Re: [Phys-L] How to become a good theoretical physicist?

[John Denker <jsd@av8n.com>]

Before we go too far down this road, let me offer an
observation and a hypothesis:  For starters, note that
there are verrry few theoretical chemists.  With rare
exceptions (e.g. Roald Hoffmann), chemists are expected
to do their own experimenting /and/ their own theorizing.
Ditto for biology;  for example, Charles Darwin did 
plenty of fieldwork;  he was not an armchair theorist.

Also note that many of the people we think of as
theorists have good connections to experiment, and
to industry.  Planck invented quantum mechanics in
the course of answering a question that came from 
the light-bulb industry.

We should consider the hypothesis that the world would
be better off if schools made less of a distinction 
between theoretical physics, experimental physics, 
applied physics, et cetera.  In particular, Galileo
is called the father of modern physics, and indeed
the father of modern science, because he demonstrated
the tremendous power that comes from /combining/
experimentation with state-of-the-art mathematical
analysis and modeling.

I can't prove this hypothesis is true, but I've 
never seen any good evidence against it, so we 
ought to give it further consideration.

========

On 06/12/2014 01:47 AM, Savinainen Antti wrote:

> <http://www.staff.science.uu.nl/~Gadda001/goodtheorist/index.html>

I call particular attention to the question of why
there are so very few (if any!) self-taught physicists 
in academia:
  http://www.staff.science.uu.nl/~Gadda001/goodtheorist/responses&questions.html

First of all, I attribute most of this to the fact that
universities are not nearly as enlightened as they claim
to be.  Top-notch self-taught scientists exist, but they
tend to be affiliated with non-academic institutions, 
e.g. Michael Faraday, Oliver Heaviside, et cetera.

Combining this with the previous thought, we expect
there to be relatively few completely-unaffiliated 
physicists who are any good, because experimentation 
requires laboratory facilities and other resources
that are hard for a lone wolf to obtain.  Sometimes
self-taught wizards find a home in industry (Steinmetz)
or start their own industry (Edison, Westinghouse).

===============

IMHO the most important thing we should be teaching is
how to think, and how to learn.  That is, given that
most people /initially/ cannot teach themselves, the 
#1 goal is to get them to the point where they can.

> I would like to think that we can teach more in HS than "the childish
> anecdotes that they call science" .

Indeed!  Actually, I think that "childish anecdotes" 
would be an improvement.  In most schools I see around
here, the primary emphasis is on conformity and obedience,
i.e. /not/ thinking for yourself.  This is antithetical
to the aforementioned #1 goal, and is disqualifying for
ever becoming a good scientist (theoretical physicist or
otherwise).

This is not a new problem, or a US-only problem.  Einstein
had a word for it:  /Kadavergehorsamkeit/.

=============

Gerard 't Hooft's list is weird in a number of small
ways.  Listing probability as if it were a sub-topic 
of Gaussian integrals is weird:
  http://www.staff.science.uu.nl/~Gadda001/goodtheorist/primarymathematics.html
IMHO probability is so important that it deserves a
line-item unto itself.  If it must be lumped in with
something, it belongs with set theory.

It is weird to classify partial differential equations 
as "primary" mathematics, while postponing any mention
of vectors to "advanced" mathematics:
  http://www.staff.science.uu.nl/~Gadda001/goodtheorist/primarymathematics.html
  http://www.staff.science.uu.nl/~Gadda001/goodtheorist/advmath.html

The ideas of Lorentz contraction and time dilatation
are 100 years out of date.  Mentioning them is weird,
especially coming from a guy who does relativity for 
a living.
  http://www.staff.science.uu.nl/~Gadda001/goodtheorist/specialrel.html

The Bohr atom is almost 100 years out of date.  Mentioning
it is weird.

Not mentioning phase space or Liouville's theorem is
weird.  Not mentioning Noether's theorem is weird.

On a far more serious level, modern science is like a
three-legged stool, depending on experiment plus theory
plus numerical methods ... by which I mean to include
both simulation and visualization.  Numerical methods
have been super-important for several decades now, and
are becoming even more important with each passing day.
I mention this because IMHO 't Hooft grossly under-
emphasizes this.
  http://www.staff.science.uu.nl/~Gadda001/goodtheorist/compphys.html

If you want to see a REALLY GOOD physicist in action, I
recommend watching Kip Thorne give a colloquium:
  http://digitalassets.lib.berkeley.edu/physicscoll/ucb/video/col.streaming.11-14-11.mov
Note that doing physics involves a loop, including:
  a) hard-core analytic theorizing
  b) simulation and visualization
  c) more hard-core analytic theorizing
  d) and so on, iteratively

===============

The links to references are valuable, but I must disagree
with some of the choices, and some of the omissions.
 -- _The Feynman Lectures on Physics_ are online, free for
  all.  You can't pretend to be a physicist, let alone a
  "GOOD" physicist, unless you have read these.
          http://www.feynmanlectures.caltech.edu/
 -- Sussman & Wisdom 
  _Structure and Interpretation of Classical Mechanics_
  is online, free for all.  It handles correctly a lot of
  stuff that other books get wrong.
         http://groups.csail.mit.edu/mac/users/gjs/6946/sicm-html/book-Z-H-4.html
 -- 't Hooft mentions a couple of the Landau & Lifshitz 
  books.  Some people swear by these books, but mostly
  they just swear at them.  They seem to appeal mostly to
  people who already have a sophisticated understanding of
  the subject.  To say the same thing the other way, they
  are approximately the last thing I would recommend to
  anyone who was studying on their own.  As much as Feynman
  exudes style and panache, L&L exude the opposite thereof.
  They are dry without being rigorous.

===============

The biggest problem with 't Hooft's list is the lack of
anything resembling pedagogical sequencing.  The only thing
that has ever made sense to me is the /spiral approach/.
Within that approach, there are lots of reasonable ways 
of arranging the details ... but arranging the material
by topic is absolutely NOT a reasonable pedagogical
sequence.  That may be how an expert looks at the material,
but it is not how anybody learns the material.

Maybe an encyclopedia should be organized by topic, but 
learning should not.  Arranging things in a "logical"
sequence is a tremendous mistake, pedagogically speaking.
I've made this mistake often enough to know exactly how
bad it is.