Re: expressing a number as the sum of 3 triangular numbers
[email protected] Sun, 16 Nov 2025 18:36:59 +0100
| Newsgroups | gmane.comp.mathematics.pari.user |
|---|---|
| Message-ID | <[email protected]> |
On 2025-11-16 03:28, Max Alekseyev wrote:
> You can replace sqs3 with
>
> { sqs3_(N) = my(s = qfsolve(matdiagonal([1,1,1,-N])));
> vector(3,i,abs(s[i]/s[4])); }
>
> Regards,
> Max
>
Wow, your solution is a oneliner for "sum of three squares":
? N=254;my(s = qfsolve(matdiagonal([1,1,1,-N])));
vector(3,i,abs(s[i]/s[4]))
%12 = [2, 5, 15]
?
(A) Are the entries s[i]/s[4] guaranteed to be integers?
[for n \neq 4^a (8b+7) with a,b positive integers (*)]
? n=254;qfsolve(matdiagonal([1,1,1,-n]))
%14 = [-2, 5, 15, 1]~
?
(B) Can it ever happen that s[4] is not 1 for n as in (*)?
? ?qfsolve
qfsolve(G): solve over Q the quadratic equation X^t G X = 0, where G is
a
symmetric matrix.
?
(C) Is it guaranteed that the first three entries of
qfsolve(matdiagonal([1,1,1,-n]))
with n as in (*) are always integers?
Regards,
Hermann.