Conclusion
Why division by zero ($$\frac N0$$) is undefined in the reals
Short answer
Because our maths teacher said so/Because maths said so
Medium answer (Credit to Prometheus)
$$0 \times N=0$$ for any N
Therefore $$\frac N0$$ is undefined for any N
Longer answer (Credit to James R)
Using the axiom (actually it's partially correct, this is explained in the fundamental answer section)
$$0 \times N=0$$ (where N is any arbitrary number)
Assume zero has a multiplicative inverse (For newbies, assume we can divide by zero)
$$\frac 10$$
Multiply both sides by the inverse you get
$$0 \times N \times \frac 10=0 \times \frac 10$$
Now we run into the problem of $$\frac 10 \times 0$$
If we said
$$\frac 10 \times 0=1$$
Then the result is
$$N=1$$
This contradict the given condition that N is an arbitrary number
Therefore zero has no multiplicative inverse (i.e. We cannot divide by zero)
Bit longer answer (Credit to Alphanumeric)(Some portions inspired from Alphanumeric)
Fundamentally, there are only addition and multiplication.
Subtraction can be defined as follows:
E.g. Let x be a UNIQUE number such that
$$x+3=0$$ (Additive inverse of 3)
Now find
$$5+x$$
We know
$$5=2+3$$
Therefore
$$5+x=2+3+x$$
Since 3+x=0, therefore
$$5+x=2+3+x=2$$
In the above process, notice there is not even a single instance of -3
Division can be defined as follows:
E.g. Let y be a UNIQUE number such that
$$y \times 4=1$$ (Multiplicative inverse of 4)
Now evaluate
$$92 \times y$$
We know
$$92=23 \times 4$$
Therefore
$$92 \times y=23 \times 4 \times y$$
Since 4y=1, therefore
$$92 \times y=23 \times 4 \times y=23 \times 1=23$$
In the above process, notice there is not even a single instance of $$\frac 14$$
Now for the case of the n/0s:
E.g. $$\frac 10$$
Let q be a UNIQUE number such that
$$q \times 0=1$$
E.g. $$\frac 20$$
Let r be a UNIQUE number such that
$$r \times 0=2$$
Begin proof:
secret said:
Let q,r,s and h be four UNIQUE numbers such that
$$q \times 0=1$$ ---(1)
$$r \times 0=2$$ ---(2)
$$s \times r=1$$ ---(3)
$$h \times 2=1$$ ---(4)
secret said:
Multiply (3) by 0 and h and using the axiom $$0 \times N=0$$ (except when N=q or r as defined)
$$s \times r \times 0 \times h=1 \times 0 \times h$$
$$s \times 2 \times h=0$$ or $$s \times r \times 0=0$$
$$s \times 1=0$$ or $$s \times 2=0$$
$$s=0$$
For the other case, multiply by h
$$s \times 2 \times h=0 \times h$$
$$s \times 1=0$$
$$s=0$$
Therefore both case are equivalent
$$s=0$$ ---(5)
secret said:
Put (5) into (3)
$$s \times r=1$$
$$0 \times r=1$$
$$r \times 0=1$$---(*)
Now for the case of $$\frac 00$$:
Let d be a UNIQUE number such that
$$d \times 0=0$$
Note this is identical to the axiom
$$0 \times N=0$$
Therefore any d except q and r satisfy the above equation
As there are at least two distinct d which can give 0 when multiplied by 0
Therefore there is no UNIQUE d
Thus d does not exist in the reals
For (*), there are two interpretations:
1. Since
$$r \times 0=1$$ and $$r \times 0=2$$
It can be said that r is a number which when multiply by 0 equals 1 and 2 (or by modifying the definition of r, equal to any arbitrary number except 0). Therefore r is not a multiplicative inverse of 0 as it does not give a UNIQUE answer.
Since
$$q \times 0=1$$
Therefore there are at least two numbers which give 1 when multiplied by 0.
Therefore 0 has no multiplicative inverse
2. $$r \times 0=1$$
Using the axiom $$0 \times N=0$$
r cannot be anything other than q (As other numbers will give zero when multiplied by zero)
Therefore
$$q=r$$
(By modifying the definition of r, we can show $$\frac n0$$ are all equivalent to $$\frac 10$$ thus are all equivalent to each other for any n except when n=0. Therefore there are only two distinct types of division by zeros $$\frac n0$$ where n=/=0 and $$\frac 00$$)
Since we have shown
$$q=r$$
Therefore there are two numbers that can give 1 when multiplied by zero
Therefore neither q and r are multiplicative inverse of zero
Therefore 0 has no multiplicative inverse
Regardless of which interpretation you use, we still showed
Therefore there is no UNIQUE number n such that $$n \times 0=1$$
Therefore $$\frac n0$$ does not exist in the reals and remain undefined
Fundamental/TRUE answer(Credit to Alphanumeric)
No, they explicitly do not exist by construction. We want a number system with the following properties (among others)
Let S be some set of objects and # and @ be two binary operations.
- If x and y are in S then x@y and x#y are too
- If z is in S then x@(y@z) = (x@y)@z and likewise for #, so we don't need to write the brackets, x@y@z is well defined etc
- There is some e such that e#x = x = x#e for all x in S and likewise there is some f such that f@x = x = x@f for all x in S
- Commutativity in #, x#y = y#x
- Distributivity of @ over #, x@(y#z) = (x@y)#(x@z)
To give an example, just set # to + and @ to *. Then e=0 and f=1, 0+x=x=x+0 and 1*f = f = f*1 and x*(y+z) = x*y + x*z.
No mention of division yet, these are valid in set ups which don't have division like rings (ie such as integers or polynomials). We define division and subtract in similar ways to one another but we need to be careful. Thus far we've said a few things about what S must contain so we're kind of building it as we go.
Given an x suppose we just say that there's an X such that x#X is the associated identity, which for # is e, so x#X=e defines X from x. Likewise y@Y=f defines Y from y. Could we just say that S must contain all such defined X and Y? Seems plausible enough but in fact it isn't because though we haven't stated it explicitly we have actually defined a system where anything multiplied by 0 gives 0, is 0*x=0 for all x in S, while we might think we can say there's an y in S such that x*y=1.
I'll do this formally, so I want to show e@x = e (ie 0*x=0) for all x in S, while we might have thought there's a y such that y@x=f (oe y*x=1). We start with the tautology x=x and go from there
x = x@f - By definition of f
x = x@(f#e) - By definition of e I've used f = f#e.
x = x@(f#f#F) - Here F is the # opposite of f (ie if f=1 and #=+ then F=-1)
x = (x@f)#(x@(f#F)) - Distributivity
x = x#(x@(f#F)) - Definition of f under @
x#X = (x#X)#(x@(f#F)) - #X on both sides and using commuting of # terms
e = e#(x@(f#F)) - Definition of X under #
e = (x@(f#F)) - Definition of e# identity
e = x@e - Definition of f#F
Similarly you can get e = e@x. Thus e@x cannot be anything other than e, regardless of which x you pick from S, so unless e=f (which it only does in the trivial ring) there is no x such that x@e=f or more commonly, there's no x such that 0*x=1. I haven't had to assume it exists and constructed a contradiction, I've shown it cannot exist within those axioms.
To allow for such an element in S you have to throw away one of the axioms I used in the step by step derivation. For example, what if there is no X such that x#X=e (ie x+X=0)?
Suppose we do throw away an axiom to let us include this magical number, let's call it Q. We don't want to break things too much so let's say Q is the only number such that 0*Q=1 (or rather Q@e = f). Well then we are led to the annoying property that Q+x=Q for any x in S. We know by the axioms Q+x is in S so which element is it? We note e@(x#Q) = (e@x)#(e@Q) using distributivity. If x is not Q then e@x=e so using Q@e = f, e@(x#Q) = (e@x)#(e@Q) = e#f = f. So now we have both x#Q and Q satisfying e@y = f but Q is unique, so Q=Q#x for any x not equal to Q. So adding anything to Q doesn't change things so now you have a system where x+1=x can be true, despite it being very akin to 1=0.
As a result of this the 'nice' mathematical number systems people work with are those which allow for x#X=e and y@Y=f to define X and Y in all cases except when y=e. Hence all numbers in the Reals have a negative such that x+X=0 and all numbers except the additive identity (ie 0) have inverses such that y*Y=1.
More simply:
Let R be some set of objects and # and @ be two binary operations
Let x, y and z be the elements (the objects) of R
R has the following axioms:
1. Given x,y are in R, x#y and x@y are also in R
2. Commutativity in @ i.e. x@y=y@x
3. Commutativity in # i.e. x#y=y#x
4. Distributivity of @ over # i.e. x@(y#z)=(x@y)#(x@z)
5. # identity: a (a#x=x#a=x)
6. @ identity: m (m@x=x@m=x)
9. Associativity in @ i.e. (x@y)@z=x@(y@z)=x@y@z
10. Associativity in # i.e. (x#y)#z=x#(y#z)=x#y#z
Using these axioms we then define:
7. Negative numbers (additive inverses)
Let X be some elements in R such that x#X=a
8. Reciprocals (multiplicative inverses)
Let Y be some elements in R such that y@Y=m
It seems plausible that all X and Y are defined in R
However based on the above axioms of R, we actually implicitly prove that
$$0 \times x=0$$ for any x
Formal prove:
x=x
x=x@m=m@x (using 6.)
x=x@(m#a)=(m#a)@x (using 5.)
x=(x@m)#(x@a)=(m@x)#(a@x)(using 4.)
x=x#(x@a)=x#(a@x)
x#X=x#(x@a)#X=x#(a@x)#X (#X on both sides)
a=a#(x@a)=a#(a@x) (using 3. and 7.)
a=x@a=a@x (using 7.)
Replacing a by 0 and @ by * you will get
$$0 \times x=x \times 0=0$$ for any x
Since from 6. m@x=x@m=x while from the above we proved that a@x=x@a=a
Therefore
$$a=/=m$$
Therefore
a@x=x@a=m
Or present in a more familiar way
$$0 \times x=x \times 0=1$$
Is impossible in R
Therefore the multiplicative inverse of 0 (i.e. $$\frac n0$$) DOES NOT EXIST in R (which happened to be the reals) due to the consequence of the axioms of R
External resources (Which also deal with other messy zeros such as 0! and $$0^{0}$$)
http://www.physicsforums.com/showthread.php?t=530207
As for what happens if the axioms were modified in another set S to allow the $$\frac n0$$ to exist is beyond the scope of this thread