Thursday, 18 October 2012

Research lines that lead nowhere (I): Quantum Discord


Hi rats,

These days I’ve been searching in the garbage bin (i.e., arxiv) for a nice topic to chew. I didn’t have to dig deep. A preliminary search shows that only this year there have been uploaded 129 articles on quantum discord, and the number is growing as I write this post. This puts discord into the category of pandemic. The stakes are high: if we don’t act soon, quantum discord could destroy thousands of promising scientific careers, and possibly even cause the end of civilization!

But let’s go by parts.

The quantum discord of a bipartite state was first defined by Ollivier and Zurek as the difference between its original quantum mutual information and the same quantity after we perform a rank-1 projective measurement on one part. 

Now, what does that mean? Probably, nothing. But lack of motivation has never prevented investigation at international scale. And so we ended up with one more research topic that clearly goes nowhere, in the line of entanglement sudden death, NMR quantum computing and quantum game theory (ah! You don’t remember quantum game theory? Then suffer!!).

The first victims of the epidemic were the hundreds of PhD students who were forced to calculate quantum discord in the weirdest physical scenarios: ground states of inexistent spin chains, non-inertial frames, cavity QED, nuclear spins… This extraordinary proliferation of unnecessary calculations may remind some of the negativity fever we experienced in QI just seven years ago. There, the goal was to compute the negativity of every bipartite two-level system in the Universe, and their proponents nearly succeeded at this task.

[A reflexion: if you are not sure about the meaning of a quantity, is it worth calculating it?]

Nevertheless, there is a big difference between entanglement and quantum discord.
An experimental demonstration of entanglement can be quite challenging, both for the experimentalist and the theorist. Experimentally, you need a degree of control of your physical system in order to prepare the state and perform the appropriate tomographic measurements. Theoretically, you have to prove that a particular density matrix (or a family of density matrices with certain features) is out of the set of separable states, whose approximate characterization is a strongly NP-hard problem.

In contrast, as shown by Ferraro et al., the set of states with zero quantum discord has zero measure. This is very convenient from an experimental point of view, because it makes quantum tomography obsolete: any experimentalist that prepares a bipartite quantum system can immediately claim that it has non-zero quantum discord on the grounds that he was able to produce it!!

Now, you may think that the 2009 article by Ferraro et al. put a stop to discord research. That was actually the original intention of the authors (I happen to know them). However, rather than killing the bacteria, the vaccine made them stronger. Check for yourselves! The number of yearly papers on the subject has doubled since 2009. More interestingly, Ferraro et al.’s paper has now 195 citations, all from articles on quantum discord.

How can that be? I recently witnessed a talk about discord where this paper was cited. Naturally, I was curious about why someone would like to mention a paper that discredits all his research. This is the explanation I got:

“Ferraro et al.’s result shows us that discord is everywhere. Consequently, discord is at the heart of every quantum communication protocol”.

Now THAT is an argument! Let me rephrase it in a more casual way:

“Dude!! This is amazing!! There is discord between your mouth and my penis!!!”.

Unfortunately, the fact that everything has discord erases the meaning of the quality “non-zero discord”. Imagine a world where everybody is blonde. Would the word “blonde” have a meaning? Wouldn’t it be like “wireless”?
A world where everybody were blond could be very spooky.
One of the latest mutations of The Plague has the form of papers “proving” that certain separable states with non-zero discord constitute a resource in bipartite communication protocols. The truth is: that can only be true in very artificial scenarios. First, if your figure of merit depends linearly on the initial state, or is just convex, any result you may achieve with a separable state (with zero or non-zero discord) can be equalled or improved with a pure product state (of zero discord). Second, even if your figure of merit is not convex, any communication protocol with steps 1, 2, …, n that requires a separable state rho with non-zero discord at step 1 can be implemented too if, at the beginning of the protocol, both parties share a correlated classical state (of zero discord) and use the classical correlations of their zero-discord state to prepare rho locally before proceeding to step 1.

Rats, I don’t know what to do. Men greater than this rodent have tried to stop this madness, but have failed miserably. Were they too subtle in expressing their dislike for discord? Shall we throw our excrements at the speaker’s face next time we attend a talk on the topic in order to make our point clear? I can’t say that I approve this. But, should you do it, send me the video.

Ok, I’m tired. These are the take-home messages:

1) Kids: say "no" to quantum discord! It's just not worth it!

2) Discord Stus of the world: if my arguments haven't convinced you to stop spreading your disease, then at least stop calling it "quantum correlations". The term "quantum correlations" has been disputed for many years by the entanglement and nonlocality communities. The last thing we need is a new third party messing things even more.

Yours truly,

Schroedinger’s rat

81 comments:

  1. DISCLAIMER: I'm not and I have never been a discord person. It is not and has never been my field of research.

    While I enjoyed your comments on the O-W "work", some statements that you make here are bold (in the negative sense) or plainly idiotic.

    I do hate the brainless sheep crowd run and "PhD factories" probably even more than you do, but let's try to stay a bit more objective.

    1)"Now, what does that mean? Probably, nothing."

    How do you know? For me what undermines at the moment the understanding of discord (btw, I do hate this word and avoid it like a plague) is that the only operational interpretation of the quantum mutual information itself is the amount of randomness used to *destroy* all correlations. As far as I know, there is no complementary constructive interpretation as the amount of something needed to *construct* the state (like e.g. entanglement of formation), so discord cannot be seen as a "correlational black hole" just like bound entanglement. Does it mean it means nothing? I'd be carefull here. There are some more or less operational approches showing that some tasks can/cannot be perfomed with non-zero-discord: the no local broadcasting thm. and the entanglement activation come to my mind. Both those thms. are "if and only if".


    2) " Nevertheless, there is a big difference between entanglement and quantum discord.
    An experimental demonstration..."

    You are asking here for some sort of both theoretical and experimental virtuosity, a very common thing nowadays. The fact is that there are things in Nature which require an ultra-high degree of virtuosity but remain an (intellectual) masturbation and others which are plain simple yet ingenious (for example the act of procreation). So this is a purely aesthetical statement, and as such it should be served.

    3)"In contrast, as shown by Ferraro et al. ..."

    Claiming that if something is omniperesnt makes it uninteresting is plainly idiotic to me. You come from the institute dedicated to the research of light. Light is omnipresent, everyone can produce it. Does it make it uninteresting?

    4) "Consequently, discord is at the heart of every quantum communication protocol”

    This is not your statement, but a quote. Anyway I take my pleasure to point out how imbecilic and narrow minded is the omnipresent attitude (or rather religion) in the field that something becomes interesting only if it can be used for something, and better yet immediately to get another 3 high impact papers and 5 fat grants. Where electromagnetic waves thought to be usefull at its discovery? Did Einstein envision GPS while inventing the clock synchro? Who forces us to see the word "understand" through the glasses of "use"? Solo nosotros mismos...

    ReplyDelete
    Replies
    1. >1)"Now, what does that mean? Probably, nothing."
      >How do you know?

      I don’t. But it doesn’t correspond to me to prove that discord means nothing. The burden of the proof is on discord researchers.

      From the point of view of communication, I doubt that a quantity that you can make zero and again non-zero by local operations can be of any use.

      >There are some more or less operational approches showing
      >that some tasks can/cannot be perfomed with non-zero-
      >discord: the no local broadcasting thm. and the
      >entanglement activation come to my mind. Both those thms.
      >are "if and only if".

      I thought you were discussing the meaning of the quantity called discord, not its delta function. For me, the fact that some tasks are only possible with CC or CQ states does not justify research on quantum discord. Suppose that your state has discord 0.25, and therefore cannot be locally broadcast. What does that number mean, and how does it compare with a state with 0.5 discord?

      >fact is that there are things in Nature which require an
      >ultra-high degree of virtuosity but remain an
      >(intellectual) masturbation and others which are plain
      >simple yet ingenious (for example the act of procreation).
      >So this is a purely aesthetical statement, and as such it
      >should be served.

      I don’t see anything ingenious or imaginative in the definition of quantum discord. And I admit that this is an aesthetical statement.

      >Claiming that if something is omniperesnt makes it
      >uninteresting is plainly idiotic to me. You come from the
      >institute dedicated to the research of light. Light is
      >omnipresent, everyone can produce it. Does it make it
      >uninteresting?

      Researchers at the Institute of Photonic Sciences, where I got my PhD, don’t get all excited whenever they detect ANY intensity of light. Rather than focusing on the absence/presence of light, they pay attention to other quantities and qualities, e.g.: its frequency, or its potential for commercial applications.

      >Anyway I take my pleasure to point out how imbecilic and
      >narrow minded is the omnipresent attitude (or rather
      >religion) in the field that something becomes interesting
      >only if it can be used for something, and better yet

      I work in quantum foundations. How could I disagree with this statement?

      Delete
    2. > For me, the fact that some tasks are only possible with CC or CQ states does not justify
      > research on quantum discord. Suppose that your state has discord 0.25, and therefore
      > cannot be locally broadcast. What does that number mean, and how does it compare
      > with a state with 0.5 discord?

      In the same way in which the overlap between two pure states can tell you something about how well you can approximate cloning (in terms of fidelity), the non-zero value of discord (or a non-zero value of some other parameter meant to capture the non-classicality of correlations) may tell you something about how well a task like local broadcasting can be approximated.

      > I don’t see anything ingenious or imaginative in the definition of quantum discord.
      > And I admit that this is an aesthetical statement.

      There is no particular reason to focus on the original definition of discord, in the same way in which there is no particular reason to focus on _one_ entanglement measure or _one_ Bell inequality. Discord--and by this I mean the specific quantity--was introduced in a specific context.

      I do believe that the idea of studying how correlations work in the quantum world---e.g., the limits on how they can be shared/redistributed---has meaning and merit. At the same time I agree that the hype is excessive, and that the explosion of papers on discord in not justified given the present status of knowledge.

      Delete
    3. Thanks for your useful input Marco!

      >In the same way in which the overlap between two pure states can tell you something about how well you can approximate cloning (in terms of fidelity), the non-zero value of discord (or a non-zero value of some other parameter meant to capture the non-classicality of correlations) may tell you something about how well a task like local broadcasting can be approximated.

      do you know if that has this be done?


      Delete
    4. No, I am not aware of published works that do this. I am actually working on that with partial results so far. It is also a matter of defining the problem specifically. In my case I am quite interested in an information-theoretic version of the no-local broadcasting theorem---that is, cast in terms of mutual information of the broadcast copies. One of the tools that would be very welcome would be a generalized version of the celebrated Petz's theorem used, for example, in http://arxiv.org/abs/quant-ph/0304007

      Delete
    5. "Anyway I take my pleasure to point out how imbecilic and narrow minded is the omnipresent attitude (or rather religion) in the field that something becomes interesting only if it can be used for something..."

      This is good: and reminds that this civilization clearly distinguishes "Intelligence" vs. "technical Intelligence". Personally, I'm not itnerested in persuading 'technicians'...

      Delete
  2. Dear Rat,

    I couldn't agree more! In fact I recently refereed a paper for a high impact journal on quantum discord. The manuscript under revision was as bad as the almost any other paper on the subject. I took the opportunity to give the editor my opinion on discord and I feel it might be a good idea to share a slightly revised version of it here (even though I unfortunately do to share your rodent eloquence...). Here it goes:

    "for the reasons I explained above I cannot recommend publication of this manuscript at ....
    I would like to take the opportunity to express I view I have heard from many colleagues and that I also share myself concerning quantum discord.

    There seems to be a lot of hype recently in the topic of quantum discord. Discord is one of many quantities one could look at when studying quantum correlations. And indeed for almost a decade the particular choice of quantum discord as a tool to study quantum correlations had very little impact. However recently there has been an explosion of papers on the subject. Although in principle this could be a good development, with the community having recognised a deep concept that was being ignored before, I am afraid it is not the case.

    Almost all papers on quantum discord are shallow, involving only a trivial reinterpretation/restatement of well-known facts in the theory of quantum information, many times accompanied with a non-sensical "physical interpretation" (the interesting aspect related to quantum discord, which is to understand better the concept of data hiding in quantum states, is almost completely ignored).

    Moreover it has become usual to state that quantum discord is related to many topics of interest, in order to justify its further study. This is unfortunately misleading.

    For example, one frequently hears that quantum discord is responsible for quantum speed-ups in a model of quantum computation called one-clean-qubit model. This is however incorrect, as you could verify yourself by asking people who study quantum algorithms or quantum complexity theory. In (arXiv:0709.0548) quantum discord was computed numerically in a few examples of quantum computation in the one-clean-model and found to be non-zero and larger than the amount of entanglement. Then for some strange and unjustified reason the authors concluded from this that discord is the "fuel behind quantum speed-ups" in the model. This fallacy in logic has since been quoted in almost every paper on discord as a reason for studying the quantity.

    In another example, it has also been argued recently that quantum discord has a key role in quantum communication (with two papers on the subject just have appeared in Nature Physics arXiv:1203.1629, arXiv:1203.0011). Again the connection is based on very specific examples and contrived settings and is misleading, to say the least (these papers are also an example of another bad practice that should be avoided: trying to publish weak theoretical results in high impact journals by backing them up with equally weak experimental realizations). Here too I would kindly suggest that you consulted people working on quantum communication and quantum information theory. From my experience I believe the majority would share a similar view to mine.

    The reason why I am exposing these facts to you is because I think it has become a shame to the field that so many uninteresting papers on quantum discord have been given a lot of attention in high prestige publication venues. In part this is due to the common practice in the field of hyping the results. However the quantum information community is also partially responsible for not stating loudly enough that the concept does not warrant all the interest it has been given and that research in the area has been mediocre and many times scientifically misleading. Therefore I decided to write this message in order to incentivate you to look further into the question and ask more opinions about it."

    ReplyDelete
  3. All this "discord revolution" is depressing. For example CQT just released a highlight on a discord review paper:

    http://www.quantumlah.org/highlight/121129_correlations.php

    The title of the highlight is "everything you always wanted to know about quantumness". Reading just the title I was expecting to find a blank page...

    The same institution highlighted before the non-sensical works published in Nature Physics mentioned above

    http://www.quantumlah.org/highlight/120807_discord_papers.php

    and even funded a _whole_ workshop on discord!

    http://www.quantumlah.org/highlight/120119_discord.php

    I find it sad. First because it damages the reputation of CQT, which otherwise has a lot of great activity in QI. Second because it sends the wrong message to other researchers in the field that look up to the institution. I can understand why Nature, Science, etc are fooled into publishing stupid papers in our field, but it's much harder to justify why centers devoted entirely to QI would support, and even celebrate, bad research in the area. They ought to know better.

    Of course CQT is certainly not the only institution to blame...

    ReplyDelete
  4. I have to admit to the guilty pleasure I felt when reviewing arXiv:1208.2857 and writing the following report for PRL:

    -----
    This is another in the long line of papers promoting alternatives to entanglement such as discord as a panacea for all things quantum. The authors do a tedious computation on two qubits and show that, lo!, it is not entanglement but dissonance which is the key to understanding assisted optimal state discrimination (AOSD). Oh, you don't know what AOSD is? That's because it's a completely peripheral thing which utterly fails to achieve the PRL standard of broad interest and importance. In fact, it is barely even of interest within the quantum information community. The only people this paper will appeal to are other discord aficionados, who, while legion, do not constitute a sufficient community to meet the PRL broad interest criterion. The paper is probably correct, but that is hardly enough to put it over the bar. This is a clear rejection.
    -----

    And in the comments to the editor, I wrote:

    -----
    "We [the authors, in reply to the first referee,] sincerely believe that proving that quantum entanglement between the principal qubit and the ancilla in the assisted optimal state discrimination protocol is completely unnecessary is a major advance that deserves publication in PRL."

    This means that you can look forward to reviewing an experiment with two photons in a separable state where they observe non-zero dissonance. Please reject that paper as well, because I promise you it will be intrinsically boring.
    -----

    I really wish that the discord bubble would pop soon... I don't know how much more I can take.

    ReplyDelete
    Replies
    1. I may be stating the obvious but I think your report is unprofessional. Even supposing you were right from the point of view of the content, there is no possible excuse for insulting/making fun of the authors in that way.
      By the way my name is Tommaso Tufarelli

      Delete
    2. Tommaso is right. This kind of highly non-scientific attitude is actually a much bigger problem than the "discord bubble". Editors shouldn't take into account these abuses of the referral process. Instead of "guilty pleasure" and proud you should feel shame. My name is Carlos Sabín. Please add it to your black list.

      Delete
    3. To Tommaso, I don't see at all how the report makes fun of the authors. There is nothing personal at all in the report. The report only discusses the scientific merit (or lack of merit) in the paper. I think you are just bothered by the tone.

      To Carlos, what is non-scientific about the report? Yes, the style is not the typical style of excessive politeness. But the only points made in the report are all precisely relevant to whether it should be published or not. I think it is a much bigger abuse of the referral process to submit crappy papers to top-tier journals to take advantage the chance of getting a referee who doesn't know enough to say that the paper sucks and shouldn't be published. I find such bad papers in top journals to be much more insulting than an impolite referee report.

      And given that there is *nothing personal whatsoever* in the report, I find it hard to understand why you feel persecuted and wish to be added to some kind of blacklist. You are not being persecuted... bad papers are being persecuted, and rightly so.

      Delete
    4. In my opinion, a `professional report' should be written along those lines:

      - brief summary of the paper, including a brief description of the problem attacked, the result achieved and if possible a sketch of the techniques used to do so.

      - evaluation of the scientific correctness

      - evaluation of the importance of the results within the field. discussion of the possible "general interest" and potential impact worthy of PRL

      - final evaluation with suggestion for publication/changes/transfer to other journal/rejection


      If we look at your report:
      - you start with a generic rant against discord which is not specific to the paper you are reviewing
      - you don't really explain what's happening inside the paper: you just mention a "tedious calculation" and say that it is somewhat connected with AOSD which you judge uninteresting.
      - another rant about the discord community.

      You say you judge the "scientific merit" of the paper. in my opinion scientific merit= correctness+relevance

      in terms of correctness you just superficially state that the paper is "probably correct".
      in terms of relevance you don't give well argumented motivations but just those rants.

      Then, there is a general lack of politeness in the way the report is written.

      This is why I judge your report unprofessional.

      Note, I am not discussing on the issue of accepting/refusing the paper.

      Ciao,

      Delete
    5. Tommaso is right. That's how every decent referee report should be written. Kudos!

      Delete
    6. Are you guys even thinking about how every decent *paper* should be written? Especially if it is going to be submitted to the top physics journal, PRL?

      If the authors had merely written a bad paper, then they would have deserved a more polite review. But the truth is, they are purposely sending every little exercise that they complete to the top journals, trying to ride the wave of hype surrounding a trending field. (Well, at least one author...) This is patently obvious from the quoted comment to the editor. I consider this behavior at least as disrespectful as the above report.

      True, two wrongs do not make a right, and the tone of the review is impolite. And Tommaso is 100% correct that each report should include the four points that he lists. Notice, though, that the report really does address at least three of those points (not so much the first point). This report was also solicited as a "tiebreaker" third opinion, and the editor requested less detail than a full report. So while you are making valid points, it's hard to see how you aren't reacting because you take personal offense to the tone of the report.

      Bottom line: if you want better referee reports, write better papers.

      Delete
    7. Tomasso is again obviously right. Regarding this:
      "And given that there is *nothing personal whatsoever* in the report, I find it hard to understand why you feel persecuted and wish to be added to some kind of blacklist. You are not being persecuted... bad papers are being persecuted, and rightly so."
      At a different point below you justify the ridiculous tone of your report because one of the authors is a "serial offender". I don't even know who the authors are but to me it's clear that the report is personally biased and non-scientific. I won't say anything else. You can write me to my email address. But I would feel bad if you do so and you are distracted from your mission of writing good papers, oh Messiah. :-)

      Delete
    8. Tomasso is wrong.
      A referee need not dwell at length at each and all of the points he mentions. If he does not object to the correctness of the paper, it suffice to say that he accepts that it is probable correct. A referee needs to expand at length ONLY on the issues where he finds the paper deficient. The referee did a wonderful job explaining at length and depth why he thinks that the whole subject is not of broad interest. It is up to the authors to make the case why their paper in particular is of broad interest in spite of the referee general criticism.

      Delete
  5. I think that we should all instantly accept such papers with glowing reports. When the journals are overflowing with completely pointless articles they might learn their lesson, as well as the people that fund the work! All they want is hype, we may as well give it to them.

    ReplyDelete
  6. "I think that we should all instantly accept such papers with glowing reports..."

    Isn't it exactly what we are doing at the moment?

    ReplyDelete
  7. Hi, I do (also) work in the area of "quantum discord". I agree that the hype on the topic is excessive. Nonetheless I do believe there is merit in adopting a point of view that considers (the possibility) that there is more (or rather, less :-) ) than entanglement in the non-classical properties of correlations.

    I do not intend to argue about merits and shortcomings of the entire subfield. I can only assume responsibility for my body of work. If you are for any reason familiar with the latter and feel like sharing your opinion/asking questions about it, feel free to contact me (mpiani AT uwaterloo DOT ca); non-anonymous correspondents are preferred.

    ReplyDelete
    Replies
    1. Hi Marco,

      Two things:

      a) I agree with you that there may be more than entanglement in correlations. However, so far I don't see any evidence of discord being meaningful in this sense (as opposed to any other quantity that vanishes for CQ states). What I do see is a lot of people, mainly with a condensed matter or quantum optics background, struggling to compute a quantity that they don't even know what it means.

      b) If it were only for your papers, I would not have written this post.

      The Rat

      Delete
    2. Fine Miguel so rectify your statement. Their research leads nowhere, referred to the people in point (a). Not the field in itself. If you value Marco's work then you cannot state that research in the field leads nowhere. One useful service to the community would have been "what to study and what not to do about working with quantum discord" or something of the sort. Not just a complete debunking. Your point (b) makes you incoherent. There is no hint to anything to be saved in your original post. So, as such, I dislike it and disagree with it, and I don't take you seriously. (BTW You were funnier when you put Angelina Jolie and Liz Hurley on pink posters, rather than pontificating away of things you don't even try to make sense of)

      Delete
    3. Let me clarify b).

      I do think that basic research on quantum discord, as defined by Ollivier and Zurek, leads nowhere. You don't just define a quantity and then try to find a meaning for it. Usually, it's the other way round, that's why I think that this line of research is doomed.

      But I could be wrong (look at quantum conditional information), so I find it healthy that some people work on the topic. Here by "some people" I mean a small number, five or six. Problem is, we're speaking of hundreds.

      Delete
    4. Fine. then if you have time give me critical, rodent's feedback on this
      http://arxiv.org/abs/1212.2214
      as anyway by now it is clear who I am.

      Delete
    5. how are you?

      Delete
    6. Hi Miguel,

      thanks for the clarification, also in reply to Anonymous.

      I agree that there are a lot of papers written with little thought and often of poor quality altogether. This is not specific of the discord subfield, but for a number of reasons discord has fueled some of the worst practice. Believe me: a good chunk of the frustration you---and other people in the comments---express is shared also by people trying to do good work---which I think is possible---about discord and the like.

      I will publish works in the area of discord as long as I see scientific value in doing so. I am OK with being shot down if this is due to the lack of quality to my own work, but I am afraid that the reception and consideration of the latter may suffer from the bad reputation of the "discord field" itself may have. In this sense I share the worries of Anonymous, and I would appreciate care in stating a general criticism of a whole research area.

      Let me say that considering a quantumness of correlations* more general than entanglement has helped me in improving my understanding of the quantum, also in very "concrete" terms related to more standard and "respectful" topics. E.g., without the thinking it went into http://arxiv.org/abs/0707.0848 and http://arxiv.org/abs/0901.1280 there would not have been something like http://arxiv.org/abs/0904.2705

      One more thing: reflecting on the "discord bubble" may help us improve the way science works, e.g., the refereeing process or else. That it has been possible for a "discord bubble" (http://dabacon.org/pontiff/?p=6776) to grow is to a large extent independent of the specific topic. It would be great to find ways to reduce the risk of such bubbles. In particular I would like to avoid reaching the point where I risk being hit by excrements the next time I give a talk about topics close to discord :-)

      Thanks for triggering an interesting discussion. I may not agree on the general tone of the blog (in particular I believe that Stephanie and Jonathan's work deserved a better treatment), but it is good to have places where to discuss some of the issues affecting the community of quantum information---and science in general. BTW, if you do not know it, you may want to have a look at Valerio's blog: http://spreadquantum.wordpress.com/

      *Any concrete suggestion on what to call the kind of general quantumness of correlations that discord and the like try to capture? Is the addition of "general" good enough?

      Delete
  8. this discussion, the rat's arguments, and the replies are some of the lowest expression of modern science. not discord, but this kind of hatred. when did theoretical physics become a bunch of haterz gonna hate, fanboys etc? people bragging about how they reject papers etc. there is a huge percentage of crap out there on discord, guaranteed. but the statements in this post are flawed. about the fact that discord is everywhere, then it is not important. ok restrict to pure states. then entanglement is everywhere, but presence and amount of entanglement still matter. any nonzero amount of entanglement in pure states allows a better than classical teleportation fidelity. so no dick and mouth criticism there? look, it is even more suprising to see such a community of anti-discordians rather than the discord pandemic itself. is hating discord going to be a status symbol, like apple vs android or stuff like that? i wish all the people who posted up here put the necessary effort and not prejudice in reviewing papers. it is not ethic to reject papers just because they are on discord. i agree that they were perhaps really not good papers, in that case it's fine, but this happens all the time and one does not usually goes proudly to copy his reports away anonymously. i praise marco up here who put his name and his face on the discussion. who would have a similar courage? lay down your names, at least if somebody writes a paper with a foundational advance or a real breakthrough, which involves discord, then that author would ask the editor not to consider the people up there as referees as their lack of professionality and impartiality is evident. i invite marco if he ever gets contacted about his works (which by the way are mentioned in a reply as examples of good practice, about the no-local-broadcasting, and about the activation, among all) to copy here the factual discussion. who feels that entanglement and nonlocality are too "purist" and perfect things for anyone ever to challenge beyond, is just showing a narrow mind. you stopped at a layer of quantumness which is not the deepest one. and you are missing a lot. nobody knows what and how much, but just for this, why stop questioning?

    ReplyDelete
    Replies
    1. I welcome your comment, as it would be nice to start a more
      constructive discussion on discord here.

      > people bragging about how they reject papers etc.

      generally is a positive thing for people to state what they think
      about scientific ideas publicly. I don't think the intention was to
      brag about rejecting bad papers (which in itself is not a bad atitude
      in my opinion), but to use previously written arguments against the
      current research on discord. Rat's arguments and the subsequent ones
      touched on the following:

      (1) Since discord is only zero for cq states, it's hard to think it
      might have any more fundamental meaning
      (2) the significance of discord behind potential speed-ups in DQC1 is flawed
      (3) all operational interpretations given so far for discord are
      misleading as they
      consider very contrived settings
      (4) most of current research on discord consists of computing it in
      every imaginable
      situation, which does not lead anywhere (the same being true for a large body of
      research done on entanglement a few years ago)

      Do you disagree with some of these points? I'd be happy to hear your
      counterarguments!

      > i praise marco up here who put his name and his face on the discussion. who would have a >similar courage?

      Miguel Navascues

      >lay down your names, at least if somebody writes a paper with a foundational advance or a >real breakthrough, which involves discord, then...

      >you stopped at a layer of quantumness which is not the deepest one. and you are missing a >lot.

      Could you please give some references on discord where there were
      foundational advances or a real breakthrough? My dislike for discord
      might well be due to my own ignorance, so I'd appreciate knowing what
      papers people in the area consider good, as the most famous ones might
      not make justice to the subject. It would also be extremely useful to
      me, and perhaps to other readers, if you could quickly say why you
      think they are good papers. I understand you might not want to do it
      or not have the time, but I think it would be a very good service to
      the discord community and I would really appreciate it.

      Delete
    2. Thanks. I was more referring to highlighting "guilty pleasures" in a referee's job, which I found inappropriate.

      About the list of arguments, I find (1) flawed. As I said before, if you restrict to pure states, then the same thing applies to entanglement (entanglement=discord there). Only product states, a null measure set among pure states, have zero entanglement. Yet having entanglement, and its amount, are meaningful. Any nonzero amount of entanglement provides e.g. a better than classical teleportation fidelity, etc. All entangled states are useful etc etc. Even if in pure states entanglement is everywhere. So I don't see why this becomes a problem for mixed states. Just that entanglement needs to be replaced by discord.

      Point (2): by the way it is constructed, it is clear that DQC1 relies on the fact that measuring one subsystem affects, or reveals info about, the other subsystem(s). So it is almost intuitive that this phenomenon makes the protocol work, or, better, if this does not happen (read: the states are CQ) then the protocol cannot work. However, it is true that there is no quantitative statement which holds. the question is, do we have a "figure of merit" for a speedup? we don't even know if classical algorithms can be found which calculate the trace efficently. same thing as for factoring etc. so unlike contexts such as quantum communication, one cannot connect the amount of discord to the performance of the protocol. and I fully agree that quoting Datta, Caves et al as proving that discord is a resource for quantum computation is incorrect.

      3) I don't find all of them misleading. I find the thermodynamical interpretations of discord (please read discord as the concept of quantumness of correlations, not just the measure by Zurek) insightful. The ones in terms of communication protocols such as state merging etc, those I find artificial or contrived. Especially the one for geometric discord in terms of remote state preparation, we know that geometric discord is flawed so that protocol has been made flawed as well, since one aims at preparing a pure state without correcting for the mixedness in the shared resource.. so yes more or less I agree that the operational interpretations have, typically, followed this pattern: I take a quantum info setting which works well with entanglement, and then add limitations and constraints until i make it seem that it works with discord. If this is the case, it's a paper to reject. We agree. But I find other approaches to understand discord insightful. We can debate if you call them operational or not, but I am thinking to the papers by Streltsov et al and Piani et al on "activation", alias discord as being equal to entanglement created in a premeasurement. That picture is very general. It also provides a clear dominance relation, as all measures of discord defined through that picture are bigger than corresponding measures of entanglement. So that one sees that they quantify a more general notion of quantumness in the correlations, which reduces to entanglement on pure states. The links are (for the arxiv versions)
      http://arxiv.org/abs/1012.4903
      http://arxiv.org/abs/1103.4032
      http://arxiv.org/abs/1110.2530
      I would appreciate your opinion on those.

      (4) Here nobody can disagree. It is shameful. especially the countless papers on discord in some dynamical evolution. But what is the real motivation behind this blog post? are we going to give it a go at saving this line of research, or do we want to just give up as it is at a too advanced status of disease? I hope I am doing my share in addressing deeper questions and avoiding just overloading the crap in the literature. But if this general opinion of dislike remains among respectable colleagues, like you surely are, it does not help at all. There is the risk that people just reject good papers because they mention discord, so rejecting gives guilty pleasure.

      [continues]

      Delete
    3. You ask for references with foundational advances or real breakthrough. First I'd like to ask you how many papers published, say, on PRL, about entanglement and/or nonlocality really contain what you would qualify as such advances or breakthrough. You will have to agree with me that we see a lot of mediocre papers, or quite technical, still accepted there. But this makes no scandal. I am sad about those too, and I am sad if a paper makes it to PRL just because it is felt that discord is fashionable (actually for what I read above it is more fashionable to dismiss it). Do you then dislike entanglement?

      Anyway, the papers mentioned above, I think they give foundational advances.

      Then about breakthrough, there are many levels of breakthrough. If you think Wootters' paper on EOF for two qubits was a breakthrough, just because it calculated a closed formula, then there are very few papers which calculate neat closed formulas for discord. I would not call them a breakthrough but they get to come handy in the long term.

      What we are lacking is a truly general setting where discord shines on its own. True! one should stop thinking about what we already know in QI and add imperfections and modifications to force discord to appear. One should take different points of view. Miguel also criticises whether something you make nonzero and then zero again with local operations makes sense. Well what about squeezing then? I agree there is also a matter of semantics, maybe discord is a notion of local quantumness of subsystems, as discussed in the papers by Oppenheim etc., rather than quantum correlations, which might be inappropriate (as measures of correlations, most people would argue, should decrease under local operations on both parties). But again, if they dominate entanglement, and the difference from total correlations can be understood as classical ones, how would you call them?

      Anyway settings where discord makes sense without "forcing" it exist. Locking, Data hiding, are some of them. I am sure there are more. If you want we enter into personal contact and try to share our research interests and we might come with an idea. Very recently, I think there was some attention to the role of discord, e.g. in metrology. Maybe this is a possibly novel advance, what do you think?
      http://arxiv.org/abs/1212.2214

      Now let me end with a question: what is your understanding of discord? do you accept it as a concept, is it clear what it means? not the measure, but the fact that states are not entirely classical, and in what respect this manifests, and what it implies? Perhaps something as basic as this should be written here somewhere. If Miguel does not understand it, why everybody should just not try? Tell me what would be your way to explain discord, to yourself or to somebody else, and I can tell you maybe other ways which might be more insightful.

      Delete
    4. >Point (2): by the way it is constructed, it is clear that DQC1 relies on the fact that measuring one >subsystem affects, or reveals info about, the other subsystem(s). So it is almost intuitive that this >phenomenon makes the protocol work, or, better, if this does not happen (read: the states are CQ) >then the protocol cannot work. However, it is true that there is no quantitative statement which holds. >the question is, do we have a "figure of merit" for a speedup? we don't even know if classical >algorithms can be found which calculate the trace efficently. same thing as for factoring etc. so unlike >contexts such as quantum communication, one cannot connect the amount of discord to the >performance of the protocol. and I fully agree that quoting Datta, Caves et al as proving that discord is >a resource for quantum computation is incorrect.

      I think there _is_ a meaningful way of studying the connection of discord and speed-ups in DQC1. One could show, using some suitable chosen way of measuring discord, that unless there is a large amount of discord in the history states of the quantum computation, the computation can be simulated efficiently classically. This was done for entanglement in the case of pure state quantum computing (quant-ph/0301063) and gave a better understanding of the role of entanglement in quantum speed-ups: while it does not make sense to say that entanglement is behind speed-ups, we now know it is a necessary ingredient; if something like this could be done for discord, it would be a nice result. Unfortunately this question is not even brought up in the recent literature and in fact the most serious paper in this direction I am aware of (arXiv:1006.4402), has been completely ignored by the discord community.

      >So that one sees that they quantify a more general notion >of quantumness in the correlations, which >reduces to entanglement on pure states. The links are (for >the arxiv versions)
      >http://arxiv.org/abs/1012.4903
      >http://arxiv.org/abs/1103.4032
      >http://arxiv.org/abs/1110.2530
      >I would appreciate your opinion on those.

      I agree they are better papers than the others on operational interpretations. However why is the entanglement produced between the state and the measurement apparatus such an interesting thing to look at? To me this seems again in the line of research of trying by all means to find an operational interpretation for discord. Also I have a problem with the technical content of the papers. Since the problem is not of such great interest, it doesn’t help that the results could be derived in half and hour by anyone with basic training on quantum information theory… (1103.4032 is stronger technically, but I find the setting considered more contrived). I might have been overly criitical here. They are good papers, but are just not strong enough to be considered breakthroughs or to serve as a seed for a large amount of further study on the matter.

      >(4) Here nobody can disagree. It is shameful. especially the countless papers on discord in some >dynamical evolution. But what is the real motivation behind this blog post?

      I have no idea on the real motivation. I liked the post a lot as it’s a shout out of the unsatisfaction of many people I know on the boring research being done on discord and all the hype accompanying it.

      Delete
    5. continuation...

      >are we going to give it a go at saving this line of research, or do we want to just give up as it is at a too >advanced status of disease?
      >I hope I am doing my share in addressing deeper questions and avoiding just overloading the >crap in >the literature. But if this general opinion of dislike remains among respectable colleagues, like >you >surely are, it does not help at all. There is the risk that people just reject good papers because they >mention discord, so rejecting gives guilty pleasure.

      I think there _is_ a real risk of rejecting a good paper just because it mentions discord, but the risk is incomparably lower than the risk of accepting a bad paper just because it mentions discord. And this is the main problem that Rat’s post is addressing.

      >You ask for references with foundational advances or real breakthrough. First I'd like to ask you how many papers published, say, on PRL, about entanglement and/or nonlocality really contain what you would qualify as such advances or breakthrough.

      Just looking at the references of quant-ph/0702225 I’d count more than 50 references as breakthroughs and more than 100 as real significant advances. Looking at 1112.6238 I’d count less than 50 references as significant advances, of which more than 45 are directly about entanglement. In fact I cannot think of a single paper on quantumness that I would consider a breakthrough.

      >You will have to agree with me that we see a lot of mediocre papers, or quite technical, still accepted there. But this makes no scandal. I am sad about those too, and I am sad if a paper makes it to PRL just because it is felt that discord is fashionable (actually for what I read above it is more fashionable to dismiss it). Do you then dislike entanglement?

      I agree with you. I do dislike large parts of research on entanglement, but I still find it the most important aspect of quantum mechanics. On the other hand, while there are a few papers which are interesting about discord, it’s hard to justify having an entirely area around it just because of them. In my view entanglement is certainly a much more important, interesting, and richer concept than more general forms of quantumness.

      >Then about breakthrough, there are many levels of breakthrough. If you think Wootters' paper on EOF for two qubits was a breakthrough, just because it calculated a closed formula, then there are very few papers which calculate neat closed formulas for discord. I would not call them a breakthrough but they get to come handy in the long term.

      It’s debatable whether Wootter’s paper is a breakthrough, though it’s certainly very nice. There is no paper on an analytical formula for 2 qubit discord. That would be an interesting technical advance, though it would open a Pandora box for the calculation of discord in another infinity of meaningless physical systems. What is known is closed formulas for special classes of states; i wouldn’t call these papers neat results.

      >Anyway settings where discord makes sense without "forcing" it exist. Locking, Data hiding, are some of them.

      I agree it’s interesting to study locking and data hiding more. But I don’t think the recent literature on discord has contributed anything in this respect.

      >I am sure there are more. If you want we enter into personal contact and try to share our research interests and we might come with an idea. Very recently, I think there was some attention to the role of discord, e.g. in metrology. Maybe this is a possibly novel advance, what do you think?
      http://arxiv.org/abs/1212.2214

      Thanks, it has been nice to discuss with you, but I’m not confortable yet going public… I don’t know the content of the paper you sent.

      >Now let me end with a question: what is your understanding of discord? do you accept it as a concept, is it clear what it means?

      It’s clear to me what it means, but thanks for offering further explanations!

      Delete
    6. fine, just few notes. eastin's paper is not ignored, as far as I know. i just happen not to work on quantum computation so it is not directly relevant for me, but people do acknowledge him. i believe that a mixed jozsa's theorem with discord would be the thing to aim for, as you say, it is just not done yet.

      about closed formulas, i don't believe it will be possible to calculate discord for two qubits in closed form. there is a formula for all two-mode gaussian states. then the last paper i pointed above contains a measure akin to discord, but computable for all two-qubit states. you don't know the content of that paper: give it a go then! perhaps you'll get inspired. it might be the breakthrough you were looking for.

      and finally a general statement.

      the world is quantum mechanical as far as we know, yet sometimes we don't reveal its quantum features and the objects under our investigation are as well modelled with classical probability theory. the question is what it takes to detect unambiguously the quantumness of our system.

      consider a bipartite state.
      if it violates a bell inequality (i.e. there is nonlocality) then you'd say it is definitely non classically accountable for.
      ok but then there are some states which do not violate a bell inequality. then perhaps the states are entangled, and again this is a fundamental departure from classical physics as you say. (werner states are examples of entangled states which do not violate any bell inequality) so again you say the state is non classical.
      but even states which are not entangled, actually almost all of them as we all know, do not comply with classical probability theory. so you see those features which point again at a departure from classical description of reality, and then you have to conclude that all states with nonzero discord also are nonclassical.

      so there is a hierarchy there, between nonlocality, entanglement, and discord. you might see nonlocality as the strongest form of nonclassicality, and this weakens all the way down to discord. but you can see discord as the most general manifestation of nonclassicality, and this narrows up to nonlocality. entanglement is in the middle. why entanglement is interesting and the most important? or why entanglement and nonlocality are deserving of attention, but discord has to be useless or bullshit? thank god all quantum states, almost all, are really quantum and not just useless classical junk. why should we waste so much of the hilbert space, when still quantumness (e.g. coherence) happens in there even if there is no entanglement?

      it took 30 years to understand that nonlocality is physical and not just a puzzle (with bell's theorem), ten-twenty more years to understand that entanglement can be exploited for information processing (before it was just a philosophical paradoxical issue, spooky etc. much at the same level as what we now tag discord, just without the hype!!). now you might acknowledge that it takes a bit of time to realize how to make the best use of this more general, and if you want weaker, or more elusive, form of quantumness which happens in states with no nonlocality, with no entanglement, but with discord.

      The way the field is progressing, we are not getting close to this, unfortunately, and we all know. but the scope is broader, not narrower, than entanglement. whatever is found for discord, includes all the known results for entanglement, not the converse. just think about it ok?

      Delete
    7. >fine, just few notes. eastin's paper is not ignored, as far as I know. i just happen not to work on quantum computation so it is not directly relevant for me, but people do acknowledge him. i believe that a mixed jozsa's theorem with discord would be the thing to aim for, as you say, it is just not done yet.

      Well Eastin paper has 2 citations… Data et al had more than 100 citations in the same time period. I call it being ignored. What’s jozsa's theorem?

      >about closed formulas, i don't believe it will be possible to calculate discord for two qubits in closed form. there is a formula for all two-mode gaussian states. then the last paper i pointed above contains a measure akin to discord, but computable for all two-qubit states. you don't know the content of that paper: give it a go then! perhaps you'll get inspired. it might be the breakthrough you were looking for.

      As I said I don’t consider even Wootters’ paper a breakthrough, so I don’t think I will consider the analytical expression for a variant of discord a breakthrough. However I didn’t read the paper and it may well be a nice result. I tried to read the intro and found very confusing, but I’ll try harder.

      >The way the field is progressing, we are not getting close to this, unfortunately, and we all know. but the scope is broader, not narrower, than entanglement. whatever is found for discord, includes all the known results for entanglement, not the converse. just think about it ok?

      I like a lot some of the research on non-locality, but I consider it a study of an aspect of entanglement. It seems to me that you see the study of entanglement as the study of distinguishing separable from non-separable states. I consider it much more broadly. There are several aspects of entanglement, e.g. separable versus non-separable, non-locality, distillable entanglement, entanglement as a barrier in describing physical systems more efficiently, etc. All of them are interesting aspects of the same concept.

      I don’t buy your point that since discord is more general any result on entanglement is also a result on discord and therefore the latter is more relevant. Define a new measure on Hermitian matrices called Arbitrariness. It’s equal to discord on quantum states and equals to your function of choice on non quantum states. I claim everyone should be studying Arbitrariness, and citing me for that, since clearly everything being done on discord is just a special case of the deeper study of Arbitrariness; I could definitely use the extra citations to get a job!

      For me the aspects of quantumness that are not already covered by study on entanglement are marginal, and although a few of them might warrant further study (as the Rat suggested by a group of perhaps 6 people), they are certainly not on the same level of interest, and richness, as the many aspects of entanglement

      Delete
    8. thanks. these are personal opinions and it is fortunate enough that everyone has their own. i have worked ten years on entanglement, read zurek's paper when it came out and did not understand it, took me 4 years between 2007 and 2011 to understand the notion of quantumness of correlations beyond entanglement, now i like it and see it as deeper than just entanglement itself. nothing against entanglement but it is not the ultimate frontier in my opinion. it matters a lot, it is completely necessary in most current protocols, it has a lot of foundational value. but it is not the end of the story in my opinion. the extras are not marginal, to me. but we can reassess this only maybe in a decade. i don't really see the point in establishing a hierarchy of importance between nonlocality, entanglement and discord. interest, richness, and diversity in aspects of entanglement did not come from day one, we constructed it, we as the community, providing advances and breakthroughs. let us give us time and see whether we do the same for more general notions of quantumness.

      what i find completely uninteresting, for instance, is all the study of nonlocal PR boxes, postquantum theories etc, all things which are not physical. but within the realm of quantum mechanics, i'd not feel right, now that i know there is more than entanglement, to rank that as b-class quantumness. again personal feeling. i feel i have a more comprehensive understanding of quantum mechanics since the moment i realized that in composite system i do not need entanglement to reveal relevant, and distinctive quantum features. it's like having more sensitive glasses. certainly entanglement is still a fat red dot, but i even see in a quantum shape, maybe foggy, belonging to discordant separable states.

      jozsa's theorem: http://arxiv.org/abs/quant-ph/0201143v2

      about the confusing intro: that paper looks at something more fundamental, and then bumps into discord. it is not the useless study let us define a quantity and then try to make sense of it. it goes the other way around, for this reason it warrants attention i believe. let me know if you find it interesting or have questions about it.

      thanks for these discussions, i am disconnecting now. i respect your opinion. i hope those 6 people, not necessarily including myself, might surprise you in the future. surely you need an open mind to have the predisposition for that. five years ago i would have concluded like you did in your post. now i know better. i don't think i became a worse scientist. i only find it disappointing that i have to fight sometimes to defend my work as Rats and superficial readings of discussions like these ones (not your posts) as well as superficial readings of papers etc and word of mouth, have sealed the scarlet letter of shame on discord. i have sent less deserving papers to prl in the past and never got harsh and unprofessional reports like those above or like some i got from manifest discord-haters.

      Delete
    9. Thanks for the discussion. You are certainly entitled to work on whatever you find interesting, and I hope I'm a wrong and that you find great things about discord!

      If you are passionate about discord, and want to protect the field from generalised rage against it, why you don't start making your contribution to clarifying what's interesting and what's not in the field? For example in 1212.2214, which I'm assuming you're the senior author, why mention DQC1 if you agreed it's not interesting? Same goes true for e.g. 1208.5964, 1110.5083. In my opinion this is contributing to the hype in the subject, which you seem to agree is detrimental to it?

      Delete
    10. @Anonymous 21 December 2012 11:46,

      As the author of the "guilty pleasure" report, let me clarify my intention. I did not recommend rejection on the grounds that they were studying discord (or dissonance, in this case), but rather because it follows the same formula that we all seem to dislike in this thread, namely: compute a quantity on two qubits, declare it to have a meaning which is deep and interesting, submit to PRL/Nat Phys. My recommendation was purely on the merits of the paper. The tone, while harsh, is also appropriate in my opinion, because at least one of the authors is a serial offender in this regard. My "guilty pleasure" was just that of Schadenfreude, seeing the downfall of my enemy. (My enemy is bad papers in reputable journals; I have nothing against the authors personally, and I will only ever consider papers on their merits.)

      Let me also add that, while I agree with Miguel and others that discord is likely to be a dead-end, I would love to be wrong about this.

      Delete
    11. hi thanks for these replies. i am not convinced whether to find overall this post and some stupid replies (like the "full of shit" below) amusing, offensive, superficial, or anyway containing something constructive. i believe the debate is better be settled in scientific venues and anyway i don't intend to answer more on this blog. just to answer the one above the one above this post. (!) I and my group are doing our best to contribute good physics and not overhyping the field. "mentioning DQC1" does not mean overhyping. and i never said DQC1 is not interesting. it is a good piece of science, a nonuniversal protocol which still does some particular quantum computation better than any classical processor, and uses highly mixed states, conventionally thought to be rubbish. so it is a starting point to investigate what really is needed to achieve better than classical computation, if, when dealing with mixed registers, it is definitely not entanglement which is the essential ingredient. what i said above is that claiming that discord is the holy grail because one can compute it in dqc1 and find that it is nonzero or has some scaling properties, is not the way to go. in those papers mentioned above, apart from the most recent one which has a limited number of refs in the arxiv version, we quote Eastin as well, and as i said we are not ignoring it. so mentioning dqc1 is not taboo, if you give a less superficial reading you see how and in what measure we refer to that and to what discord has to do with it. in the specific, dqc1 and some instances of it realized experimentally may offer a good example where to test alternative measures of discord, or to study their accessibility, which was the subject of the studies you mentioned.

      i wish to clarify this because a fast reading of those posts might lead to the incorrect conclusions that those papers you mention are examples of bad practice, which i firmly object to, and this might damage especially the students who worked on it.

      this said, please direct any other comment to me personally (GA) and merry christmas to all. i'm out of here.

      Delete
    12. "who feels that entanglement and nonlocality are too "purist" and perfect things for anyone ever to challenge beyond, is just showing a narrow mind. you stopped at a layer of quantumness which is not the deepest one. and you are missing a lot. nobody knows what and how much, but just for this, why stop questioning?"

      INDEED.

      Delete
  9. ah ok so founding fathers and sacred monsters like Zurek, Caves, the Horodeckis, Ralph, DiVincenzo etc. all sudden became morons as they started investigating more natural and general notions of quantumness of correlations?

    ReplyDelete
    Replies
    1. Just because they did great work in the past doesn't mean everything they do is gold. To be more concrete: Ollivier and Zurek first paper on discord, as well as Horodecki et al paper on quantumness (quant-ph/0410090), are actually good papers; Caves et al paper on DQC1 is not of much value (just computing discord in some DQC1 circuits and realizing discord might be larger than entanglement...), and it really led to a lot of horrible hype (but the author cannot be blamed for that); Ralph et al experimental papers on discord are as bad as it gets, with boring experimental realizations of even more boring experimental ideas, and highly hyping it in order to get it published in a high impact journal; I'm not aware of DiVincenzo work on the subject...

      This comment is actually representative of another bad practice in science: Of justifying studying something because well-known researchers do it, or because it's fashionable.

      Delete
    2. Thanks. DiVincenzo et al worked on locking and defined the "classical mutual information", which if subtracted from the quantum mutual information gives a symmetrized version of discord, the same thing that Marco calls Delta_CC in his no-local-broadcasting paper
      http://prl.aps.org/abstract/PRL/v92/i6/e067902

      I am not saying that doing what big names do justifies it, but I am saying that people who know what they are doing are not ashamed to work on discord, possibly providing good pieces of knowledge. I admit it is really hard to filter the literature and I as much as I can keep on rejecting bad papers on the subject (but without any "guilty pleasure", it is just in my duty). I am not happy of the reputation of this subfield, but I defend the value of this research at least at the level of motivation. I also agree that the DQC1 thing is very weak and that overhyping is only doing bad things to the field. Also I would have rejected both nature physics papers, although I work on discord. At least one of them is even flawed.

      Delete
    3. I think DiVincenzo et al paper is also a good one. Actually it seems we agree quite a lot on the topic!

      My main concern is that while DiVincenzo et al, Ollivier and Zurek, Horodecki et al, and a _few_ others are indeed good papers, the interesting follow up questions are generally hard ones (such as understanding data hiding better, or solving some of the open questions of quant-ph/0410090) and the recent research on discord is largely going in an orthogonal direction. Generally I have the feeling that the study of notions of quantumness more general than entanglement, while potentially interesting if done with properly, is too of a specialized topic to have an entire research area around it, and this is an important reason for the flurry of embarrassingly lame results we see today.

      > Also I would have rejected both nature physics papers, although I work on discord. At least one of them is even flawed.

      How is it flawed? (I think it would be good for people seriously working on quantumness if they tried to stop the hype on the subject, so it would be useful to have this information openly, if not on http://blogs.nature.com/, at least here on the Rat's).


      Delete
    4. good. Ok then the problem with discord is that the field boomed because of Datta and Caves' paper which is about "calculating discord in something to try sometihng different from entanglement" so people stuck with this trend. Most people indeed ignore the works on quantumness done in the decade while the original discord paper was mostly ignored. and this means those good papers by oppenheim et al, divincenzo et al, etc. so one should stick to the roots and really look at deep questions like those. but not everybody is as skilled. but then i don't think that the field does not deserve to be investigated further. and i don't see it more specialized than entanglement, why should it be so?

      about the nphys papers, i am not sure about the continuous variable one, I only think that the one on geometric discord and remote state preparation is doubly flawed. one, because geometric discord is not a good measure, the hilbert schmidt norm is not well behaved. and then the protocol is made such that geometric discord works for it. one is trying to prepare a pure state on bob's qubit, without allowing for any correction of the mixedness of the shared resource state (unlike teleportation).

      Delete
    5. >good. Ok then the problem with discord is that the field boomed because of Datta and Caves' paper which is about "calculating discord in something to try sometihng different from entanglement" so people stuck with this trend. Most people indeed ignore the works on quantumness done in the decade while the original discord paper was mostly ignored. and this means those good papers by oppenheim et al, divincenzo et al, etc. so one should stick to the roots and really look at deep questions like those. but not everybody is as skilled. but then i don't think that the field does not deserve to be investigated further.

      i agree.

      >and i don't see it more specialized than entanglement, why should it be so?

      entanglement is certainly more interesting than discord. It's comparing some nice open questions in a few nice papers, as it's the case of quantumness, against the deepest concept in quantum theory. E.g. recent research in entanglement include the development of better methods to describe low energy states of many particles systems; non-locality and device-independent cryptography; obtaning better codes for classical and quantum information transmission; etc, etc. In fact it's difficult to find an area of quantum information and computation which could not be seen as studying a given aspect of entanglement.

      Delete
    6. see the reply above. recall that e.g. for MPS and PEPS you are talking about simulating ground states. so MPS and PEPS are pure states. entanglement might just as well be called discord there. in most of the above cases, it is the same thing. the necessity of entanglement for quantum computation, it's again pure-state entanglement, i.e. discord.

      "entanglement is certainly more interesting than discord" is really not sounding right. entanglement is contained in discord, discord goes beyond it. read my reply above thanks

      Delete
    7. Please see my answer in the previous comment.

      Delete
  10. Merry Christmas, fags!

    ReplyDelete
    Replies
    1. I think this comment actually raises the level of discourse in the comment section.

      Also, anonymous above who says that discord is more interesting than entanglement because all entangled states have discord is full of shit. For example, consider the set of positive semi-definite trace-one matrices. That set is included in the set of hermitian matrices, but infinitely more interesting, since it's exactly the set of physical states.

      Delete
    2. Yet, arbitrary hermitian matrices are useful for representing quantum observables and many other things in Mathematics. Definitely, your comment doesn't raise the level of the discussion but hey, you use the word "shit", so you are probably a very cool guy!

      Delete
  11. Dear Rat,

    On a similar note, do you have any opinion on the sudden death of entanglement? What about the sudden death of discord?

    ReplyDelete
    Replies
    1. Oh, you mean this paper, http://arxiv.org/abs/0905.3376, which has been cited 263 times? Discord cannot die suddenly, because the set of states with zero discord have measure zero.

      Delete
    2. Ouch, I was just mindlessly combining buzz-words.

      Delete
    3. Oh, sorry! I realized you were combining buzz words... my harsh tone was directed at the authors of that shitty paper, not at you. :)

      Delete
    4. at the time 0905.3376 was submitted, the result of Ferraro et al. (the zero measure thing) was not known. Of course in hindsight it is easy to say that something is trivial. Will you people stop being insulting to other people's work just because you're being anonymous?

      Delete
    5. Fair enough, Tommasso: Ferraro et al.'s paper appeared three months later than 0905.3376, so one can perhaps understand that the latter was cited three times in 2009 (source: NASA ADS).

      Now, can you explain me why 225 people have cited 0905.3376 between 2010 and 2013?

      Delete
    6. Because there is still the freedom to cite all the papers that you wish. And it is a common bad practice to cite papers without reading them. Yet, it is not peculiar of discord researchers. You admitted to have cited the O-W paper: was it so difficult to check the bibiliography of your own work?




      Delete
    7. So, in your understanding, 225 people erroneously copied the reference to 0905.3376 from article to article. Man, that's not a reference! That is a virus!

      I wonder if after several more citations it will start mutating... Think about it! In some generations, it could start "eating" the other references, or jump to the main text! Eventually, it ends up in a law code and starts affecting human affairs. And at the end... it is ruling the world!

      This is not very appropriate for a post, but I believe I can turn it into a nice short sci-fi story. I will call it: "the extra reference". Thanks for the inspiration, Anonymous!

      Delete
    8. No, I said: "Because there is still the freedom to cite all the papers that you wish. And it is a common bad practice to cite papers without reading them." It's not difficult to understand, should I make a comics?

      Anyway, I don't suggest you to start a new career in sci-fi. Your posts are not funny and the paradigm "Everyone (grants/papers referees, Nature/Science editors, QIP organisers, discord people) is a bunch of idiots or there is some conspiracy" has been overused. Keep strong with the one of (good) quantum researcher.

      All the best

      Delete
    9. >It's not difficult to understand, should I make a comics?

      Yes, please. Your explanation has so many words... it is very confusing!

      > Anyway, I don't suggest you to start a new career in sci-fi.

      You don't like the premise of the "extra reference"? But it is pure gold!

      >Your posts are not funny and the paradigm "Everyone is a bunch of idiots or there is some conspiracy" has been overused.

      ... and, in spite of all that, you keep reading my blog. I feel flattered. God bless you, Anonymous!

      Delete
  12. Thanks to the Rat, who ignited an interesting discussion.

    Personally, I don't see any point in disputing the intrinsic value of "discord", "entanglement" or whatever other concept or research line. I read that entanglement is the deepest concept in quantum theory. Actually, one can make a great course in Quantum Mechanics without any mention to entanglement. High energy physicists, who works daily with quantum mechanics, couldn't care less about entanglement.
    Does it mean that entanglement is useless?

    In the same way, maybe discord is useless for Quantum Computation but useful for other fields. It is a property of quantum states currently under investigation. The Rats can express their legit opinions on it. I even agree with them on some points (excessive hype, many bad works, etc.). But only scientific progress and acquisition of further knowledge will determine, as always, if discord will be someway useful, a research line leading nowhere, a breakthrough, or other. The attitude emerging in some comments here is more suitable for art or literature than science. In the 70's at some point no one was working on electro-weak interaction theory, was considered useless because didn't allow to make predictions as QED, that you would have called a much more interesting theory...

    Finally, sorry but I find very sad that even PRL referees, who are supposed to be experienced researchers and clever people, still don't understand why journals keep them anonymous. This measure is taken to guarantee freedom and independence of judgement, not to cure their everyday childish frustrations by publishing their reports. Well, one can have "guilty pleasure" in the way he wants!

    ReplyDelete
    Replies
    1. > I read that entanglement is the deepest concept in quantum theory. Actually, one can make a great course in Quantum Mechanics without any mention to entanglement.

      Deepest might have been an exaggeration (from my part) to make the point that entanglement is
      more meaningful than discord. However I really don’t think you can make a great course in quantum mechanics without any mention in entanglement. It would be an outdated course, depriving the students of _one_ of the deepest concepts in quantum mechanics.

      >High energy physicists, who works daily with quantum mechanics, couldn't care less about entanglement. Does it mean that entanglement is useless?

      I remember John Preskill mentioning that he recently went to one of the yearly Solvay conferences and almost everyone there, including several leading high-energy physicists, were talking about entanglement. But I agree currently entanglement is a much more emphasized in quantum computation than in other fields of physics.

      > The Rats can express their legit opinions on it. I even agree with them on some points (excessive hype, many bad works, etc.). But only scientific progress and acquisition of further knowledge will determine, as always, if discord will be someway useful, a research line leading nowhere, a breakthrough, or other.

      We are in agreement. But since the world we leave has limited resources it’s legitimate to try to allocate them using our partial information at any given time. At the moment, taking into consideration what research on discord has achieved and its future promise, it seems a good strategy to have 6 people studying it, as the Rat proposed. Everyone is free to work on whatever they find interesting, but then they cannot complain if people don’t see the point in what they are doing.

      >The attitude emerging in some comments here is more suitable for art or literature than science.

      ????

      Delete
    2. continuation...

      >In the 70's at some point no one was working on electro-weak interaction theory, was considered useless because didn't allow to make predictions as QED, that you would have called a much more interesting theory...

      not really. No one knew how to make a good theory for the electro-weak interaction, one as successful as QED, but certainly everyone thought it would be a great achievement to have one. It’s true that some of the approaches that turned out to be successful (after the idea of symmetry breaking was eventually introduced) were largely ignored for a long time, with only a small group of people working on them. But I don’t find it a problem. It’s just the way science works. It would have been a shame if this small group of people weren’t allowed to work on what they found to be an interesting research area, but it would also have been a bigger shame if 10000 researchers would be working full time on the research area, writing papers that go nowhere, hyping the results in order to fool people with less knowledge on the subject, and ultimately using the science-is-about-free-thinking card when pressed against the wall.

      >Finally, sorry but I find very sad that even PRL referees, who are supposed to be experienced researchers and clever people, still don't understand why journals keep them anonymous. This measure is taken to guarantee freedom and independence of judgement, not to cure their everyday childish frustrations by publishing their reports. Well, one can have "guilty pleasure" in the way he wants!

      To say “guilty pleasure” might have been wrong, but it’s not a bad thing to public show the report of a bad paper. In fact one could make a point that referee reports should always be public. I agree with you that it’s a good practice to have anonymous refereeing. But what’s the relation with not showing the reports in public?

      Delete
  13. > Deepest might have been an exaggeration (from my part) to make the point that >entanglement is
    >more meaningful than discord. However I really don’t think you can make a great >course in quantum mechanics without any mention in entanglement. It would be an >outdated course, depriving the students of _one_ of the deepest concepts in quantum >mechanics.

    I agree that entanglement is more important than discord! Ok, in a fifty hour course I would dedicate two-four hours to entanglement/Bell inequalities. And of course I would not talk about discord.

    > I remember John Preskill mentioning that he recently went to one of >the yearly Solvay conferences and almost everyone there, including >several leading high-energy physicists, were talking about >entanglement. But I agree currently entanglement is a much more >emphasized in quantum computation than in other fields of physics.

    Well, I looked at the ad of that event. I suppose that everyone talked about everything about quantum mechanics, from quantum information to string theory, on general lines. I was referring to state-of-the-art research.

    >We are in agreement. But since the world we leave has limited >resources it’s legitimate to try to allocate them using our partial >information at any given time. At the moment, taking into >consideration what research on discord has achieved and its future >promise, it seems a good strategy to have 6 people studying it, as the >Rat proposed. Everyone is free to work on whatever they find >interesting, but then they cannot complain if people don’t see the point >in what they are doing.

    Are we now talking about funding/positions allocation? I don't see such a plethora of grants/postdocs assigned to work on discord. It's fine to discuss about it, but If there are seven or seven hundred good grant proposals on discord, then they should be funded, and in general be judged without bias. I think that here no one complained about lack of interest on discord, but about the confusing merging of good points, open issues, and superficial and not professional behaviours.

    >not really(...) the wall.

    But here you are mentioning very general problems of modern science, and then linking them with discord. There is greater accessibility to a scientific career in many countries, thus the average quality of researchers and papers is lower. Also, hyping the results is a bad practice of some authors, some press offices and some scientific writers, that surely didn't start with discord. Again, I don't see this overwhelming political power of discord community you are afraid of.

    >To say “guilty pleasure” might have been wrong, but it’s not a bad >thing to public show the report of a bad paper. In fact one could make >a point that referee reports should always be public. I agree with you >that it’s a good practice to have anonymous refereeing. But what’s the >relation with not showing the reports in public?

    Sorry, but here you have misunderstood my point. In principle I have nothing against public reports. But first, the authors have to be informed
    that the reports on the paper they want to submit will be made public. Second, they should have the opportunity to reply.

    ReplyDelete
  14. When can we expect the next part of "Research lines that lead nowhere"?

    ReplyDelete
  15. how about superstring theory? does it lead to anywhere?

    ReplyDelete
  16. "Miguel Navascues 22 October 2012 14:49
    I work in quantum foundations. How could I disagree with this statement?"

    Two (bona fide) questions:

    1. What have we learned about the "foundations" from QuInfComp (QIC)?
    2. Might it happen that without such discord-like stuff, the QIC field might have already been dead?

    ReplyDelete
    Replies
    1. >1. What have we learned about the "foundations" from
      >QuInfComp (QIC)?

      I will mention four insights:

      a) You can design a quantum experiment that refutes all (non-local) realistic theories in which an influence propagates at finite speed v>c, even if you don't have a clue on the magnitude of v.

      b) If our universe were a little bit more non-local than quantum mechanics allows, it would violate macroscopic locality (the principle that "natural" macroscopic experiments admit a local hidden variable model) and information causality (the efficiency of random access codes is the same as in classical physics).

      c) You can recover quantum mechanics if you assume local tomography, continuous reversible dynamics and that all systems that effectively carry the same amount of information have equivalent state spaces.

      d) Free energy quantifies the amount of useful work which can be extracted from asymptotically-many copies of a quantum system when using only reversible energy-preserving transformations and a thermal bath at fixed temperature.

      >2. Might it happen that without such discord-like stuff, the
      >QIC field might have already been dead?

      Heavens, no! Have a look at the arxiv: you'll find device-independent quantum cryptography, randomness amplification, topological memories... and all kinds of QI-inspired foundational research. People are doing cool stuff out there!

      Delete
    2. A reference on b, please.

      On d: what's temperature, quantum mechanically?

      On2: how can I be sure I'm not implicitly accounting for some correlations?

      [And, PLEASE, stop checking if I'm a comp - my astigmatic eyes can't read it easily!!!]

      Delete
    3. >A reference on b, please.

      http://arxiv.org/abs/0905.2292
      http://arxiv.org/abs/0907.0372

      >On d: what's temperature, quantum mechanically?

      Most papers in quantum thermodynamics assume (actually define) that your bath is in a thermal state, i.e., rho\propto e^-H/KT, where H is the hamiltonian of the bath. T, the temperature, is just a parameter.

      >On2: how can I be sure I'm not implicitly accounting for
      >some correlations?

      I don't understand the question.

      >[And, PLEASE, stop checking if I'm a comp - my astigmatic
      >eyes can't read it easily!!!]

      Sorry, Mr Blogger's rules. Try posting with a gmail ID, maybe that works.

      Delete
    4. On Temperature:

      It's easy with "bath" - I like it. I am not sure about an atom. Or even worse: how can I DERIVE the parameter T from the QM principles?

      On "On2": how can I be sure that "device-independent quantum cryptography, randomness amplification, topological memories..." is not merely about quantum correlations?

      Delete
    5. >It's easy with "bath" - I like it. I am not sure about an
      >atom. Or even worse: how can I DERIVE the parameter T from
      >the QM principles?

      You got me. People in the field always take Boltzmanians for granted, but so far I haven't seen a convincing argument to justify them in the quantum case. Can any expert in quantum thermodynamics help us here?

      >On "On2": how can I be sure that "device-independent
      >quantum cryptography, randomness amplification,
      >topological memories..." is not merely about quantum
      >correlations?

      I don't really understand the expression "...is merely about quantum correlations". All device-independent stuff requires entanglement (and, yes, in particular, non-classical states). However, not any kind of entanglement is useful here: DI protocols using just two inputs and two outputs won't work with PPT entangled states, for instance.

      Saying that "quantum discord" or "quantum entanglement" is at the heart of such protocols because it is necessary is like saying that "things" are at the heart of quantum computation because you need "things" (ions) to codify your qubits and use other "things" (a laser) to make them interact. While true, the philosophical study of the "thing" concept is not going to give you any insight on how quantum computing works. Do you see my point?

      Delete
    6. "Do you see my point?"

      Sorry for being imprecise. I'm not talking about "hardware" or any philosophical issue.

      My question, in (over)simplified terms, reads: if we "remove" quantum correlations from a (COMPOSITE) quantum system, may it be that there nothing non-classical remains in the system? Which non-classical characteristics exist independently on correlatrions? [Or other way around: what is "quantum" (in COMPOSITE systems) EXCEPT the non-classical correlations?] Is it more clear now?...

      Delete
    7. Quantum mechanics was born out of the need of explaining a series of 1-partite phenomena, like the black body radiation. The multipartite paradoxes came later.

      For me, the fact that the observables X and P have a continuous spectrum, whereas L_z, or X^2+P^2\alpha^2 don't, is already a quite non-classical feature (that can be observed in states of zero discord). True, it is not device-independent. But neither is entanglement.

      Delete
    8. "... 1-partite phenomena, like the black body radiation"

      - Sorry, but thermal radiation is a collective phenomenon.

      "The multipartite paradoxes came later."

      - I'm talking about correlations. Where are correlations in single particle phenomena? So, again: how can I be sure that there are quantum phenomena for bipartitions without correlations?


      Delete
    9. >Sorry, but thermal radiation is a collective phenomenon.

      ... which is typically observed at one location, not two or three separate ones. Ergo, it is a 1-partite phenomenon (although it may involve many particles).

      >So, again: how can I be sure that there are quantum
      >phenomena for bipartitions without correlations?

      Again: tell Bob to go for a coffee and make Alice measure X, P, X^2+P^2 and X^2+P^2/2 on her subsystem. There you have a quantum phenomenon (and it works with three parties, too!).

      If what you mean is: "can I conduct ANY quantum multipartite protocol with states of zero discord?", have a look at arXiv:1304.2667.

      Delete
    10. ". which is typically observed at one location, not two or three separate ones. Ergo, it is a 1-partite phenomenon (although it may involve many particles)."

      Exactly! It's LOCAL - but for the "internal degreeS of freedom" of a multiparticle system, not for a single-particle system. [And yet requires non-canonical ensemble averaging though]

      "If what you mean is: "can I conduct ANY quantum multipartite protocol with states of zero discord..."

      Maybe one can do "non-classical" without quantum correlations. But THAT nonclassical does not necessarily mean it's "quantum".
      [e.g. you may have quantum tunneling, but this does not necessarily mean your system is "quantum".]

      So, my question seems sharpened, not answered yet.

      Delete
  17. Why Science would take any care about "application": why should one worry too much about "operational meaning of discord"?
    Better sounds the moto:

    "'To the electron -- may it never be of any use to anybody.' -- JJ. Thomson's favorite toast"

    ReplyDelete
  18. Hey Miguel, I just realised that your post created a lot of discord between you readers.
    Ha!

    ReplyDelete
  19. Hey Miguel, your post might help quantum discord to grow up.
    Ha!

    ReplyDelete