Dear Michel, Jerry and Loet,
Welcome back to the fray, Jerry, but I recall a kind of "gentlemen's
agreement" we made at our meeting in Liège, namely, that I can find a place
for your theory, but you should reciprocally find a place for mine! In the
following, I will try to disentangle two major issues in the recent
exchanges.
1. Jerry's theory of Perplex Numbers, underlying his comments, is not a
physical theory. It is a model derived from some of the numerical
characteristics of the atomic structure of elements due in reality to
underlying physical constraints (e.g., the Pauli Exclusion Principle).
2. Mathematics captures some of the "essential features of the information
content of chemical structures" but by no means all of them. Are the
dynamics of atomic and chemical structures, and their potential for reaction
not also information?
3. There is no problem in talking about "parity of iconic representations"
as irregular, but if you say electrical, you bring in physics, the iconic
representations are no longer applicable, and "modern chemical logic and
category theory" are no longer adequate.
4. No "practice of mathematics" or proof theory could have applied to the
results of my own research nor could apply to recent major advances in, say,
organometallic catalytic chemistry (see any recent issue of SCIENCE).
Combinatorial chemistry and its efficacy for screening, in which I see Jerry
was personally successful, is only one, limited domain of chemistry.
5. Jerry's critique of Loet is perhaps justified, and I will pass on the
debate as whether chemoinformatics is a part of information theory or not.
My view is that talking about the "identity of matter" and three-tailed
Peircean graphs is diversionary. Jerry understates Rosen's contribution,
even if he is correct about the chemoinformatics aspects. Rosen's work is
valuable because his vision went beyond thermodynamic considerations to
concepts like anticipation which underlie some current systems approaches.
6. To conclude, the "physical basis of chemical logic" may be well
understood, but this "chemical logic" is an abstract, partial model of what
is going on. It cannot be an adequate basis for the informational processes
that occur in real chemical systems.
7. Loet and I can get back to a "debate" about morphology and information
theory on other grounds. As a reminder, on Oct. 14 Loet wrote: "It seems to
me that the issue of morphology and its evolution is a red herring in a
discussion about information theory. A shape (e.g., a network) can be
described as a graph or also numerically. This numerical description can
easily be evaluated in terms of information theory. Information theory, also
offers options to develop measures for the evolution over time (such as,
Kullback-Leibler divergence, cf. Theil (1972).)" This statement implies that
morphology or shapes cannot be dynamic processes and, again, if not fully
describable mathematically are "lost" to information theory. This takes us
back to the question of the primacy of quantitative over qualitative
properties, or, better, over qualitative + quantitative properties. This for
me is the real area for discussion, and points to the need for both lines
being pursued, without excluding either.
Thank you and best wishes,
Joseph
----- Original Message -----
From: "Michel Petitjean" <petitjean.chi...@gmail.com>
To: <fis@listas.unizar.es>
Sent: Monday, October 17, 2011 1:39 PM
Subject: Re: [Fis] Chemo-informatics as the source of morphogenesis - both
practical and logical.
Dear Loet and dear Jerry,
2011/10/17 Loet Leydesdorff <l...@leydesdorff.net>:
> Dear Jerry,
>> ...
> It may be easiest to raise some questions:
>
> 1. What is the equivalent in chemo-informatics of a bit of information?
> Can
> this be operationalized as a formula like Shannon's H?
> 2. Can one compute with this formula in fields other than chemistry? For
> example, in economics; without using metaphors? ("As if")
> ...
If (1) can be answered, thus chemoinformation enters in the field of
information theory. That would be a very strong result.
Alas, I am afraid that it can't. Sets of flexible 3D realized graphs
seem hard to give raise ti bits of information.
But I didn't proved that. Who knows, if a good mathematician can
answer to (1), it would be a great advance in the field.
And I did not speak about (2) ...
Best,
Michel.
_______________________________________________
fis mailing list
fis@listas.unizar.es
https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
_______________________________________________
fis mailing list
fis@listas.unizar.es
https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis
