Letter ##1: 15 January 2004 (partial)
Dear Dr. Rondoni,
.......................
Please note that I am asking for a clear statement about your opinions
or results on OUR paper or that you attribute to us .....
Please note that from now on I WILL MAKE PUBLIC your answers via my
web page: therefore, please, write in english.
...........
Sincerely GG
Letter ##2: 15 January 2004 (verbatim, reply to 1)
From: Lamberto Rondoni
Dear Giovanni,
I hope this will help clarify the situation once and for all.
If it doesn't, and I think e-mail is not the best way of
solving this problem, we will have to think of other ways of
proceeding. I take this opportunity, however, to apologize to
anyone to whom I may have caused any incovenience.
>>From your letter of yesterday evening (in italian) and from at least
> another letter (in english) that you wrote yesterday I understand
>
> 1) you have withdrawn your cond-mat/0312353 paper from the archive
This is correct.
> 2) in spite of your acknowledgements (in italian) that all the
> statements in cond-mat/0312353 about the work of
> Prof. Cohen and me that I had challenged were, in fact, incorrect
> you are keeping the submission of cond-mat/0312353 to PRE running
> and waiting for referee reports
Yes. The journal indicates that the referees have recevied the paper
some time ago. They most likely have already noticed and written up
things which we would prefer to avoid in the future. This would spare
the referees from working twice on the same issues. I do not think
that it would be a responsible thing to withdraw the paper from the
journal at this stage. But that paper will not be resubmitted. I
made this clear in our correspondence.
However, I must explain that I never acknowledged that the content
of our paper was incorrect. That may well be the case, and I am
ready to acknwoledge it, if that is the case.
I acknowledged the fact that the paper was incorrectly formulated.
A new paper should express better our views, and then it will become
clear whether the content of this paper is correct or incorrect.
> 3) unlike what you wrote me, you are still convinced that your paper
> is "basically correct" and that you derive this feeling (? or
> certainty ?) from a discussion with Prof. Ruelle, as you wrote
> yesterday.
> In fact I see no other reason to keep the paper under review
I explained why we keep the paper under review. It is because we believe
that it has been partly or fully reviewed. However, I have explained in
my past e-mails that this paper will not be resubmitted. A new paper is
being written now.
Yes, I got the impression that the content of our (incorrectly formulated)
paper may be correct. I got this from various sources, including my
correspondence with Prof. Ruelle. If I misunderstood anyone's remarks, I
am sorry. But, certainly, I did not want to use their words to claim that
my paper is correct, especially if they don't think so. I only expressed
what I believe from what I have gathered. I mentioned my correspondence
with Prof. Ruelle, in my e-mail to Prof. Cohen, only because he knew of
it. But I did not say that Prof. Ruelle had told me that the paper is
correct. If I used Prof. Ruelle's words improperly, that was
unintentional, and I am sorry for that.
Whether the content of this paper is correct or not, will have to be judged
when the, hopefully correctly formulated version, will come out.
> I am, of course, very interested in discussing your claim with
> Prof. Ruelle himself. I am at the moment in Paris and this is an
> opportunity that I do not want to miss. Particularly as you are asking
> me (in italian) to help you in making your paper "less offensive" (as
> you may imagine I think instead that, after your own admissions (in
> italian), a more appropriate wording would be "less wrong" also
> because I am not at all offended).
This is a relief for me. I thought otherwise.
> Please send me the exact conclusions that you claim to have followed
> from your correspondence with Prof. Ruelle which make your statements
> about the work by Prof. Cohen and me, or by me, still "basically
> correct" in your paper which, as the title itself makes clear, is
> mainly about our work.
My impression is that there are some molecular dynamics models which
do not satisfy the Chaotic Hypothesis or that do not satisfy the
fluctuation relation of the Gallavotti-Cohen theorem. To explain this
point is the only purpose of our paper, and the idea comes in part from
my correspondence with Prof. Ruelle. But it has other motivations as well.
This impression may be right or wrong, but I have been told that the old
paper didn't make this point. I was told that that paper claimed we had
found mistakes in the existing proofs of the Gallavotti-Cohen theorem.
This was not the point we wanted to make. Therefore, we are preparing a
new paper for the scientific community to judge. If our point turns out
to be wrong, I will gladly acknowledge that.
> After the letters that you sent around (at least those that, once more
> being in italian, you sent only to an apparently selected audience) I
> am confident that you realize that if your paper is "basically
> correct" (as you write to some, in english) this means that the work
> by Prof. Cohen and me is basically wrong: I refer for instance to the
> comments in the abstract and to the statements in appendix B (only to
> fix the ideas, as you know from my previous letters).
I have sent the e-mails in italian only doing a reply-all. If I
should have not done so, I am really very sorry. But I thought it
would be all right. I did not see any reason for sending these e-mails
to others. But of course, as acknowledged also in my e-mails, I had had
some correspondence with others.
In any event, I do not understand why you say that if I am right, you
must be wrong. I don't see any reason for that. You, Prof. Cohen, Prof.
Ruelle and others have proven a theorem under some assumptions. The
theorem is definitely correct. But I ask the question of which systems
fit the assumptions, and which do not. That is all.
I don't see how this could imply that you did any mistake.
Now, I acknowledged that the old paper did not say this correctly, that
its formulation could upset and offend some people. I apologized, and
apologize for that. It was wrong on my side to write something that
suggested that you made any mistake. For that reason, the paper has been
withdrawn, and will not be resubmitted. If there is anything else I should
do, please let me know.
> Please note that I am asking for a clear statement about your opinions
> or results on OUR paper or that you attribute to us: the SAME
> comments, PLEASE, that could be presented to me, to prof. Cohen, to
> prof. Ruelle as well as to interested colleagues. Science should not
> depend on the language one uses.
I agree. Your results are correct. If I suggested a different view,
that is because I did not express my views correctly.
I hope I have been clear enough now.
> Please note that from now on I WILL MAKE PUBLIC your answers via my
> web page: therefore, please, write in english. I feel forced to do so
> as you write different statements to different colleagues (in
> different languages) and confusion is increasing and most people
> cannot take the time to check what is true and what not.
This is fine with me. I hope I made clear the source of the
misunderstandings. In a few words, what I wanted to say was:
You did not make any mistake. I agree that my paper was incorrectly
formulated, but, at the same time, I believe that the point in it
is correct. When the new paper will be ready, you and anyone will
be able to judge.
I hope this helps
My best regards
Lamberto
--------------------------------
Lettera ##3: From Giovanni Gallavotti, 15 gennaio 2004
Dear Dr. Rondoni:
I pick up two points in your letter below which refer to your paper
0312353.pdf
(still submitted to PRE)
>Whether the content of this paper is correct or not, will have to be judged
>when the, hopefully correctly formulated version, will come out.
I say that if you are right our work must be wrong because you say in
the abstract that our theorem predicts an unsymmetric relation in the
hamiltonian case. This can be judged now.
I say that if you are right our work must be wrong because you
attribute to the papers [4] a result, p. 30, that is the basis of the
criticism mentioned in the abstract.
>However, I must explain that I never acknowledged that the content
>of our paper was incorrect. That may well be the case, and I am
>ready to acknwoledge it, if that is the case.
therefore I say that if you believe, as you write, that the content of your
paper
is still correct then we are wrong (see above).
Question:
Do you now acknowledge that the abstract of your paper and your related
claims
about our work are wrong ? I thought you had done so in your letters in
italian:
if so little is left of the paper if I understand it.
I am sorry to ask that: but you distributed your paper
to the world stressing that we were predicting things we did not! and after 20
days
you still do not want to acknowledge that; and in fact you keep your paper
under refereeing.
Sincerely: GGallavotti
ps I attach a copy of your paper where you can check my statements.
Lettera ##4: (From F. Bonetto)
Dear all
I'm trying to follow the discussion. I do agree with Giovanni that the
paper is written in a terrible way and that many of the statement it
makes are wrong. On top of this the work of GC is misquoted and the use
of term like theorem or hypothesis is highly confusing.
I just want to try to understand what was the point of the authors in
writing this paper. They are, in my opinion, good scientists and so I
belive they had an idea that they probably expressed in a wrong and
confusing way may be helped by some personal animosity.
In what follows I try to paraphrase their point as I understand it.
Please correct me if I make mistakes.
The FT, has any theorem, is composed by some hyptheses from which a
thesis follows. The hypotheses are chaoticity (Anosov), reversibility
and that the large deviation functional for the phase space contraction
is analytic in the intervall [-p^*,p^*] and the average of the large
phase space contraction is positive. The corresponding thesis is the
usual large fluctuation simmetry.
Clearly the thesis of a theorem can be verified also when the hypotheses
are not verified. This is the content of many numerical works where one
sees that the fluctuation relation (so I call the thesis of the FT)
holds in systems that are surely not anosov (in the mathematical
acception of the term). E.g. the several biliards we studied where the
dynamics is not even continuous.
The above mentioned numerical results have been used to "invert" (note
the quotation mark) the FT. I mean that the fact that the fluctuation
relation (FR) is verified is an inderect evidence that the systems under
consideration behave as if they were Anosov. This in turn give
confidence that the theses of other theorems (e.g. GK formula) proven in
the Anosov setting (which hipotheses are not verified for the systems
under consideration) may turn out to be verified. This more or less what
the authors call Chaotic Hypothesis.
Thus it may be interesting to look for systems that although chaotic in
some weak sense (e.g. existence of one positive Lyapunov exponent) do
not satisfy the FR. Note that asking whether the FT is correct for this
systems makes no logical sense. The author start with the question: what
about a reversible Anosov system where the average phase space
contraction is 0 but the sup of the time average of the phase space
contraction on a trajectory of lenght t is bounded away from 0
uniformely in t. This mean that we can find points for which the time
average of the phase space contraction for an infinite time on the
trajctory issuing from that point is non 0. The argument goes on saying
that, from the proof of the theorem (? which?) follows that one can
identify p^* with the above mentioned maximum of the time average of the
phase space contraction. This seems to lead to a contradiction.
Clearly no contradiction is present. I think that this is due to theorem
6.4.3 of the book by Gallavotti Gentile and myself from wich it follows
(I belive) that no such system can exists. But it is also possible that
the above identification of p^* with the maximum time averaged entropy
production is not correct. In any case it would be interesting to see if
such Anosov systems exists.
Clearly if one relaxes the Anosov condition it is easy to find example
where the maximum of the time averaged entropy production is non 0,
endeed infinite, but the average is 0. This is for example the case I
discussed in my previous mail where you have 2 particles in a biliard
interacting through a potential and an isokinetic thermostat. In this
case the average phase space production on a trajectory is the potential
difference between the initial and the final point. This implyes that
the probability of seeing an average phase space contraction larger that
a given S is given by the SRB probability of the set of points where the
potential is larger than S. If the potential is not bounded any value S
can be observed. Moreover one can add an external electric field (and
some obstacles to avoid trivial motion) to the two particle to create a
non zero phase space contraction. These seem to be the systems that the
authors have in mind.
These systems represent flows that are clearly not Anosov. Indeed the
dynamics is syngular (as it immediatly follows form the fact that a
regular function on a compact set is bounded). Is the FR (again not the
FT, the systems are not Anosov) verified for these systems? the question
is interesting and seems that is has been addressed in some numerical
works with a negative answer. This is justified by the fact that the FR
cannot hold at 0 external field so that it will not hols at small
external field. I had no time to check those works but I'm a little
surprised for the following reason. I can chose as a poincare' section
for the above flows the surface defined by the fact that the to
particles are at a given (small distance) and their distance is
increasing. I can call such a situation an (after)collision. If I look
at the dymanical system generated by the intersection of the flow with
this poincare section I have that the phase space contraction between
two collisions is always bounded. I would be very surprised if the FR
resulted to be wrong for this (descrete time) dynamical system. This
should be easy to check to whoever did the previous simulation.
My impression is that the presence of the singularitites make it
necessary to go to very long time to see the FR for the above systems in
continuum time. And the lenght of time needed diverges with the
smallness of the external field. I think I have an argument for this
that I can give you if you are interested. So I tend to think that the
difficulties in seeing the FR in the numerical work for the above system
are due to simulation not long enough. But this is only an intuition.
I'm sorry for the long mail but is was useful at least to me to
understand discussion. I would like to ask Lamberto if he agrees that
this is the content of their paper, at least very schematically.
Best
Federico
P.S.: I think we used the two different expression FT (fluctuation
theorem) and FR (fluctuation relation) in the paper with Joel and Kolia
really to avoid some of the confusions we are discussing about now.
Letter ##5: From Dr. Rondoni Fri Jan 16 10:23:08
Dear Giovanni,
I think Federico's e-mail clarifies the situation in all its
aspects. I only repeat once more that our paper has been
withdrawn because we understood, thanks to your observations,
and those of others, that it made incorrect statements (even
if it was not our intention to do that).
My best regards
Lamberto
Lettera ##6: Reply to #5, 16 Jan 2004
Dear Dr. Rondoni
NOTHING IS CLARIFIED!! In fact the letter by Bonetto shows that you are
wrong if you agree with it.
PLEASE answer the questions: I asked. Is your paper correct in
attributing the theorem in appendix B to GC?
If not is the theorem correct?
I claim that is is NOT. If so your paper is wrong (and I do not
understand why you keep its submission)
otherwise my work is wrong. If it is correct but not proved by us please
tell me where is it proved.
You cannot play with the words: your paper has not been withdrawn; it is
still submitted to PRE. Please
do a scientific discussion based on logical assertions. After you wrote
that all my work is wrong I have the right to know why.
Sincerely: GGallavotti
Letter ##6: 17 Jan 2004 (answer to the previous + my reply)
Dear Giovanni,
I do not know what to say more. If the paper has been withdrawn
it does not exist any more. You insist to say that this nonexisitng
paper told that all your work is wrong. So let me make it clear:
YOUR WORK IS RIGHT
Clearly, a different interpretation of my paper was possible.
Therefore, the paper has been withdrawn. I have also acknowledged
that I made a mistake in not realizing that our paper could be
read the way you did. I already said that it made incorrect
statements.
I really don't know what else I could say about a nonexisting
paper.
Now, you insist that the paper exists because is under review.
I explained that we don't to waste the referees' time. We think
this is the most responsible attitude. But we may be wrong. If
you have something better to suggest, please let us know.
I hope you are not bothered by the thought that we may not
do as we promised, resubmitting the paper.
My best regards
Lamberto
-------------------------------
Reply
Dear Dr. Rondoni
Please ANSWER (once and for all) the questions I asked:
you are avoiding them.
You write that we have proved a theorem (Appendix B):
is the theorem really proved by us?
where?
if not by you? and
where is the proof?
The theorem is WRONG as I wrote you, and it is the basis
of many critiques to the work of CG in your paper (eg abstract).
As a consequence I want to know why you say that you
have not made mistakes and why you believe that CG is correct if you
think that it contains that theorem. PLEASE answer.
I am afraid that I do not understand why you keep the paper on PRE if
it is
wrong.
Sincerely: GGallavotti
ps: the paper HAS NOT been withdrawn: as you say it is still
submitted to PRE! please use language appropriately
In any event it is still on my web page.
Letter ##7: Friday 16 January 2004 and answer
Dear Giovanni,
I really hate to upset you. Simply I thought that I had answered your
questions. That's all. I'll try again.
It is true that in the paper we have discussed, the terms Gallavotti-Cohen
FluctuationTheorem, or GCFT, were misused. When they appeared in the
text they incorrectly referred to your theorem. They meant to refer to your
theorem, but your theorem had not been correctly reported in that paper.
Had that paper been formulated more precisely, it would have been obvious
that it contained no critique of your work.
I have acknowledged that to read that paper as a critique of your work
was possible. But the paper has also been withdrawn from the archives.
As to the PRE submission, I explained my reasons. But I am ready to
listen, if you have better suggestions. As far as I am concerned, it
would be irresponsible to withdraw the paper from PRE at this stage.
However, if you strongly feel that it would be better to withdraw it as
soon as possible from PRE, I will consider this option with my
co-authors.
I hope this answers your questions
Best regards
Lamberto
----------------------------
Answer
Dear Dr. Rondoni
I have no suggestion on what you should do about your paper. In fact I really
hope that your paper will be published as it is now so that I can comment on
it in front of a wider audience. I only mentioned the improper use of the
word "withdraw" because you seem to be playing with words (here as well as
elsewhere) and I find that insulting.
What I am concerned with is: YOU DO NOT ANSWER.
You say that your paper is correct: hence the theorem in appendix B that you
atrribute to us. But that theorem is false. I asked
1) where did we prove the theorem?
2) if we did not then ou proved it? if so your paper s wrong and you should
acknowledge because that is the basis of your statements about our work?
PLEASE ANSWER once and for all. I am not upset: I think that you are not
behaving appropriately. I remind you that I claimed that your paper (hence a
large portion of it) is WRONG and that you denied that: hence I have the
right to an answer to the above questions.
Sincereley: GGallavotti
Lettera ##8: From rondoni@calvino.polito.it Sun Jan 18
Dear Giovanni,
>
I thought that what I told you clearly meant that the paper we discussed
must be wrong, one way or another (at least, this is my opinion).
In my previous e-mails, I did an effort to explain what I meant by wrong
in that specific case, addressing the issues that you had raised. For
instance, you asked about appendix B, and I acknowledged that incorrect
(wrong) statements reported in that appendix were incorrectly attributed
to you (being incorrect, of course, nobody could have proved them). This
was part of of my previous e-mail. In another e-mail you pointed out that
that paper accused you of having done some mistake; then I explained that
if that was the case, it was due to improper (wrong) formulation of our
ideas, as, in reality, you had done no mistake.
>
However, if you feel that this is playing with words, and you only want
to hear the word "wrong", with no explanation, that is also fine.
In my opinion, that paper was WRONG.
>
And indeed it has been withdrawn form the archives, and will not be
resubmitted to PRE.
I sincerely hope that I have satisfied you with my answer, this time.
My best regards
Lamberto
-----------------------------------------------
I stop the discussion here as I see no point continuing it: hopefully
this will have contributed to clarify the relationship between the
work of CG, criticized in the paper cond-mat/0312353 and others, see
also J. Stat. Phys.96,
1343--1349, 1999
GG