2

Recently I was studying a linear algebra class which introduce some basic concepts of fields, groups

Once again I tried to divide by zero by constructing an algebraic structure from scratch using just a few axioms because I stumbled across this when browsing the site

We knew that the important theorem in fields that

$$0\cdot n=0$$

holds for all n

and one of the shortest proof of this is as follows

$$\text{Take any }a \in S$$ $$a\cdot 0=a\cdot (0+0) \text{ (Additive identity)}$$ $$a\cdot 0=a\cdot 0+a\cdot 0 \text{ (Distributivity of }\cdot n\text{ over }+)$$ $$a\cdot 0=0 \text{ (Cancellation using additive inverses of }a\cdot 0) $$ $$\square$$

In my following attempt, I have not assumed anything for this set $S$ except

$$\begin{matrix} \text{Associativity} \\ \text{Additive and multiplicative identities}\\ \text{Multiplicative inverses}\\ \text{Distributivity} \end{matrix}$$ and the following was set as an axiom inspired from the link

$$0\cdot m=n \text{ (Call it "Zero axiom") }$$

The following 13 Theorems was then obtained (in order) from $S$ after some algebra

$$\begin{matrix} \text{Theorem 0.1,0.2,0.3} \\ 0 =n\cdot m^{-1} \\ n^{-1}\cdot 0=m \\m\cdot n \cdot m^{-1}=m \cdot 0 \end{matrix}$$

$$\begin{matrix} \text{Theorem 1} \\ m+n=m \\ \text{Proof: } m=1 \cdot m=(1+0)\cdot m=m+0\cdot m=m+n \end{matrix}$$

$$\begin{matrix} \text{Theorem 2} \\ n+n=n \\ \text{Proof: } n+n=(0 \cdot m)+(0 \cdot m)=(0+0)\cdot m=0\cdot m=n \end{matrix}$$

$$\begin{matrix} \text{Theorem 3} \\ n+m=m \\ \text{Proof: } n+m=(0 \cdot m)+m=(0+1)\cdot m=1\cdot m=m \end{matrix}$$

$$\begin{matrix} \text{Theorem 4} \\ m=m+1 \\ \text{Proof: } n\cdot m=(n+0) \cdot m=n\cdot m+0\cdot m=n\cdot m+n=n\cdot (m+1)\Rightarrow m=m+1 \end{matrix}$$

$$\begin{matrix} \text{Theorem 5} \\ 1=1+m^{-1} \\ \text{Proof: } m\cdot n=m \cdot (n+0)=m\cdot n+m\cdot 0=m\cdot n+m\cdot n\cdot m^{-1}=m\cdot n \cdot (1+ m^{-1})\Rightarrow 1=1+m^{-1} \end{matrix}$$

$$\begin{matrix} \text{Theorem 6} \\ (1+n^{-1})\cdot 0=m^{-1} \\ \text{Proof: } (1+n^{-1})\cdot 0=0+ n^{-1}\cdot 0=n^{-1}\cdot 0=m^{-1} \end{matrix}$$

$$\begin{matrix} \text{Theorem 7} \\ n\cdot 0=0 \\ \text{Proof: Multiply n on the left both sides of Theorem 6 and then use Theorem 0.1} \end{matrix}$$

$$\begin{matrix} \text{Theorem 8} \\ n\cdot n=n \\ \text{Proof: Using Theorem 7, } n\cdot n=n \cdot (0 \cdot m)=(n \cdot 0)\cdot m=0 \cdot m=n \end{matrix}$$

$$\begin{matrix} \text{Theorem 9} \\ m^{-1}=0 \\ \text{Proof: Apply Theorem 7 to Theorem 0.2} \end{matrix}$$

$$\begin{matrix} \text{Theorem 10} \\ n=1 \\ \text{Proof: Multiply m on the right both sides of Theorem 9 and then use the Zero Axiom} \\ \text{and multiplicative inverse property}. m\cdot m^{-1}=1 \end{matrix}$$

$$\begin{matrix} \text{Theorem 11} \\ 1+1=1 \\ \text{Proof: Apply Theorem 10 to the Zero Axiom followed by} \\ \text{Theorem 9 to obtain } \\1=0\cdot m=(0+0) \cdot m=0\cdot m + 0\cdot m=m^{-1} \cdot m+m^{-1} \cdot m=1+1 \end{matrix}$$

$$\begin{matrix} \text{Theorem 12} \\ m+m=m \\ \text{Proof: Factor out m and then apply Theorem 11} \end{matrix}$$

$$\begin{matrix} \text{Theorem 13} \\ 1+m=m \\ \text{Proof: Apply Theorem 10 to the Zero Axiom and then factor out m to obtain}\\ 1+m=0\cdot m+m=(0+1)\cdot m=1\cdot m=m \end{matrix}$$

Sumarise all the findings in a Caley table we get the following:

enter image description here

Now since m remains unspecified by the axioms of $S$, if $m=0$ then we got a contradiction in Theorem 1 and Theorem 4

$$\begin{matrix} \color{red}{\text{Theorem 1 & 4} \\ 0+1=0\Rightarrow 1=0 \\ 0=0+1\Rightarrow 1=0 }\end{matrix}$$

However I don't understand, $\square$ was never proven in $S$ due to lack of additive inverses to apply cancellation property (and $S$ is possibly non integral domain, because $m \cdot m$ is unspecified and can be set to $0$ thus cancellation may not work in $S$).

But the contradictions of S clearly suggests that $\square$ must still be able to be proved in $S$ even without the cancellation law

Question:Why $$0\cdot 0=1$$ is not allowed, even if we lacked the cancellation law in this set to prove that $\square$ holds for all elements?

Secret
  • 2,420
  • 2
    Tried to read this (except the images) a few times, and am confused, possibly because the symbols ($n$, $n0$, $m$) are defined only implicitly or after they are used, or their meanings change from place to place. It also seems your entire thought process is shown (including dead ends). It's assumed $+$ is commutative, but its Cayley table isn't symmetric. You write "$0 = 1$ is a contradiction", which isn't true; it just means you have an algebraic system with only one element (which is what you'd expect if $0$ is invertible). Perhaps the last observation (partially) answers your question? – Andrew D. Hwang Mar 28 '15 at 13:23
  • 1
    The image is pretty illegible. – Pedro Mar 28 '15 at 13:23
  • 2
    Not clear what you are asking. Unless you are asking for a minimal set of field axioms from which it follows that $0x=0$ for all field elements $x$ (and if you meant to say that, why not just do so?), I don't think there is anything clear that you could be asking; it is just confused. Voting to close. – Marc van Leeuwen Mar 28 '15 at 13:31
  • 2
    Please shorten this down to the essential minimum. – mvw Mar 28 '15 at 13:38
  • The question was edited and the major issue highlighted now in the title, refined with goblin's answer. I hope it is legible enough – Secret Mar 28 '15 at 15:56

1 Answers1

0

I'm not sure what the question is, but the easiest way to prove $a0=0$ from the field axioms is:

$$0+a0 = a0 = a(0+0) = a0 + a0$$

$$\therefore 0 + a0 = a0 + a0$$

$$\therefore 0 = a0$$

The same argument goes through from the following axiom set:

  • take the axiom list for semirings with $0$. (Google "semiring axioms" if you don't know them).
  • delete $a0=0$ from the list to avoid triviality
  • adjoin the additive cancellation law:

$$a+x = a+y \rightarrow x=y$$

Note that, in semiring theory, we usually just take $a0=0$ as one of the axioms. This makes perfect sense, because $a0 = 0$ actually completes the distributive law. Can you see the pattern?

$$a(x+y+z) = ax+ay+az$$

$$a(x+y) = ax+ay$$

$$a(x)= ax$$

$$a0 = 0$$

Hint: The empty sum is $0$.

goblin GONE
  • 69,385
  • I am in the middle of editing the question now to highlight the problem given the comments said the image is too confusing.

    Specifically, does the additive cancellation law require additive inverses to exist?

    – Secret Mar 28 '15 at 14:08
  • @Secret, good question; no, the classic example is that $\mathbb{N}$ satisfies the additive cancellation law, but no element of $\mathbb{N}$ has an additive inverse, except $0$ of course. – goblin GONE Mar 28 '15 at 14:10