Saturday, July 9, 2011

A simple challenge to theorists

[I am] concerned with the spirit that seems to pervade much of theoretical physics nowadays, and with the interests and preferences shown by some of its practitioners. Of course, theoreticians depend on their experimenter colleagues for the bread and butter of their work,  ..... And yet when I try to read some of the current literature or listen to some of my colleagues, I can't help asking: Are they in touch with reality? Are they seriously interested in physics? As an example let me cite antiferromagnetism, which is again a popular topic since the discovery of the superconducting copper oxides. Very sophisticated models of antiferromagnetism have been proposed and worked out in great detail by people who have neither heard of, nor do they care to learn about, hematite, or what everybody else knows as a rusty nail....
... Even among theoreticians the natural order of things should be electron-spin resonance in hemoglobin first, anti-unitary operators and quaternions second....
As a referee for scientific journals I have led a losing battle with authors who don't want to (or can't) explain their general ideas in terms of simple experimental systems, or even in terms of high school geometry as the best in our trade did, from Einstein to Feynman.
Martin Gutzwiller, Two questions physics can no longer avoid, Letter to Physics Today, August 1994.


  1. This is a general problem we have been discussing as theoretical physicist "wannabe"s for a long time. Thank you for raising this topic, maybe you or the commenter can help me with my confusions, I really hope anyone to break my argument in a satisfactory way.

    1) The two theories of theories in general: A theory that does not give an "experimental result" "spot on" is not necessarily a bad theory. Also, a theory that gives an experimental result "spot on" is not necessarily a good theory.

    The biggest example I have is TDDFT and TDDFT with exact exchange. TDDFT is an ab-initio and exact theory, but like DFT, it is limited by the kernel, currently ALDA kernel is good enough for excitons with "electron and hole" are spatially very close (TDDFT is a theory for molecular systems, I approach its application to crystals highly sceptically, especially in the presence of better suited light-matter interaction approaches for bulk such as TDCFT.) In the literature I keep seeing people using hybrid XC like B3lyp and reporting good agreement with experiment. The problem is, although B3lYP might be a good approach for ground state DFT, it is a mixture of elements that are time reversal symmetric and anti symmetric. Thus in the end, the kernel does not obey the time reversal symmetry. I am quite confused how such a kernel can be used in a time dependent schrodinger equation. In my opinion, considering the number of external parameters you impose on the system, with the same number of parameters, you can have better results with chicken bones and some stones.

    2) The branding, fashions, and belief driven science: Something published by a high ranking author/group is automatically good science and should be believed %100 percent even if doesn't make sense (for your case). If you do not have a "brand name" in your paper, it is almost impossible to get yourself heard (publish in a good journal). If you are not backed up by a "brand name" challenging another "brand name" is equivalent to academic suicide.

    Well, this is a problem that existed for a very long amount of time, in science in arts, etc. See Emil Rupp's story
    The more scary one is Schön scandal, please notice how the lab workers had to hide their identity, so that they are not "erased from science" I have friends that were "erased from science" because they have stand against the theory of their bosses in a very theoretically sound matter, I really wonder if it is a common practice to "erase people from science" when they challenge your theories in big groups. I mean, can you considered to be a scientist if you are not open to discussion.

  2. 3) Scientists are evaluated in a bureaucratic manner.
    How can you distinguish a good scientist from a bad scientist if you are in a bureaucratic position and does not understand a thing about what the work you are to evaluate is about? You look at the numbers. What are the numbers, number of papers, probably weighted by where they were published. Combine that with branding... Hmmm, maybe Terry Prachet is a better writer than W. Shakespeare, Stephenie Meyer is definitely better than F. Kafka.

    4) The American way of science (science as a business).
    Consider you are a big group boss. In order for your group to exist, and you to be scientist, you have to have funds. In order to have funds you have to be evaluated in a bureaucratic manner, in order to have good evaluations you have to have many good publications. Added phenomena: a member of your group failing has almost no consequence to you, you are like an infinite potential to someone that has lesser academic rank than you, only other professor of equal or more calliber can challange you. People failing in your group are rarely heard outside. Natural consequence and common practice: Instead of spending enormous amounts of time to train a student in the way of science like the old German or Russian system, hire as many students as you can, assign them with some tasks in "hot" and "fashionable" topics, without even bothering to check if they make physical sense, and do business tactics like "group meetings" "task appointments" to keep them working and on edge. Lets exaggerate a bit, if you have 20 students, even if a huge number fails, there is little or no consequence to you, you will still have a big number of "high ranking" papers added to your cv, and you can hire 25 of them next term. The failed are forgotten, even if it was just their naivety, inexperience, and bad luck that they ended up spending a huge portion of time trying to do something that does not have physical sense.

    As long as there is an infinite number of PhD. students available this system works, at least seemingly much better than old system.

    Maybe you can hire some good scientists in your group that publish only once in a while, but make good papers, and did not drift so far from science that they can still answer the students questions.

    5) 1,2,3,4 combined together: Good science becomes an exception not norm, at the exceptional sacrifice and goodwill of foolishly idealistic people. The divide between physical world and physics become deeper and deeper. Science begins to resemble catholic church in medieval times.