3

Jacobi and Madden found infinitely many primitive solutions to,

$$a^4+b^4+c^4+d^4 = (a+b+c+d)^4$$

using an elliptic curve. We will use a different approach that, like the method for,

$$a^4+b^4+c^4 = d^4$$

discussed in this post, also yields pairs of solutions.


I. The system

Define,

$$x^4+y^4+z^4+1 = (x+y+z+1)^4\tag1$$ $$\frac{x^2+x+1}{(x+y+1)(x+z+1)}=u\tag2$$ $$\frac{y^2+y+1}{(y+z+1)(y+x+1)}=v\tag3$$ $$\frac{z^2+z+1}{(z+x+1)(z+y+1)}=w\tag4$$

The variable $w$ is dependent via a rather complicated expression on $(u,v)$ so $(4)$ is redundant. Use the first three equations to solve for the three unknowns $(x,y,z)$. After some algebra, we find they are roots of quadratics hence yields pairs of solutions. The discriminant of the quadratic is,

$$D^2 = -3(2 - u + u^2)^2 + 6(2 + u + 3u^2 - 3u^3 + u^4)v - 3(5 - 6u + 6u^2 - 2u^3 + u^4)v^2 + 6(1 - u)(1 - 2u - u^2)v^3 - 3(1 - u)^2v^4$$


II. Table of known u

If there is rational $(u,v)$ such that $D$ is also rational, then the quartic in $v$ is birationally equivalent to an elliptic curve. Furthermore, if $(u,v)$ is a solution, then $\big(\frac{u+1}{u-1},v\big)$ is also a solution and which slightly complicates things. As of 2015, there are only eight $u$ (Update: Now twelve as of 2024) of small height (numerator and denominator < $1000$) that are known, and not counting its partner $u'=\frac{u+1}{u-1}$. Namely,

\begin{array}{|c|c|c|c|c|} \hline \text{#} & \color{red}u & u'=\frac{u+1}{u-1} & \color{red}v & \text{Discoverer}\\ \hline 1 & \dfrac{511}{450} & \dfrac{961}{61} & \dfrac{1651}{126} & \text{Brudno}\\ \hline 2 & \dfrac{193}{18} & \dfrac{211}{175} & \dfrac{619}{450} & \text{Wroblewski}\\ \hline 3 & \dfrac{31}{6} & \dfrac{37}{25} & \dfrac{6619}{5550} & \text{Rouse} \\ \hline 4 & \dfrac{211}{150} & \dfrac{361}{61} & \dfrac{2041}{150} & \text{Tomita}\\ \hline 5 & \dfrac{157}{150} & \dfrac{307}{7} & \dfrac{8467}{150} & \text{Rouse}\\ \hline 6 & \dfrac{181}{150} & \dfrac{331}{31} & \dfrac{277567}{31675} & \text{Tomita}\\ \hline 7 & \dfrac{373}{150} & \dfrac{523}{223} & \dfrac{9785779}{952879} & \text{Tomita}\\ \hline 8 & \dfrac{121}{96} & \dfrac{217}{25} & \dfrac{6250987}{506400} & \text{Tomita}\\ \hline \end{array}


III. Example

From the table, choose #4 and let $u=\dfrac{211}{150}$ and $v_1 =\dfrac{2041}{150}$ where $u$ was also employed by Tomita here using a different method. Solving the system,

$$x^4+y^4+z^4+1 = (x+y+z+1)^4$$ $$\frac{x^2+x+1}{(x+y+1)(x+z+1)}=\frac{211}{150}$$ $$\frac{y^2+y+1}{(y+z+1)(y+x+1)}=\frac{2041}{150}$$

yields the pair of solutions,

$$ \left(- \frac{1984340}{107110}\right)^4 + \left( \frac{1022230}{107110}\right)^4 + \left( - \frac{1229559}{107110}\right)^4 + 1 = (x+y+z+1)^4$$

$$ \left(-\frac{3597130}{1953890}\right)^4 + \left(- \frac{561760}{1953890}\right)^4 + \left(- \frac{1493309}{1953890}\right)^4 + 1 = (x+y+z+1)^4$$

where $(x,y,z)$ are the first three addends. These are the 3rd and 5th smallest known solutions, also found by Tomita in the link above. From the initial $v_1$, one can find infinitely many other $v_k$.


IV. Question

So given,

$$D^2 = -3(2 - u + u^2)^2 + 6(2 + u + 3u^2 - 3u^3 + u^4)v - 3(5 - 6u + 6u^2 - 2u^3 + u^4)v^2 + 6(1 - u)(1 - 2u - u^2)v^3 - 3(1 - u)^2v^4$$

can you find a $u$ of small height not in the table above?

1 Answers1

1

Thanks to a newer version of Tomita's table here, we find that Allan MacLeod found in 2017 four additional $u$ of small height, namely,

\begin{array}{|c|c|c|c|c|} \hline \text{#} & \color{red}u & u'=\frac{u+1}{u-1} & \color{red}v & \text{Discoverer}\\ \hline 9 & \dfrac{511}{150} & \dfrac{661}{361} & \dfrac{7219123}{2878098} & \text{MacLeod}\\ \hline 10 & \dfrac{499}{474} & \dfrac{973}{25} & \dfrac{80069461}{1388425} & \text{MacLeod}\\ \hline 11 & \dfrac{79}{54} & \dfrac{133}{25} & \dfrac{29549171683987}{25656103349287} & \text{MacLeod}\\ \hline 12 & \dfrac{49}{24} & \dfrac{73}{25} & \dfrac{138551171933011575944603377}{41031556739549840108788225} & \text{MacLeod}\\ \hline \end{array}

with the last two also discussed in this new MO post and in the last section of MacLeod's 2017 paper. Given the equation $a^4+b^4+c^4+d^4 = (a+b+c+d)^4 = e^4$, then there are about $36$ known primitive solutions with $e<10^{15}$. Statistics are,

\begin{array}{|c|c|c|} \hline \text{Range #} & \text{Range of e} & \text{# of sol} \\ \hline 1 & 10^3-10^9 & 12\\ \hline 2 & 10^9-10^{15} & 24\\ \hline - & \text{Total} & 36\\ \hline \end{array}

Compare to the similar table for $a^4+b^4+c^4 = d^4$ in this MSE post where Range 2 is also double that of Range 1. However, since the list below is incomplete (a brute-force search was done by Wroblewski only for $e<2\times10^5$, tables here), then only #1 and #2 are certain, with the correct #3 probably $e<2\times10^6$. The list is arranged $(e;a,b,c,d)$ starting with the smallest $e$, with most results found before 2017.

  1. 5491; 5400, -2634, 1770, 955, (Brudno, u = 961/61)
  2. 51361; 48150, -31764, 27385, 7590, (Wroblewski, u = 193/18)
  3. 2084559; 1229559, -1022230, 1984340, -107110 (Tomita, u = 211/150)
  4. 2852957; -2434795, 1945570, 1483582, 1858600 (Tomita, u = 2851/1626)
  5. 3698309; 561760, 1493309, 3597130, -1953890 (Tomita, u = 211/150)
  6. 3856263; 841263, -792940, -44410, 3852350 (Rouse, u = 157/150)
  7. 41357889; 39913670, -23859495, 15187700, 10116014 (Rouse, u = 31/6)
  8. 58169009; 53902630, 2542025, 35847220, -34122866 (Rouse, u = 31/6)
  9. 387938059; 378573600, 145514934, 65167315, -201317790 (Tomita, u = 193/18)
  10. 448785721; -336869940, 178944510, 210240721, 396470430 (Tomita)
  11. 765548027; 732896170, 303742360, 189854902, -460945405 (Tomita, u = 331/31)
  12. 904758187; 753684930, 294589950, 558360120, -701876813 (Tomita, u = 709/450)
  13. 1347505009; 1338058950, -89913570, 504106884, -404747255 (Jacobi-Madden, u = 961/61)
  14. 2056299853; 500764020, 1768211850, 1297734853, -1510410870 (Tomita, u = 709/450)
  15. 3187575799; 3095408880, 1655829870, -157072326, -1406590625 (Jacobi-Madden, u = 961/61)
  16. 4907568427; 719130355, -2889516060, 4672341330, 2405612802 (Tomita)
  17. 5243198761; 4625798910, 46140636, 3744195265, -3172936050 (Jacobi-Madden, u = 193/18)
  18. 19606701703; 16515508578, -10824551825, 15627586290, -1711841340 (MacLeod)
  19. 49363426793; 7784423350, 2943361793, -10692790190, 49328431840 (Tomita, u = 157/150)
  20. 66279566663; 50627178820, 1357751663, 55867457830, -41572821650 (Tomita, u = 373/150)
  21. 120237738541; 115711769730, -64829623500, 10424211666, 58931380645 (Jacobi-Madden, u = 1651/126)
  22. 183337258649; 106185491830, 80795489585, 146163232960, -149806955726 (MacLeod)
  23. 447496521629; 123140611690, 446604426005, -96985017746, -25263498320 (MacLeod)
  24. 646013554673; 35835675310, -168853510327, 134075405440, 644955984250 (Tomita, u = 157/150)
  25. 1154082080211; -6714317914, 994485789915, -698106854980, 864417463190 (MacLeod, u = 511/150)
  26. 1398023584459; 1058103081810, 535945811334, -1140105961325, 944080652640 (Tomita)
  27. 1583248235841; 259448373800, 1526478290216, -889698809680, 687020381505 (Tomita, u = 121/96)
  28. 1777049886537; -150723250810, 1751113229630, 802797814305, -626137906588 (Tomita, u = 331/31)
  29. 1944503291223; 1796256123098, 1372056547455, -494995788100, -728813591230 (Piezas 2024, u = 2851/1626)
  30. 21290372350701; 3868630767650, 895775733285, 21271390911326, -4745425061560 (MacLeod, u = 499/474)
  31. 32122121484371; 29175553438600, 5059968816155, -21271610809130, 19158210038746 (Tomita, u = 31/6)
  32. 49975792844931; 34998027446475, 32309023830920, -42457132181770, 25125873749306, (Tomita, u = 31/6)
  33. 214645386004377; -116500933664998, -4781626731970, 136060396818225, 199867549583120 (Piezas 2024, u = 2851/1626)
  34. 566873307258041; -237321095011880, 558974521862416, 22424373335225, 222795507072280 (Tomita)
  35. 584510799294687; 530920858665230, 377970149282480, 35966749745415, -360346958398438 (Tomita, u = 331/31)
  36. 686096520792029; 662971279500154, 309770790508565, 85290604949260, -371936154165950 (MacLeod)

Note 1: The largest in this list is $e \approx 6.86\times10^{14}$.

Note 2: Only the smallest parameter $u = 31/6$ has four known primitive solutions within this range.

  • Piezas. Wolfram Alpha, gives alternate (approximate) value for 'v'. This means that there are numerical solutions, but the exact numerical ratio needs to be worked out. For example, another value is v=9.62 (for u=49/24) and v=11.78 (for u=79/54). Also, for [u=(67/42),(97/72),(103/78)] we get, [v=(10.82),(13.34),(13.86)] respectively. – David Feb 20 '24 at 16:41
  • 1
    @David I assume you are referring to this MO post. Approximate solutions are not acceptable when dealing with elliptic curves. And Rouse has answered that those last three $u$ may not even have solutions at all. – Tito Piezas III Feb 21 '24 at 05:53