Other Research Projects: Apollo Bioinformatics Materials Research Molecular Evolution
Physics of Scale Activities

Dzyaloshinskii interview, part II

### PoS

Can we move back to your early collaborations with Gor'kov, and to some extent with Alexei A. Abrikosov as well? You all became quite enthusiastic about the prospects for field theoretic methods within statistical physics in the late 1950s and early 1960s.

### IED

Yes. It was not only us, because basically there was also Arkadii Beinusovich Migdal, he studied it, and he did everything at zero temperature. And then we read the paper by Takeo Matsubara, and then we came with this idea to use--

### PoS

--temperature Green's functions.

### IED

No, Matsubara did it, he defined everything. Our contribution, besides practical use -- we used it extensively in different phenomena, like in our book you may see that we applied it -- but we found this way to use, instead of imaginary times, to use imaginary energies. So in fact in the Matsubara approach you get diagrams, the diagrams were somewhat unworkable. When you integrate stuff in the finite imaginary time region, it's hard to work, you never work automatically, you get exponentials, and God knows what, and they proliferate. Then separately I improved the stuff, how to move from Matsubara imaginary frequencies to real frequencies, and I published a paper. They don't use it now, they switched to the Keldysh formulation. I did it almost all in short order, because there was a problem: When you calculate thermodynamics, you cannot easily move to real time, if you sit just on this finite stretch of imaginary time from 0 to 1 over temperature. At zero temperature you go easily: it is nothing but the Wick rotation from Minkowski to Euclidean space in QFT. At finite temperatures you see that conventional diagrams are not enough to build the field theory in real time. You need more, either more functions like Keldysh, or you need more diagrams. The easiest way was with more diagrams. Starting with image -- Feynman diagrams -- you have to rearrange them according to the order of times at different vertices and each time sequence gives different contributions analytically. It's equivalent in fact that you use more propagators. {I. E. Dzyaloshinskii, "A diagram technique for evaluating transport coefficients in statistical physics at finite temperatures," Sov. Phys. JETP 15 (1962): 778-783.}

### PoS

So the article you wrote with Gor'kov and Abrikosov on the applications of quantum field theoretic methods in statistical physics [Sov. Phys. JETP 36 (1959): 636-641], that became chapter 3 of your textbook?

### IED

Yes, in 1958, and then we wrote a book and published it in four years' time, three years' time. {The preface to the Russian edition is dated 1961. An English translation appeared in 1963.}

### PoS

And there was nothing unusual about that...?

### IED

We wrote it extremely fast. I wrote my part, I believe, in two weeks. We sort of lived with the stuff. Actually I don't remember that I consulted anything when I wrote this. It was our life and work at that time. It was really easy to write.

### PoS

Were there other people in Moscow besides, say, Efim Fradkin, who were also interested in this kind of cross-fertilization?

### IED

Yes, he did the same thing, in fact, but Efim, our difference with him was fundamentally, I would say, technical. Efim always worked in [Julian] Schwinger's framework. He likes functional derivatives, and OK, so he did not go farther. You know, it's hard, I know cases where Schwinger's way is perfect -- in the case when you may solve something exactly, and like Schwinger himself. But if you want to calculate a small correction (he did it, too), it's much more tedious work.

### PoS

Do you have any recollection of what the Landau group at the Institute of Physical Problems thought of the paper by Schwinger and Paul Martin from 1959?

### IED

I never read it. I did not pay attention. I came to know the paper afterward.

### PoS

But for much the same reason, it probably didn't offer a calculational tool...

### IED

Yes, sure, that's why we never mentioned it. I eventually became aware of it. It was sort of just a statement of relations. But that's why I did not use it. Now you may say in the same way that the Keldysh work was originated by Schwinger, but you look at an actual Schwinger paper (not just the references), and then you would not see anything like this. Schwinger was already not interested in that.

### PoS

The original Schwinger-Martin paper was supposed to be the first of two papers. Schwinger couldn't be bothered with the second part, so it was Martin and Leo P. Kadanoff who wrote the second paper.

### IED

Well, but that's different. But again, it's usually if you... [The paper by] Martin and Kadanoff, it's basically, you should use higher correlations to [Low?], and then it works. But that's uncontrollable. We were trained never to use uncontrollable approximations. Landau would kill anyone who was using them. In fact, Schwinger himself never did anything like this in his lifetime, if you look at his work. He gave formulations, he solved rather normal cases in strong magnetic field, and that sort of stuff, where you may really use those functionals of his in an effective way. So in fact he tremendously improved what [Werner] Heisenberg and [Hans] Euler did. They did everything in a rather simple non-relativistic way. They used the spectrum of a charge in a magnetic field, or an electric field. Schwinger did this perfectly. I enjoyed his work. But I never use functional integrals in my work, because... Besides this stuff, either you may solve the problem's exact integral, or you work with the single-loop approximation, whatever you're doing. The last thing I did in this respect I told you about. It's first-order, and all fixed points are unstable. Then you draw the phase portrait of the flow, and then you go -- there are instability lines when your initial Hamiltonian is unstable. You may calculate a lot in this way -- discontinuity, entropy, and stuff: everything is possible. And finally, I was fascinated by writing down in conventional RGs -- in the sense that they are not simple one-parameter RGs or whatever. Unfortunately it was in the context of high Tc superconductivity, and of course it does not have relevance. In fact I attended all the conferences in 1987, but then I stopped doing anything. Now I realize that there is nothing which is unusual in that.

### PoS

Can we discuss again how you got started with work on ferromagnetic materials? You have a piece from 1959 with Abrikosov on spin waves in a ferromagnetic metal [Sov. Phys. JETP 35 (1959): 535-537].

### IED

Ah, 1959, that paper's wrong. It's my only mistaken paper.

### PoS

Where do you think you went wrong with that?

### IED

Unfortunately we were--

### PoS

I get the impression that everyone read Landau's theory of Fermi liquids and found that an immensely productive idea, and tried to find--

### IED

The idea was good, but the result is wrong. The point was that we missed a term in the functional. It was shown by Conyers Herring that we were wrong. We wrote down the spin dependency F-sub-Landau. When you calculate spin wave spectra, the answer is that the energy is proportional to the square of the wave vector. Then you see that you missed your initial functional, the term of the same order. Landau was always right in doing this for zero sound, because zero sound [writes on board]... and then it's enough. But here you may write it down, here you may add the term which is, say, H, and here you have (dn/dx)^2, and this H is a new phenomenological parameter. So in fact our answer for the mass of the spin wave was wrong. It is not defined solely by F-sub-Landau. And then I (with my student at the time), I analyzed it with all the machinery of field theory, for then it can be solved in actuality, when the spontaneous moment is small.

### PoS

You returned to the subject later. {Reference is to "The theory of weak ferromagnetism of a Fermi liquid," Sov. Phy. JETP 43 (1976): 1036. [confirm]}

### IED

And so this is the correct answer.

### PoS

With the additional term.

### IED

Yes. We did not do it this way. It's not enough, anyway. Here we were able to pass through the transition point and to find so-called paramagnons. It's excitations with energies which are much larger than saturation magnetization, which is small here, so we used the fact that it is small. They are not excitations, because they are imaginary. The spectrum of paramagnons is diffuse, with energy proportional to ik^3. It's a known fact. Unfortunately we did not find in this way anything new. This ik^3 was found by Seth Doniach at Stanford, but they consider simply classical oscillations of magnetization in a paramagnet, and then you have this stuff. But we were able to write the whole spectrum. In 1995, 1996, I finally came to consider all this to be crazy. They are generalization of RGs, so I came up with one problem which I thought was relevant to high Tc. Now I understand that it could be a model of non-Fermi liquid behavior, because this type of stuff destroys the Fermi surface in the sense that technically the residue of the propagator goes to zero when you renormalize. I am not too happy with this. And another thing, it was written in our ancient style. Nowadays people...

### PoS

How do you mean, "ancient style"?

### IED

It's a propagator with [word?] diagrams, and that sort of thing. People now, they prefer to write down the RG. From what I have seen, they are usually wrong, and then they solve numerically, which is absolutely unnecessary, because I was able to solve analytically. I do not know, but anyway, it had something to do with decreasing mathematical culture, at least in condensed matter physics.

### PoS

Do you think the training of condensed matter theorists has become more pragmatic in mathematical terms, or what do you mean?

### IED

No, I mean, when you just write down a model, and then you try to solve it numerically, obviously the model is relevant -- it cannot be solved.

### PoS

That's sort of an axiom of condensed matter theory?

Yes.

### PoS

If the model is relevant, it cannot be solved analytically.

### IED

No, and numerically, too. The only relevant model, you cannot solve. You have to wait for a new generation of machines. So to me, as I understand high Tc, there is nothing new, no ideas, it has the same [b-------] status. The fact that it is d-wave, the fact that it is a mixture of s and d states, and in fact liquid helium 3 was known to be p-wave decades before, so it's not a great big deal. But the main thing in this stuff, because they are poor metals, and besides they are almost two-dimensional -- in two dimensions there is no Coulomb screening. Do you know, by the way, that we still are not able to calculate the transition temperature of lead? You do it using the whole machinery of G. M. Eliashberg. It's a nice theory, but it does not take into account Coulomb repulsion. You may see, even in good metals, you expect the Coulomb repulsion to be effectively small. Either you calculate oscillations in density of dispersion functions, either from first principles, or better still, take them from experiment. You use another set of data like conductivity and normal state, you know almost everything. But then the results are usually off by a factor of five. And then comes the famous [mu]* constant which you need to account for Coulomb repulsion. And then you adjust it, directly by hand, because without it, your Tc is wrong, three to five times wrong. Well, now, in high-Tc materials, Coulomb repulsion is not known. In good metals it is known. So I think we just should wait. And anyway I do not know a single case when theory could give a guiding light about what materials they should be studying. I lost interest in all this stuff.

### PoS

Well, can we talk about phase transitions at an earlier stage of the game? I'm thinking in particular of Fairbank's work in 1957. Fairbank and his colleagues in 1957-1958 published a work that... {W. M. Fairbank, M. J. Buckingham, and C. F. Kellers, "Specific heat of liquid He4 near the lambda point" (August 1957), published in J. R. Dillinger, ed., Low Temperature Physics & Chemistry: Proceedings of the Fifth International Conference (Madison, Wisconsin: University of Wisconsin Press, 1958), 50-52. }

### IED

I do not know about it, tell me about it.

### PoS

Well, apparently that seemed to indicate that the helium transition was sharper than expected, it wasn't simply--

### IED

Ah, you mean the experimental work, Fairbank's, yes, I know. At that time they claimed there was a logarithm (which is not a logarithm), but anyway. Landau was always excited with the work of Onsager.

### PoS

Yes, tell me a little bit about Lars Onsager, when did you first learn about the Onsager solution?

### IED

From Landau. He was always -- that was his trouble. And then I will tell you: after all this Matsubara stuff, he came up with a suggestion. Actually, he used what is now called the Landau-Wilson functional -- it was before Wilson. You use this functional, you calculate propagators, and obviously in 3-D everything gets out of hand. The next term is obviously larger, if you start with k^2, but then Landau observed that you may simply assume that the exact propagator in 3-D behaves like k^3/2 -- then you have logarithms and bubbles -- that was his idea. So we did a lot of work. He came and told us, "look," and everybody got excited. So actually then [IED writes on board] we sum up this and that and that and that: it's easier. We were clever enough, so everything was solved. But it does not work, the solution is not logarithmic, it's not a good fixed point. Actually, it does not satisfy...

### PoS

This is very interesting, because in the 1958 English edition of Statistical Physics, Landau and Lifshitz are quite cautious. They say, "Onsager's result is extremely impressive, but it is for only two dimensions, and we have no expectation that it have an immediate--"

### IED

No, I'm not so sure. Landau was absolutely aware, and he was worried about it. So it was his idea, but it does not work, it's not a fixed point, it's not conformable. It is not scale invariant. In fact unfortunately the corrections are not so good.

### PoS

So were there other people in the 1950s who were interested in studying non-classical examples, exceptions from mean-field theory?

### IED

But all of them are. You define, say, the parquet summation, the new mean field theory -- they are now more and more exotic. Now I teach my students -- they are so badly prepared, so the only way for me to explain to them superconductivity and stuff, I simply teach them that there are new exotic mean fields, like superfluidity. It's for teaching poorly-prepared people. Of course you may write it down as a mean field theory, and functional integrals, or whatever. But my students, they usually do not know second quantization, so that is hard. So I cannot follow with Bogoliubov's method. It takes a long time to explain what a-- {End disc 1}
{Begin disc 2} The idea is somehow purely numerical. And of course it's OK when you are going to do engineering, or if you are going to just service, it's OK. But if you really want to do something crucial in condensed matter theory...

### PoS

In your current situation here do you find yourself teaching condensed matter theorists, materials scientists--

### IED

No, I enjoy it. As I told you, I started work as an engineer, and in fact my education was immensely higher than was needed, so I know how to go "down." It's easy to go "down," it's hard to go "up."

### PoS

But you know how to pick your level.

### IED

Yes, and since then I know what an "engineer" means.

### PoS

So in this department you're teaching materials scientists, materials researchers?

### IED

No. It's practically impossible for them, I could not even tell you the probability for any of our students to somehow succeed in academia. so that means they are bound either to go to good industry labs, which is now exceptional, you understand, because many operations are closing their labs. Lucent destroyed Bell Labs, it's a national shame, I would say. IBM did not do this, they still... But Lucent, in general... I could never understand the idea that the big operations should make only lasers.

### PoS

I have noted here that, just to circle back to Fairbank again,...

### IED

No, no, Landau was aware that the theory is wrong.

### PoS

But Alexander Voronel recalls that Landau denied the analogy between the lambda point and the critical point, the lambda point of helium and the notion of a general critical point in phase transitions.

### IED

Ah, you mean the critical point in liquid-gas transitions.

### PoS

For instance.

### IED

Well, but it is not identical.

### PoS

But is there an analogy, is it worth pursuing, or no?

### IED

Landau understood too that the result is wrong. In fact we know that exponents are different and basically it's not the same phenomenon. The singularities are not the same in the critical case. I do not know what Voronel... Well, Voronel was one of the first to measure this kind of stuff, but his measurements were wrong.

### PoS

The measurements on argon?

### IED

I do not know about the measurements, but he always claimed that he saw logarithms. I tried to convince him that it's the wrong way to look at it. You have to look for an exponent, if you claim that it is zero, then you have to measure plus and minus around the zero, and that's the correct presentation of the results. But he never wanted to do it.

### PoS

His first expectation was that it should be a logarithm.

### IED

Yes, he wanted to have a logarithm simply on the grounds that in Onsager's work, it was a logarithm. Landau did not believe in this logarithm. His idea was this, that you understand that it's not logarithms here in thermodynamics with k^3/2. No. So in fact we were well prepared in anticipation. Wilson's work came to us not unexpected. Internally we expected something like this. But somehow we did not appreciate four dimensions as mean field. And Landau obviously, too. Because his idea, judging by the Onsager result, was that the result is so far from mean field, and should be far. To me, what Wilson did, he has shown that mean field is still relevant in 4-D, that it defines the behavior, and in fact it opens up the way to calculate.

### PoS

In your work on van der Waals forces, you're clearly aware of issues in the quantum theory of fluctuations, and then, of course, Alexander Z. Patashinskii and Valery L. Pokrovsky, their paper says, look, fluctuations are the heart of the matter in some sense.

### IED

I must confess that I was not particularly impressed by the simple assertion of scaling invariance, OK? Because it could be this way, and you immediately realize that it means that there is an RG, there is a fixed point, and that sort of thing. Actually in field theory it's a single cutoff, and logarithms. And then when I read the first article by Wilson, when he used some crazy RG -- if you have read it...

1971-1972?

### IED

Yes, his first paper. He just wrote down an RG which gives you scaling invariance, and that sort of stuff. But then he switched to what's now called Landau-Wilson, and treated it around D4 and expanded it and stuff, and then I was convinced.

### PoS

If I understand correctly, Wilson visited the Soviet Union in 1971 for one of the international symposia?

### IED

Yes, he was there. I met him before. But you know, that [pointing to photograph] was the best gathering I ever attended. It was a Nobel Symposium in 1973, I believe. {See B. Lundqvist, S. Lundqvist, and V. Runnstroem-Reio, eds., Collective properties of physical systems: Proceedings of the Twenty-Fourth Nobel Symposium, 12-16 June 1973 (Stockholm, 1973).} Here's Wilson, [Walter] Kohn, Martin, [Robert] Schrieffer. Do you know who this is? Guess. It's Hubbard. Look, it's [David] Mermin, it's Doniach, [Philip J.] Anderson, that's me. [Leon] Cooper, Gor'kov. That's Ambegaokar, Toulouse, [Pierre] de Gennes, that's Hopfield, Heeger, [Brian] Josephson, Nozieres, [Michael E.] Fisher, and here, I guess that's [Bertrand L.] Halperin. I left the volume [the published proceedings] in Moscow. So that was the best gathering I ever attended. By the way, some people were talking about Wilson's work on phase transitions, but Wilson himself gave a talk on his solution of the Kondo problem. It was the only case when I really saw how his RG really worked. But he directly renormalized the Hamiltonian. In fact it is the only case which I know. Now you understand why it worked -- well, it's not exactly integrable, but when you do some tricks, and throw out this and that -- then it's exactly integrable. It [the meeting] was really exceptional. Otherwise I never met all those people in one place.

### PoS

Is Kadanoff in that group, or no?

### IED

No. I think there were some... Well, you know, even in science there can be delicate personal relations.

### PoS

And of course it's a delicate matter in retrospect. I mean, people like Cyril Domb, for instance, will say outright that, look, Wilson should not have been awarded the prize by himself, it should have been Wilson, Fisher, and Kadanoff.

### IED

Oh, no, that's wrong. I don't agree at all. You see, after time lapsed and everybody got accustomed to this sort of stuff... First of all, it seems a little bit arbitrary. Of course you may assume there is a single scale, but...

### PoS

To return to some of your own work, could you say a little more about how you got interested in the problem of van der Waals forces? Because you have these very meaty papers that you write with Lev Petrovich Pitaevskii and Lifshitz, culminating in the survey piece in Soviet Physics Uspekhi. I mean, there's a lot of work that went into those... {See I. E. Dzyaloshinskii and L. P. Pitaevskii, "Van der Waals forces in an inhomogeneous dielectric," Sov. Phys. JETP 36 (1959): 1282-1287; I. E. Dzyaloshinskii, E. M. Lifshitz, and L. P. Pitaevskii, "Van der Waals forces in liquid films," Sov. Phys. JETP 37 (1960): 161-170; Dzyaloshinskii, Lifshitz, and Pitaevskii, "General theory of van der Waals' forces," Sov. Phys. Uspekhi (Sept/Oct 1961): 153-176.}

### IED

Oh, you mean... But that's not RG. Well, the problem was that Lifshitz did it using a sort of standard thermodynamic theory of fluctuations. It was formulated for real frequencies and that sort of...

### PoS

Was it you and Pitaevskii who pushed Lifshitz to reformulate it with a problem quantum--

### IED

No, we did it absolutely independently of him. Then we wrote an essential paper, we wrote the first paper without him, and proved how all the calculations should be done. The main thing was -- in Lifshitz' work, he was able to calculate only the force between two bodies divided by vacuum. I'll tell you why, it's not that simple. If you put a liquid here [in between the bodies], then he calculated the force by simply taking Maxwell's stress tensor in vacuum. It exists, but there is no Maxwell stress tensor in absorbing liquid, or in any absorbing medium. There is no such concept. Basically the force in the medium, if you try to write it down for real frequencies and even in non-equilibrium state, you cannot express the force in terms of dielectric functions. Actually, you may do it, and Pitaevskii did it for some special cases in plasma, but you immediately see that the equivalent tensor is expressed not only through the dielectric function of the same plasma, but separately there were some mean free paths or something, so it's not the same stuff. We worked in the Matsubara case, and so we derived everything from thermodynamics, and we defined the force, and we did everything. Then we showed that you may write down the force as given by Maxwell's tensor, but written for an imaginary frequency. But it's wrong for real frequencies. If you move back, then you will not obtain this. Then we were able to calculate the chemical potential on the field and the wall. It is impossible to do it Lifshitz' way. But that's a simple-minded fluctuation theory. In a sense it means that if your fluctuations are far from equilibrium, there is no way to write down -- nothing is expressed through dielectrics and that sort of thing. So you cannot reduce it. You may calculate it, but nothing results. So that was what we did. The film [liquid films] was our main issue. Now I simply cannot adjust myself to what people are doing now, what they call Casimir forces. It's against my grain. I mean in astrophysics. The point is obviously cutoffs. But you need cutoffs only in Casimir's initial theory, when you assume that your walls are ideal metals. Then you need cutoffs. But if your walls are not -- and indeed in our condensed matter, they are not -- all my epsilons here, at high enough frequencies they go to 1. So to me the physical phenomena -- the results are so ambiguous, because of these cutoff problems. But to get rid of them, they need a theory. You build your own problems.

### PoS

I'm interested in hearing a little bit more about -- you have really a sort of intimate cohort there at the Institute of Physical Problems. For a time, there's just the six of you working with Landau: Lifshitz, you, Abrikosov, Gor'kov, Khalatnikov, Pitaevskii (and some graduate students).

### IED

That was basically the four of us. We were all basically the same age, with Pitaevskii [b. 1933] as the youngest, Abrikosov [b. 1928] as the oldest. There was a difference of five years in this case, but we were all the same generation. Fundamentally we did not always work together, but we discussed everything.

### PoS

So when each of you was finishing a paper...

### IED

Absolutely. Well, no, before -- and sometimes we discussed subjects even before we started to work on them. We were not afraid of one another. At the top we had a good policeman. Landau would not tolerate any cheating in this way. This law was severely enforced when we were young.

### PoS

You explain that very nicely in your essay when you talk about how "demokratizm" was very strictly enforced in Landau's seminar.

### IED

Yes, democracy was on the surface. Otherwise, I don't know...

### PoS

Can you say a little bit more about the relationship between the Theory Department and the other departments of the Institute of Physical Problems?

### IED

It was good. I mean I interacted a lot with people in the Kapitza institute who worked on magnetism and that sort of thing. I would never try to solve an experimental problem, but I was aware of what was going on around me, of their problems, and sometimes it happened that -- I do not know why -- I found the solution. The majority of the experimentalists, they worked on superconductivity. I never did [this]; they had contact with Abrikosov and Gor'kov. Pitaevskii returned to the institute. There was somehow not a position for him after he completed the PhD, and then he [officially] moved to the institute, which was concerned with plasma in the atmosphere, so he got work in this way. But then Kapitza himself was researching what he considered a way to control fusion. He was absolutely wrong, but for the last thirty years of his life he was doing this stuff, and he simply did not listen. He was absolutely convinced that the temperature... He heated his plasma by pumping high-frequency radiation into it, with no magnetic confinement, and then he was convinced somehow that the pumped energy had some relation to the ion's temperature. He was stubborn until he died. After his death his lab simply "blew up" and disappeared. But he employed Pitaevskii--

### PoS

--to his credit.

### IED

By that time Lev [Pitaevskii] became expert in plasma...

### PoS

Petr Leonidovich Kapitza's successor is Andrei Stanlislavovich Borovik-Romanov? At an earlier stage you had discussed ferromagnetism with him. Tell me a bit about that collaboration.

### IED

Yes, I collaborated with him, on weak ferromagnetism and piezomagnetism. {See Dzyaloshinskii, "Thermodynamic theory of 'weak' ferromagnetism in antiferromagnetic substances," Sov. Phys. JETP 5 (1957): 1259-1272; "The problem of piezomagnetism," Sov. Phys. JETP 6 (1958): 621-622.}

### PoS

What kinds of things did you ask of each other?

### IED

Well, I was strange, I never interacted in this way. He found weak ferromagnetism, and he could not explain it, and nobody at that time could, and I was available. I tried the method to work with this problem, the general problem of magnetic spin arrangement and that sort of stuff: how to analyze the possible symmetries. So then I applied them to different compounds, and accidentally I applied it -- well, not completely accidentally -- but I applied it to hematite, which was the first known weak ferromagnet, and then the only important one. It's my first long paper.

### IED

Yes, that's my thesis. Then I used a symmetry concept to analyze the possible magnetic--

### PoS

May I ask, who were your opponents for your kandidatskaia defense?

### IED

Oh, I don't remember.

### PoS

You would have had outside opponents, right?

### IED

Yes, two, there should be two. I don't remember, and I'll tell you why: because I did not give a talk [an oral defense].

### PoS

It was more a question of whether they signed a piece of paper saying, "Yes, this is good work."

### IED

Yes, they signed it. I'm sure that one of them was [Sergei] Vonsovskii, I'm not sure who the second was. It might have been Borovik-Romanov himself, I have forgotten. It was forty three years ago. In a sense the defense was not important to me. I was already engaged in other work, and we were occupied.

### PoS

And it was already clear at the time that you submitted this work for publication that you were going to stay on as a researcher at the institute, is that right?

### IED

Here, look [points to photo hanging on wall], that's the funny Landau picture. It's related to this work. The picture is from Life magazine, 1957. The situation: it was a small room, with a small blackboard. It was the first time I really told Landau about this work, and I covered the board with my writing. Then this guy with cameras appeared, so in fact he staged the picture. He put Landau in front [at the blackboard], and I'm here [in back]. The caption here is "Landau teaches at the Institute for Physical Problems." {See Robert Wallace, "First hard facts on all Russian sciences," Life Magazine 43 (16 December 1957): 118.}

### PoS

Even though he hadn't written the equations...

### IED

Yes, that's my writing. I'm about 26 at the time. {End of 14/12/01 session}

Continue reading part III of this interview, or go back to part I.