FOM: Why *should* infinitesimals be definable?

Moshe' Machover moshe.machover at
Thu Nov 27 18:44:13 EST 1997


I can't quite understand why some people are so keen to have a *definable*
infinitesimal. Why *should* there be one?

In many years of teaching Nonstandard Analysis to maths and fom graduates,
I've never been asked  by a student to provide an example of an
infinitesimal. But if I were to be asked, I would say: I *can't* give you
an actual example. These things do not really exist; they are just
fictions--but very, very useful fictions.

		`...  je leur te'moignai, que je ne croyais point qu'il y
		des grandeurs ve'ritablement infinies et ve'ritablement
		infinite'simales, que ce n'e'taient que des fictions mais des
		fictions utiles pour abre'ger et pour parler

To `abre'ger' I would add also `simplifier'.  Of course Harvey Friedman is
right: infinitesimals are as undefinabe as non-principal ultrafilter. In
fact, infinitesimals *are* ultrafilters in disguise (and the non-szero
ones are non-principal). But the beauty of this disguise is that if we
count points as objects of type 0 then ultrafilters are of type 2 (sets of
set of points); while infinitesimals are *points, ie *objects of type 0.

This is why, as I mentioned in a previous posting, the definition of
compactness which is an AE^2 statement can be reduced in NSA to a *AE^0

	` ...  de sorte qu'on ne diffe`re du style d'Archime`de [ie
	delta-epsilon] que dans les expressions, qui sont plus directes dans
	notre me'thode et plus conformes a` l'art d'inventer.'

  %%  Moshe' Machover                 | E-MAIL: moshe.machover at %%
  %%  Department of Philosophy        | FAX (office)*: +44 171 873 2270  %%
  %%  King's College, London          | PHONE (home)*: +44 181 969 5356  %%
  %%  Strand                          |                                  %%
  %%  London WC2R 2LS                 |  * If calling from UK, replace   %%
  %%  England                         |    +44 by 0                      %%

More information about the FOM mailing list