It is easy to see that no integer congruent to $4$ or $-4$ modulo $9$ can be written as the sum of three integer cubes. In view of this and Question 331163, I proposed the following conjecture in March 2019.

Conjecture. Every integer $n$ can be written as $x^3+2y^3+3z^3$ with $x,y,z$ integers. That is, $$\{x^3+2y^3+3z^3:\ x,y,z\in\mathbb Z\}=\mathbb Z.$$

This conjecture has an interesting application. Under the conjecture, my result on Hilbert's Tenth Problem implies that there is no effective algorithm to test for a general polynomial $P(x_1,\ldots,x_{33})$ with integer coefficients whether the diophantine equation $$P(x_1^{3},\ldots,x_{33}^3)=0$$ has integer solutions.

Quite recently, my PhD student Chen Wang checked my above conjecture seriously. He found that the set $$\{0,\ldots,5000\}\setminus\{x^3+2y^3+3z^3:\ x,y,z\in\{-30000,\ldots,30000\}\}$$ only contains four numbers: $36,\ 288,\ 2304,\ 4500.$ For example, he obtained that $$3772=(-20027)^3+2\times15936^3+3\times(-2739)^3.$$ Note that $$288=2^3\times 36,\ \ 2304=4^3\times36,\ \ 4500=5^3\times36.$$ So, to finish the verification of the conjecture for all $n=0,\ldots,5000$, it remains to find $x,y,z\in\mathbb Z$ with $x^3+2y^3+3z^3=36$.

QUESTION. Are there integers $x,y,z$ satisfying $x^3+2y^3+3z^3=36$?

  • 1
    $\begingroup$ I don't know if Andrew Booker's algorithm can be generalized from the sum of three cubes problem to this one, but surely Noam Elkies' algorithm can be adapted. $\endgroup$
    – Will Sawin
    May 11 '19 at 2:03
  • 3
    $\begingroup$ oeis.org/A014136 $\endgroup$ May 11 '19 at 3:10
  • $\begingroup$ @Bullet51 That is peculiar. The OEIS sequence lists 288 as a number that is not expressible in this form ... $\endgroup$
    – EGME
    May 11 '19 at 14:21
  • 2
    $\begingroup$ I don't think that your conjecture is any easier than the analogous conjecture on $x^3+y^3+z^3$. $\endgroup$
    – GH from MO
    May 11 '19 at 14:33
  • 3
    $\begingroup$ On my request, Andrew Booker obtained that the equation $x^3+2y^3+3z^3=36$ has no integral solutions with $\max\{|x|,|y|\}\le 10^8$. $\endgroup$ May 16 '19 at 19:12

Above equation shown below:

$x^3+2y^3+3z^3=n$ ------------$(1)$

Above equation $(1)$ can be written as:


Seiji Tomita has shown that for $(a+b=c)$ there are rational solution's for any '$n$'.

For, $n=36$, $(a,b,c)=(1,2,3)$, the solution is:


The above numerical solution is equivalent to:


And, 26244= (9)^3*(36)

His web page link is:


Click on "Computational number theory" & select article # 313.

  • 5
    $\begingroup$ But the question asks for integer solutions for every $n$, not rational solutions. $\endgroup$
    – Nik Weaver
    May 16 '19 at 16:35
  • $\begingroup$ There are "smaller" rational solutions like $(x,y,z)=(1/3, 10/3, -7/3)$ or $=(11/3, 2/3, -5/3 )$. $\endgroup$
    – Xarles
    Jun 16 '19 at 9:26

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.