Solutions to $a^4+b^4+c^4 = d^4+e^4$ with $d\neq e$?
问题内容
(Updated with a computer search.)
I. Question
We seek to find infinitely many primitive solutions to,
$$a^4+b^4+c^4 = d^4+e^4$$
where $d \color{red}{\ne} e$. The most well-known case when $d = e$ is,
$$a^4+b^4+(a+b)^4 = 2(a^2+ab+b^2)^2$$
where one then solves $a^2+ab+b^2 = z^k$ for $k=2$. (In fact, it can be solved for any positive integer $k$.) But there are infinitely many solutions where $d\pm e \neq 0$,
$$(6n)^4 + (4 - 3 n^4 - n^8)^4 + (4 n - n^9)^4 =\\ (4 + 3 n^4 - n^8)^4 + (2 n + n^9)^4$$
$$(1 - 2 n^4 - n^8)^4 + (2 n - 2 n^9)^4 + (3 n^3 - n^{11})^4 =\\ (1 + 2 n^4 - n^8)^4 + (n^3 + n^{11})^4$$
by Gerardin and Norrie, of deg-$9$ and deg-$11$, respectively. (Are there any of smaller degree?)
II. Numerical solutions
They seem quite rare compared to the case $d=e$. A quick Mathematica search yields only $14$ primitive solutions with all terms $x_i<100$, the smallest $10$ being,
3^4 + 20^4 + 26^4 = 7^4 + 28^4
3^4 + 26^4 + 35^4 = 17^4 + 37^4
1^4 + 25^4 + 42^4 = 17^4 + 43^4
25^4 + 26^4 + 42^4 = 37^4 + 38^4
7^4 + 29^4 + 50^4 = 21^4 + 51^4
5^4 + 42^4 + 78^4 = 51^4 + 76^4
25^4 + 76^4 + 60^4 = 35^4 + 82^4
17^4 + 52^4 + 80^4 = 48^4 + 81^4
39^4 + 79^4 + 60^4 = 51^4 + 83^4
4^4 + 57^4 + 85^4 = 11^4 + 89^4
III. Elliptic curve 1
The $3$rd has symmetries we can exploit, namely,
$$1^4 + (a + b)^4 + (2 b)^4 = (-a + b)^4 + (2 b + 1)^4$$
Expanding, this reduces to just a quadratic in $b$,
$$a^3 - 3 b + (a - 4) b^2 = 1$$
where its discriminant $D_1$ must be made a square $D_1 = 4 (a^3 - 1)(4 - a) + 9 = y^2$. From initial rational point $a=4$, we can find an infinite more rational $a$.
IV. Elliptic curve 2
The $4$th also has symmetries we can exploit,
$$(a - 2d)^4 + (a + 2d)^4 + (4c)^4 = (b - 2d)^4 + (b + 2d)^4$$
Expanding, this reduces to a quadratic in $d$,
$$a^4 - b^4 + 128 c^4 - 24 (-a^2 + b^2) d^2=0$$
Its discriminant $D_2$ must be made a square,
$$D_2 = 24 (-a^2 + b^2)(a^4 - b^4 + 128 c^4) = y^2$$
which is only a quartic in $c$. From the rational point $(a,b)=(17,\,25)$, we find initial rational point $c = 7$ then an infinite more rational (but not integral) $c$.
Note 1: A computer search for small integer $(a,b,c)$ with $\text{GCD}(a,b,c)=1$ yields,
(a,b,c) = (7, 31, 19)
(a,b,c) = (8, 16, 11)
(a,b,c) = (8, 20, 7)
(a,b,c) = (17, 25, 7)
though each of these integer $a<b<200$ will yield infinitely many rational $c$.
Note 2: To find useful symmetries on the other solutions was a bit more elusive.
V. Summary
So can we find more polynomial parameterizations or elliptic curves for $a^4+b^4+c^4 = d^4+e^4$ where $d\neq e$?
P.S. The third solution appears in Guruprasad's 2025 MO post. There are only three known with a unit term on the LHS,
$$1+25^4+42^4=17^4+43^4\\ 1+405^4+1786^4=253^4+1787^4\\ 1+3870^4+11954^4=6953^4+11632^4$$
the first two by Guruprasad, and the third by Peter Mueller. Mueller points out in his answer that if there is a fourth, then it must be enormous. There doesn't seem to be any known solution if the unit term is on the RHS,
$$a^4+b^4+c^4 = d^4+1$$
回答 (3)
$4641^4+4935^4+6722^4=7461^4+1$
$4481^4+5228^4+8175^4=8657^4+1$
$19074^4+75736^4+92865^4=101802^4+1$
I've used an algorithm that sorts on positive sums $S = x^4 \pm y^4$, and then tests for adjacent sums differing by $1$. After rearranging, that gives a solution to one of the 3 forms:
$$a^4 + b^4 + c^4 = d^4 + 1$$ $$a^4 + b^4 = c^4 + d^4 + 1$$ $$a^4 = b^4 + c^4 + d^4 + 1$$
The search is exhaustive, and currently covers a radius $r<320669$ where $r^4=S$. I've found: $$143884^4 + 166447^4 + 1014825^4 = 1015111^4 + 1$$ in addition to the smaller solutions already discovered. That gives totals of 4+3+0 solutions to each of the 3 forms above.
Equation, $(a^4+b^4+c^4)=(d^4+e^4)$------- $(1)$
Above equation has been parametrized by Seiji Tomita on his website. The solution is shown below.
$a=6p^2$
$b=4(p+8)(p-8)$
$c=2x(p+4)(p-16)$
$d=2x(p-4)(p+16)$
$e=2(p^2+128)$
There is symmetry & beauty in the above solution. In his solution we put,$(a,b)=(2,x)$ & then substitute (p=x^4).
For (x=1) we get: $(1,42,25)^4=(17,43)^4$. This numerical solution has the unit one on the (LHS), which is a near miss for the (4-2-2) quartic equation, as pointed out by "OP".
The link to Tomita's web site is given below:
http://www.maroon.dti.ne.jp/fermat
Select fourth powers & then click on Item #24.