abstractmath.org
GLOSSARY
To generalize a mathematical
concept C is to find a concept C' with
the property that instances of C are also instances of C'.
a) , for arbitrary positive integer n, is a generalization of . One replaces the ordered pairs in by ordered n-tuples, and much of the arithmetic and spatial structure (except for the representation of using complex numbers) and even some of our intuitions carry over to the more general case.
b) The concept of abstract vector space is a generalization of . To get it you forget that the elements of are n-tuples of real numbers and your produce a different, more abstract definition obtained by finding properties of such as the existence of addition, multiplication, scalar multiplication, and so on, and using these properties as axioms for the new concept of vector space.
Example (a) is an example of an expansive generalization, obtained by changing a datum in the definition of a concept (the dimension 2 in this case) into a parameter.
Example (b) is
obtained taking the concept and reconstructing
it, producing a
very different-looking definition that includes all the original examples and
others as well. This is reconstructive generalization.
Both expansive
and reconstructive generalizations, if done correctly, take a concept and introduce
another concept that includes all the examples of the original concepts and (in
general) others as well. These are legitimate
generalizations.
It can happen that all the examples you know of a specific concept have a certain property P not required by the definition. If you conclude from this that all the examples of the concept have property P you are engaging in generalization from examples.
Generalization from examples is an illegitimate mathematical method.
It does not always give true statements about math objects.
The statement “An infinite sequence gets close to its limit but never reaches it” is a MYTH. Many newcomers to abstract math believe it because in the typical examples you see of limits of sequences the elements of the sequence in fact are never equal to the limit. For example
but 1/n is never equal
to 0.
On the other hand,
but every fourth term is zero. The sequence starts out like this:
The names “expansive” and “reconstructive” are due to Harel and Tall.
The usual meaning of the word “generalization” in colloquial American English is generalization from examples, often used in a derogatory way. When mathematicians refer to a generalization it usually means an expansive or reconstructive generalization and has no derogatory intent. See cognitive dissonance..
Give is used in several ways in math English. often with the same sense it would be used in any academic text (“we give a proof...
”, “we give a construction ... ”).
“We give a proof…”
This is seen in general academic prose.
To “give a object” means to describe it sufficiently that it is
uniquely determined. A phrase of the form “give an X such that P”
means describe a object of type X that
satisfies predicate P.
“Problem: Give a function of x that is positive at x = 0.”
Correct answers to this problem could be “the
cosine function” or “the function ” .
You could provide an answer for the problem in the preceding example by
saying:
“the function given by f(x) = 2x
+ 1.”
“Given sets S and T, the intersection S ∩ T is the set of all objects that are elements of both S and T.”
This means “If S and T are sets…”
To glue two math objects together is to identify them. You make a Möbius strip by reversing the right edge of a rectangle and glueing it to the left edge, as described here. This means you identify (in the mathematical sense) the reversed right edge with the left edge. This is a metaphor that is also a literal description of what you do with a real rectangle made out of paper, too!
The word graph has two unrelated meanings in undergraduate
mathematics:
¨ The graph of a function.
¨ A (directed or
undirected) graph is a
structure consisting of nodes with directed or undirected edges that connect
the nodes, often subject to further conditions.
In both cases, the word graph is used both for the math object
itself and also for a drawing of (often only part of) the math object.