perm filename MODL.OLD[DIS,DBL] blob
sn#226052 filedate 1976-07-17 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 . SSSEC(Limitations of the Model of Math Research)
C00009 ENDMK
C⊗;
. SSSEC(Limitations of the Model of Math Research)
AM, like anything else in this world, is constrained by a mass of
assumptions. Most of these are "compiled" or interwoven into the very
fabric of AM, hence can't be tested by experiments on AM. AM is
built around a particular model of how mathematicians actually go
about doing their research. This model was derived from
introspection, but can be supported by quotes from Polya, Kershner,
Hadamard, Skemp, and many others. No attempt will be made to justify
any of these premises. Here is a simplified summary of that
information processing model for math theory formation:
.SKIP TO COLUMN 1;
.BREAK; BN; SINGLE SPACE PREFACE 0;
.COMMENT GROUP
. ONCE TURN OFF "α" TURN ON "∞→" ⊗8⊂∞α→⊃⊗*
. NARROW 2,2;
.BOXTOP(1); INDENT 2,6,3;
.MODELPAGE: PAGE;
.MODELSSEC: SSECNUM;
.MODELSSSEC: SSSECNUM;
.MODELSEC: SECNUM;
.GROUP SKIP 2;
λλ The order in which a math textbook presents a theory is almost the
exact opposite of the order in which it was actually discovered and
developed. In a text, new definitions are stated with little or no
motivation, and they turn out to be just the ones needed to state the
next big theorem, whose proof then magically appears. In contrast, a
mathematician doing research will examine some already-known
concepts, perhaps trying to find some regularity in experimental data
involving them. The patterns he notices are the conjectures he must
investigate further, and these relationships directly motivate him to
make new definitions.
.COMMENT APART;
λλ Each step the researcher takes while developing a new theory
involves choosing from a large set of "legal" alternatives -- that
is, searching. The key to keeping this from becoming a blind,
explosive search is the proper use of evaluation criteria. Each
mathematician uses his own personal heuristics to choose the "best"
alternative available.
λλ Non-formal criteria (aesthetic interestingness, inductive
inference from empirical evidence, analogy, and utility) are much
more important than formal deductive methods in developing
mathematically worthwhile theories, and in avoiding barren
diversions.
λλ Progress in ⊗4any⊗* field of mathematics demands much non-formal
heuristic expertise in ⊗4many⊗* different "nearby" mathematical
fields. So a broad, universal core of knowledge must be mastered
before any single theory can meaningfully be developed.
λλ It is sufficient (and pragmatically necessary) to have and use a
large set of informal heuristic rules. These rules direct the
researcher's next activities, depending on the current situation he
is in. These rules can be assumed to superimpose ideally: the
combined effect of several rules is just the sum of the individual
effects.
λλ The necessary heuristic rules are virtually the same in all
branches of mathematics, and at all levels of sophistication. Each
specialized field will have some of its own heuristics; those are
normally much more powerful than the general-purpose heuristics.
λλ For true understanding, the researcher should grasp$$
Have access
to, relate to, store, be able to manipulate, be able to answer
questions about $ each concept in several ways: declaratively,
abstractly, operationally, knowing when it is relevant, and as a
bunch of examples.
λλ Common metaphysical assumptions about nature and science: Nature
is fair, uniform, and regular. Coincidences have meaning.
Statistical considerations are valid when looking at mathematical
data.
.COMMENT WIDEN 2,2
. ONCE SELECT 8 TURN OFF "α" TURN ON "∞→" %∞α→$;
.BOXB;
.END;
.BEGIN NOFILL
.END
.ONCE TURN ON "{}"
THE VALUEOF LINE HERE IS {LINE}.
.ONCE TURN ON "{}"
THE VALUEOF TESTBOT (BOTTOMLINE) WAS {TESTBOT}.
<<Remove this note before final version!!>
<<Perhaps incorporate some of Iberall's metaphysics here, in final point.>