Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Does $a^3 + 2b^3 + 4c^3 = 6abc$ have solutions in $\mathbb{Q}$?

This is not a homework problem. Indeed, I have no prior experience in number theory and would like to see a showcase of common techniques used to solve problems such as this. Thanks

Edit Apart from $a=b=c=0$.

share|improve this question
1  
$a=b=c=0$ works. –  vadim123 Jun 17 '13 at 3:13
    
What makes you think this is solvable? Is it from a contest? Finding the rational solutions of a general diophantine equation is a very hard problem (cf. Fermat's lat theorem). –  Potato Jun 17 '13 at 3:13
    
@vadim123 See edit –  user71815 Jun 17 '13 at 3:21
    
@Potato (i) I have no idea whether or not it is solvable, which is why I asked; (ii) This is not from a contest, but from a textbook; (iii) Of course, but this is not a general diophantine equation –  user71815 Jun 17 '13 at 3:23
1  
@user71815 Generally textbook problems are solvable. :). –  Potato Jun 17 '13 at 3:27

4 Answers 4

up vote 16 down vote accepted

First, note that $(a,b,c)$ is a solution if and only if $(ka,kb,kc)$ is; hence we may assume $a,b,c$ are integers, with no common factor (divide by that common factor if necessary).

Because $6abc, 2b^3+4c^3$ are even, so is $a^3$ and hence $a$. Write $a=2a'$ and we have $$8(a')^3+2b^3+4c^3=12a'bc$$ and hence $$4(a')^3+b^3+2c^3=6a'bc$$

By similar logic, $b$ is even, so write $b=2b'$ and we hve $$4(a')^3+8(b')^3+2c^3=12a'b'c$$ But now $$2(a')^2+4(b')^3+c^3=6a'b'c$$ and hence $c$ is even. Hence $a,b,c$ are all even; this contradicts $a,b,c$ having no common factor.

Followup: The same proof works if the coefficients $\{1,2,4,6\}$ are replaced by $\{\alpha_1, \alpha_2, \alpha_3,\alpha_4\}$ so long as there is some prime $p$ with $\nu_p(\alpha_1)=0, \nu_p(\alpha_2)=1, \nu_p(\alpha_3)=2, \nu_p(\alpha_4)\ge 1$. (Here $\nu_p(\cdot)$ denotes the p-adic valuation). For example, apart from $(0,0,0)$, there are no rational solutions to $$7a^3+15b^3+18c^3=45abc$$ where here $p=3$.

$~$

Double followup: The same proof works with $n$ variables $$\alpha_0a_0^n+\alpha_1a_1^n+\cdots+\alpha_{n-1}a_{n-1}^n=\alpha_n(a_0a_1\cdots a_{n-1})$$ provided that $\nu_p(\alpha_i)=i$ (for $0\le i\le n-1$) and $\nu_p(\alpha_n)\ge 1$.

share|improve this answer
1  
Nice. Reminds one of a traditional proof about $\sqrt{2}$ :) +1 –  coffeemath Jun 17 '13 at 3:25
    
@user71815, alas Diophantine equations have many many techniques to solve them. Finding a divisor and making a substitution (as I did 3 times) is a common one. –  vadim123 Jun 17 '13 at 3:25
    
Yes indeed! Thanks –  user71815 Jun 17 '13 at 3:26

As it happens, the cubic form $C(a,b,c)=a^3+2b^3+4c^3-6abc$ is the norm form for the extension $K=\mathbb Q(\root 3 \of 2)$ over $\mathbb Q$. That is, if you look at a general element of $K$, say $a + b\root3\of2+c\root3\of2^2$, and take its field-theoretic norm, what you get is exactly $C(a,b,c)$. Now, the norm doesn’t vanish on an algebraic extension, except at zero. So, just because $C$ happens to be a norm form, you can say immediately that the trivial zero is the only one.

How did I spot this? By having done lots of examples, first by hand over many years, then, more recently, with symbolic algebra programs.

share|improve this answer
    
Is it true that, for prime $q = 4 u^2 + 2 u v + 7 v^2,$ so that 2 is not a cubic residue, and $C(a,b,c) \equiv 0 \pmod q,$ then each $a,b,c \equiv 0 \pmod q?$ See math.stackexchange.com/questions/329936/… Note that Noam Elkies gave me a proof for a very similar problem at mathoverflow.net/questions/127160/… –  Will Jagy Jun 17 '13 at 4:33
    
Oh man, @Will Jagy, I don’t know nothin’ about things like that. Sorry. –  Lubin Jun 17 '13 at 15:56

HINT Make use of $$x^3 + y^3 + z^3 -3xyz = \left(x+y+z \right) \left(x^2+y^2+z^2-xy-yz-zx \right)$$where $x=a$, $y=b\sqrt[3]{2}$ and $z=c \sqrt[3]{4}$.

share|improve this answer
1  
Does this help if one seeks rational $a,b,c$ as in the posted question? –  coffeemath Jun 17 '13 at 3:19
1  
Oh, I think I see where this was going. The left side is zero iff the equation of the OP holds, for the integers $a,b,c$. So a solution would imply one of the two factors on the right is zero. The first one clearly isn't, if we assume $1,\sqrt[3]{2},\sqrt[3]{4}$ are rationally independent, and the second more complicated factor might be clearly nonzero for similar reasons, though I didn't check. So I'm giving this an upvote, but would advise inclusion of some of the explanation just indicated. –  coffeemath Jun 17 '13 at 4:13
    
I downvoted this answer, and would happily remove it after seeing coffeemath's explanation above. The second term can't vanish unless $x=y=z$ (by writing it as sum of squares), and it's not hard to check that there's no solution in that case. –  user27126 Jun 17 '13 at 6:18
1  
You can show rational independence directly (no Galois) by assuming $a+b\sqrt[3]{2}+c\sqrt[3]{4}=0$ for rational $a,b,c$, and then start by isolating $\sqrt[3]{4}$ and cube both sides, then plug in to eliminate $\sqrt[3]{4}$ from the first formula into the cubed out formula. The result if zero for nonzero $c$ would give that $\sqrt[3]{2}$ is rational. –  coffeemath Jun 18 '13 at 8:06

if negative numbers are accepted, (57,63,-156) is OK

share|improve this answer
2  
Those numbers satisfy the equation $a^3+b^3+c^3=6abc$, not the OP's equation. –  Barry Cipra Apr 9 at 13:31

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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