theoretician who can tell them what the hell to do with their current optical setup. It is precisely this obsession to implement experimentally whatever is fashionable in theoretical circles what leads to surrealistic situations where four different groups report experimental boson sampling in the same week.

These are my final messages:

1) Experimentalists: for many of you, the real motivation is the experimental control of quantum systems. Such is a noble enterprise; be proud of your work and stop forcing QI applications into your papers.

2) Theorists: not every theoretical discovery must be complemented with an experiment. E.g.: it is possible to control where soldier crabs walk by projecting them predator shadows; hence, in theory, one can build a computer using swarms of crabs rather than electric currents. However, no serious researcher would attempt to perpetrate such a stupi-. Oh, no.

Yours truly,

The Rat