Non-Standard Analysis and the Hyperreals

Branden Laske
Department of Mathematics
University of California, Santa Cruz
June 13, 2012
Abstract
Introducing the notion of a new branch in Mathematics, this idea of
Non-Standard analysis was rst introduced in the 1960s by Abraham
Robinson. The concept of his ideas was using this intuitively natural
idea of innitesimals and their appeal to our natural intuition and
understanding of numbers.These innitesimals were rst popularized
by Leibniz in the late 1600s along with the discovery of innitesi-
mal calculus. Our today understanding of these innitesimals exists
in our real number system as vanishing quanities. We were not
able to give a consistent theory in analysis of which included these
innitesimals until Robinson introduced a satisfactory way in doing
so. The simplest way in understand what Robinson did to achieve this
was using a semantic approach. This was basically the study of mod-
els and developing purely set-theoretic objects called superstructures
which construct a mapping from our real numbers to this extension
of the real numbers called the hyperreals. Think of these hyperre-
als as our real number system but include innitely small quanities
called infinitesimals and quantities which are innitely large, or a
largest value of which we will call infinite. To gain a sucient grasp
in understanding our new perspective in Analysis, we will begin by
introducing Leibniz and the innitesimals that greatly inuenced ad-
vancements into Non-standard Analysis.
1
1 Leibniz and the innitesimals
A philosopher, rationalist, and mathematician, Leibniz became an impor-
tant contributer to anything he set his mind to. Much of his mathematical
successes were in the development of innitesimal calculus, his work in me-
chanical calculators with the description of the pinwheel calculator, and the
renement of the binary number system. Leibniz saw through his work that
there was much importance in understanding this notion of innite and he
began relying on a principle called the Law of Continuity. This law states
that what succeeds for the nite numbers must also succeed for the innite
numbers. Although much of his work is relevant to todays mathematics, we
are only concerned with his work within innitesimals.
We want to rst dene what we mean by infinitesimal. We can introduce
our Archimedean property which denes some number x N and x = 0 as
an infinitesimal if it satises the following.

1
x

< 1,

1
x

<
1
1 + 1
,

1
x

<
1
1 + 1 + 1 . . .
, . . .
Again by the Archimedean property we want to give a similar denition
of what we mean by the value innite. Let us dene some number x as
being infinite if it satises the following.
|x| > 1, |x| > 1 + 1, |x| > 1 + 1 + 1, . . .
Innitesimals The reason Leibniz adopted this idea of very small values
was beause of its intuitive value. The idea of an innitely small value doesnt
seem like a very natural idea of thinking as humans. Lets consider a visual
representation of an innitesimally small value.
The basic idea behind this picture is an intuitive example of innitesimal.
We take triangles and plug in a triangle inside of the triangle. We continue
2
to do this innitely many times to construct innitely small triangles down
even smaller than the known existent sub atomic particles. In order to accept
this idea you must think Pseudo-Physically passed what we actual know in
science and see in reality.
Now that we have a good understanding of the innitesimals that inuenced
the work of Non-Standard Analysis we will now begin to study and gain a
grasp in understand the construction of what we call the hyperreals.
2 Robinson and the Hyperreals
First let us dene in words what is meant by a hyperreal. Think of it as
taking our real number line, the collection of all rational numbers, integers,
irrationals, and transcendentals, and extending it to include our two values
dened earlier, infinite and infinitesimal. This term hyper-real was rst
introduced by Edwin Hewitt in 1948 when we pioneered the construction of
the hyperreals. We will denote these hyperreals as

R and we can think of the
reals, R, as a subset of the hyperreals R

R. What we know from studying

the real number system is that we dene the reals as a totally ordered set. In
order for us to gain a better grasp of our understanding of these hyperreals,
let us go through and creat our real numbers as a eld.
3 Constructing an Ordered Field
In order for us to model our reals we want to dene a couple of simple de-
nitions and properties that our reals hold for all numbers.
First let us make it clear that our reals, R, form our algebraic system called
a eld which we dene as being closed under addition and multiplication.
We will skip the tedious facts about a eld and assume we know the small
details. Now that we have our eld we want to propose a couple of denitions
and properties to begin forming our totally ordered eld.
(i) Let X be a set and let be an ordering of X then the following
hold:
(1) a a
(2) a b and b a, then a = b
3
(3) a b and b c, then a c
Now that we have an ordering on our set X we can now introduce the fol-
lowing:
(ii) Let X be an ordered set with two operations + and so that our
set is totally ordered, then:
(1) For every a, b, c X, a b then (a + c) (b + c)
(2) For every a, b X 0 a and 0 b then 0 ab
Another way of stating this is saying our set R satises the axioms of an
algebraic eld, the order axioms, and the completeness axiom. Now that we
have obtained our totally ordered eld, we notice a crucial problem within
our axioms of our eld R with respect to our hyperreals

R.
Problem going beyond 1st Order If we look back at the completeness
of our eld (ii), we can dene this in a dierent way.
We can say our eld X is Archimedean by the Archimedean property:
(1) For every a X, b X such that a < b
In words, there is no innitely large or small elements within our eld X.
But from our denition of our hyperreals, it is the set of reals of which con-
sists of innite, our largest element, and innitesimal, our smallest element.
This is where Robinson stepped in and said lets take a minute, step back
and analyse our problem. He started to ask simple questions, specically,
what exactly we meant by a property within our eld X. He noticed
theres a heirarchy of properties within our eld. In which the property that
doesnt hold for our hyperreals, Archimedean depends on our SUBSETS
of X, not the entire eld X itself. Robinson then began to categorize these
properties into a table similar to this.
Property of type 0 - include distributive/associative/etc.
Property of type 1 - include subsets of X (archimedean/completeness/etc.)
Property of type 2 - include sets of subsets of X
Property of type 3 - include sets of sets of subsets of X
4
etc.
From this categorizing of properties by Robinson, we can easily conclude that
R shares type 0 with

R. For other property types to hold though, Robinson
went to construct the very important Transfer Principle which helps satisfy
such properties. In order for us to construct this principle we must begin by
changing the world we are working within. This will be explained more in
the next section.
4 Set Theory and our Superstructure
In order for us to construct a new world to work in we need to dene a
map which will help us map each object from our standar world onto our
non-standard world. We dene world as a very large collection of properties,
rules, and objects such as numbers, functions, sets, subsets, etc. One prob-
lem with directly mapping from our standard world onto our non-standard
world is this method would lead us into Category Theory where we would run
into many diculties. To avoid this problem we will work with a restricted
universe. In order to do this we want to work within set theory and the
construction of structures.
We will be working with what we call a superstructure which we will dene
in the following:
Let A be a set, P (A) is the powerset of A.
Let S

:= S for n = 1, 2, . . . dened inductively

S
n
:= S
o

P (S
n1
)

S :=

S
n
is a superstructure of S.
From this we dene elements of S as individuals or atoms;
elements of

S are entities.
Putting this denition into words we dene our structure,

S as a nite set
of rules, elements, etc. We dene a superstructure as a powerset of many
sets. So when we speak of values within our superstructure we are speaking
in terms of a union of structures which help construct our superstructure.
To better understand exactly what we mean by an entity, let us examine the
rules and properties which hold within entities.
5
(1) S
o
S
1
S
2
. . .

S and S
o
S
1
S
2
. . .

S. S
n
are entities
including .
(2) Each entity is transitive such that for

S if A is an entity and x A, then
x is either an entity or an atom.
(3) If A is an entity and B A, then B is also an entity.
(4) If A is an entity, then P (A) is an entity.
(5) Let A be a set of entities. If A = , then

A :=

xA
X is an entity. If
A

S, then

A :=

xA
X is an entity.
(6) If x
1
, . . . , x
k


S, then {x
1
, . . . , x
k
} is an entity.
(7) If A
1
, . . . , A
k
are entities, then

j
A
j
= A
1
. . . A
k
is an entity.
(8) If A
1
, . . . , A
k
are entities, then A
1
. . . A
n
is an entity.
(9) All functions acting between entities are themselves entities.
These entities are very fundamental once we approach our construction of
the Transfer Principle and dening our three fundamental principles which
will construct it. But rst we must also dene our formal language which
will be used within our new restricted universe,

S.
Formal Language Our formal language will be very similar to our normal
logic language used within our mathematics. We then must identify what
will be used within

S.
(i) Variables : {a, b, c} and they will be underlined to denote variables.
(ii) Constants : cns(L) which will denote the set of all constants.
(iii) Basic Predicates : and = which denote as an object in some set and
equality.
(iv) Seperation Symbols : ( and )
(v) Connectives : , , , ,
(vi) Quantiers : and
Our language will only be admissible to formulas that are well-formed formu-
las aka ws. Now that we have layed the structure we will be working with
along with the language, we can now begin to dene the Transfer Principle
which allows us to dene our non-standard world as a Totally Ordered Set.
6
5 The Transfer Principle
The idea of the Transfer Principle is to dene the mapping from our standard
world to our non-standard world so that what holds within our real numbers
also holds within our hyper-reals. We denote this mapping as:
Let :

S


S such that ;

S is our standard world so that

S


S. (1)
Basically we dene our standard world

S as being contained within the non-
standard world


S. In order for this denition to be true we must develope
it through three fundamental principles. These principles help us to make
use of a non-surjective interpretation maps to add non-standard elements to
our standard superstructure

S. This should be more clear as we dene some
important properties within this superstructure.
Standard Denition Principle For simplicity of things this principle
implies that the mapping * preserves much of the structure of

S. In other
words, * does not run into any diculties in terms of properties, equivalences,
structure, etc.
Internal Denition Principle We dene this principle as the following:
Denition: An entity A


S is internal i it can be written as
A = {x B : (x)}
Where B

S is an internal entity and is a transitively bounded internal
predicate with x as its only free variable.
In other words, A is considered internal if it contains a predicate , which
contains a free variable within any arbitrary entity, in this case B. This
arbitrary entity must be an entity of our standard world. We can understand
this by saying
Let A


S and A = (x) for some x B. Since B

S
A

S
A is internal since x A B

S


S.
Existence of External Sets We will denote

B as an external set. We
let

S contain an innite entity, is nonstandard i

B =

B some innitely
countable entity B

S. For any entity A

S the following hold.
7
(1) If A is innite, then

A is external.
(2) If A is innite, then P (

A) is external and

P (A)

P (A) P (

A)
(3) If S is innite, then

S\

S is non-empty and contains only elements

which are internal but not standard.
Now that we have constructed our Transfer Principle and our map holds for
the reals to the hyper-reals we now want to describe the non-standard map
more explicitly. To do so we must introduce special subsets aka lters, which
are similar to ideals. The special type of lters we will deal with are ones of
which properties within these lters either hold for almost everywhere within
our superstructure or almost nowhere. We dene such lters as ultralters.
To understand what we just said we will dene a lter with detail.
6 Constructing our Ultrapower
Filter: A set f of subsets of J is called a lter on J if it has the following
properties:
(1) If A f and A B J, then B f.
(2) If A, B f, then A B f
(3) / f
We also want to clearly state that these properties hold almost everywhere.
From this we get a theorem which says
Theorem: Each lter f is contained in some ultralter.
Ultrapower: We dene our Ultrapower on our superstructure

S as an equiv-
alence relation of our ultralter with respect to

S. In other words our Ul-
trapower embeds the properties of these ultralters onto our non-standard
world allowing us to dene our non-standard map as a preserved structure.
Therefore, we can now dene our non-standard map as a Totally Ordered
Set.
7 Hyperreals as a Totally Ordered Field
To conclude the method in which we took to be able to dene the Hyperreals
as a totally ordered eld:
Dene the Reals as ordered set Set theory and our superstructure
8
Dene transfer Principle Embed Ultralters Dene our Ultrapower
Results in our non-standard model
Its important to keep in mind that this process we took in produce an or-
dered eld for our hyperreals was an inconstructive approach. There are other
approaches to reach the same conclusion as we did but this process is the
best in understanding why we have to dene what we did. Its considered to
be almost constructive. We can think of the hyper-reals as implementing
an innitesimal-enriched continuum and the transfer principle implements
Leibnizs law of continuity. Arend Heyting being a strong believer in non-
standard analysis actual praised non-standard analysis as a standard model
of important mathematical research. He believed that in order to become
successful in contributing to mathematics community. Probably the most
notable eld of mathematics which innitesimals is implemented is calculus
but there hasnt been much use of them since then. So this is considered
a huge advancement in the use of innitesimals in advanced mathematics.
Robinson implementing the hyper-reals into analysis will allow us to discover
many hidden secrets not yet revealed within our mathematics.
9
References
Abraham Robinson. Wikipedia. Wikimedia Foundation, 25 Apr. 2012.
Web. 13 June 2012. http://en.wikipedia.org/wiki/Abraham robinson.
Farlow, S.J The Complete Ordered Field: The Real Numbers, Univer-
sity of Maine. http://math.umaine.edu/ farlow/sec52.pdf
Innitesimal. Wikipedia. Wikimedia Foundation, 06 Dec. 2012. Web.
13 June 2012. http://en.wikipedia.org/wiki/Innitesimal.
Non-standard Analysis. Wikipedia. Wikimedia Foundation, 06 Dec.
2012. Web. 13 June 2012. http://en.wikipedia.org/wiki/Non-standard anal-
ysis.
Vath, Martin. Nonstandard Analysis. Basel: Birkhauser, 2007. Print.
10