Let $(A,+,\cdot)$ be a ring such that there are $a,b \in A$ which satisfy $$(a+b)^2=a^2+b^2, \quad (a+b)^3=a^3+b^3$$ Prove that $(a+b)^n=a^n+b^n,$ for all positive integers $n.$

I have found the following solution, but I am not quite satisfied with it.

From the hypothesis we get $ab+ba=0$ and $ab^2+ba^2=0.$ We will prove the identity using induction. Suppose that it is true for $1,2,...,n-1, \: n \geq 4.$ We can write $$(a+b)^n=(a+b)^{n-1}(a+b)=(a^{n-1}+b^{n-1})(a+b)=a^n+a^{n-1}b+b^{n-1}a+b^n$$ It is left to prove that $a^{n-1}b+b^{n-1}a=0.$ We can write $$a^{n-1}b+b^{n-1}a=a^{n-2}ab+b^{n-2}ba=-a^{n-2}ba-b^{n-2}ab \quad (*)$$ But $(a+b)^{n-1}=(a+b)^{n-2}(a+b)=(a^{n-2}+b^{n-2})(a+b)=a^{n-1}+b^{n-1},$ so $$a^{n-2}b+b^{n-2}a=0 \Rightarrow b^{n-2}a=-a^{n-2}b$$ Plugging this back in $(*)$ gives $$a^{n-1}b+b^{n-1}a=-a^{n-2}ba+a^{n-2}b^2=a^{n-2}(-ba+b^2)=a^{n-3}ab(-a+b)$$ But $0=ab^2+ba^2=ab^2-aba=ab(b-a),$ so $$a^{n-1}b+b^{n-1}a=a^{n-3}\cdot 0 = 0$$ and this completes the indution.

Is there any other solution, maybe quicker or more beautiful?


This is the same, but maybe rearranged to see an idea of norming (non-commutative) monomials in $a,b$. As noticed in the OP, from $a^2+b^2=(a+b)^2=(a+b)(a+b)=a^2+ab+ba+b^2$ we get the "supercommutativity relation" $$ ab= -ba\ . $$ The other relation, extended as $a^3+b^3=(a+b)^3=(a+b)^2(a+b)=(a^2+b^2)(a+b)=a^3+a^2b+b^2a+b^3$ gives as in the OP $$ a^2b=-b^2a\ . $$

We start now a "new" proof. (Essentially the same.) The strategy being to write each monomial in $a,b$ in a "normalized form". First, using supercommutativity we can write any monomial $aa\dots abb\dots baa\dots a bb\dots b\dots$ in the form $\pm aa\dots a\ aa\dots a\ \dots bb\dots b\ bb\dots b\dots$ by pushing all $a$'s in the front and changing signs. Using the second rule, we can reduce the $a$ powers in front of the $b$'s to get a "normalized" monomial of the shape $\pm b^?a$. Let us show inductively the relation $a^nb=-b^na$, starting from the given one for $n=2$. For $n\ge 2$ we have $$ \begin{aligned} a^{n+1}b &= aa^n b\\ &= -ab^n a&&\text{(by induction)}\\ &= -(-1)^n aa b^n&&\text{(by supercommutativity)}\\ &=-(-1)^n a^2 \underbrace{bbb\dots b}_{n\text{ times}}\\ &=-(-1)^n(- b^2 a)\underbrace{bb\dots b}_{n-1\text{ times}}\\ &=+(-1)^n b^2\ (a\underbrace{bb\dots b}_{n-1\text{ times}})\\ &=+(-1)^n b^2\ (-1)^{n-1}\underbrace{bb\dots b}_{n-1\text{ times}}a&&\text{(by supercommutativity)}\\ &=-b^{n+1}a\ . \end{aligned} $$ The wanted relation now follows also inductively, $$ (a+b)^{n+1}=(a^n+b^n)(a+b)=a^{n+1}+\underbrace{a^nb+b^na}_{=0\text{ shown above}}+b^{n+1} = a^{n+1}+b^{n+1} \ .$$


It is no hard to see by using induction $a^nb=-b^na$ and $a^nb^n=-b^na^n$

Assume that $n$ is even. $$(a+b)^n= \underbrace{(a+b)^2(a+b)^2(a+b)^2(a+b) ^2\cdots(a+b)^2}_{\text{$\frac{n}{2}$}\ \ \text{times}}$$
$$=(a^2+b^2)(a^2+b^2)\cdots(a^2+b^2)$$ $$=(a^4+a^2b^2+b^2a^2+b^4)\cdots (a^4+a^2b^2+b^2a^2+b^4)$$ Using $a^nb^n=-b^na^n$ $$=(a^4+b^4)(a^4+b^4)\cdots(a^4+b^4)$$ $$\vdots$$

  • Finally, we reach two components by using $a^nb^n=-b^na^n$, We get $a^nb=-b^na$ if used as much as necessary we obtain $$=a^n+b^n$$ $\textbf{Note:}$ When we reach the odd number of components, we do not handle the final component.

$\textbf{Example:}$ The logic is as follows: $$(a+b)^6=(a+b)^2(a+b)^2(a+b)^2$$ $$=(a^2+b^2)(a^2+b^2)(a^2+b^2)$$ $$=(a^4+a^2b^2+b^2a^2+b^4)(a^2+b^2)$$ $$=(a^4+b^4)(a^2+b^2)=(a^6+a^4b^2+b^4a^2+b^6)$$ $$=a^6-b^4ab-a^4ba+b^6$$ $$=a^6+b^5a+a^5b+b^6=a^6+b^6$$ Similarly, we can think when $n$ is odd as following:(Actually, we encounterd above the situation below.) $$(a+b)^n= \underbrace{(a+b)^2(a+b)^2(a+b)^2(a+b)^2\cdots (a+b)^2}_{\text{$\frac{n-1}{2}, $}\ \ \text{times}} \underbrace{(a+b)}_{\text{$1$ times}}=a^n+b^n $$

  • $\begingroup$ How do you show $(a+b)^2\dots(a+b)^2=a^n+b^n$? $\endgroup$ – Wojowu Sep 17 '18 at 14:47
  • $\begingroup$ @Wojowu, I think the answer is more understandable. $\endgroup$ – 1ENİGMA1 Sep 18 '18 at 10:23
  • $\begingroup$ Would you explain why you voted negative? $\endgroup$ – 1ENİGMA1 Sep 18 '18 at 12:35
  • $\begingroup$ In the last line, do you mean $(a+b)$ instead of $(a+b)^2$ at the end? Because $n\neq 2\cdot (\frac{n-1}{2})+2$, but $n=2\cdot (\frac{n-1}{2})+1$. $\endgroup$ – cansomeonehelpmeout Sep 20 '18 at 10:14
  • $\begingroup$ @ cansomeonehelpmeout , You are right. I typed wrong. It is not related solution logic. $\endgroup$ – 1ENİGMA1 Sep 20 '18 at 10:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.