6
$\begingroup$

Can we find a cubic polynomial (except the obvious ones, i.e. cubes of linear polynomials), say, $f(x)\in \Bbb{Q[x]}$ whose values are cubes for four consecutive integers? What about five consecutive integers, or more?

To find such polynomial for two consecutive integers is very obvious, i.e. $f(n)=n^3+7n$, for $n=0,1$ takes on the value $0^3,2^3$.
To find such polynomial for $3$ consecutive integers, $n(n+1)(n-1)$ also works but although we do get same value for i.e. $0^3$ when we plug in $n=-1,0,1$.
But I hope things to become more interesting from $4$ onwards. Right?

In general, given some random positive integer, say $n$ (no matter how big) can we always find a cubic polynomial whose values are cubes for $n$ consecutive integers?

It could be an open problem, very well, but if there are some examples for $n=4,5,6$ it will be interesting to know. Thanks!


On other note, my question is inspired by the following:

There is a polynomial, like, $1+24n$ which gives squares for three consecutive integers, namely $0,1,2$ (In this case these three squares ($1,25,49$) are in A.P. too, whereas it is impossible to find four squares in A.P.).
Anyways, then there is $60n^2-60n+1$ which has square values for five consecutive integers namely $n=-2,-1,0,1,2,3$

$\endgroup$
  • 1
    $\begingroup$ Do you want to say $f(x)\in \mathbb{Z}[x]$? or it's as you stated "over real numbers"? $\endgroup$ – Elaqqad Jul 15 '15 at 14:09
  • $\begingroup$ Well if we can find it over $\Bbb{Z}$, it will be extra cool, but I asked it over reals. $\endgroup$ – Bhaskar Vashishth Jul 15 '15 at 14:18
  • $\begingroup$ If it is over $\mathbb{R}$ can you not just interpolate through $\{(n,a^3),(n+1,b^3),(n+2,c^3),(n+3),d^3\}$ ??? And continue from $x=n$ to $x=n+i-1$ for $i$ consecutive cubes? $\endgroup$ – JP McCarthy Jul 15 '15 at 14:20
  • $\begingroup$ @JpMcCarthy what if the polynomial is supposed to be of degree of $3$ $\endgroup$ – Elaqqad Jul 15 '15 at 14:25
  • 1
    $\begingroup$ Sorry, I m nly asking for cubics throughout the question and thus I hv edited it to make it clear. Plz see edit. Thanks $\endgroup$ – Bhaskar Vashishth Jul 15 '15 at 14:34
2
$\begingroup$

For $4$ consecutive integers, consider $f(x)=103\,x^3-304\,x^2+228\,x$, where $f(0)=0^3$, $f(1)=3^3$, $f(2)=4^3$, and $f(3)=9^3$. For $5$ consecutive integers, consider $f(x)=85\,x^3-84\,x$, where $f(-2)=(-8)^3$, $f(-1)=(-1)^3$, $f(0)=0^3$, $f(+1)=(+1)^3$, and $f(+2)=(+8)^3$.

Now, we assume that $k\geq 4$. Let $s_0,s_1,s_2,\ldots,s_k\in\mathbb{Z}$. There exists a cubic polynomial $f(x)\in\mathbb{Q}[x]$ such that for which $f(i)=s_i^3$ for $i=0,1,2,\ldots,k$ if and only if $$s_j^3-4\,s_{j-1}^3+6\,s_{j-2}^3-4\,s_{j-3}^3+s_{j-4}^3=0$$ for all $j=4,5,\ldots,k$. To deal with $k>5$, we have to solve cubic equations for rational numbers such as $4p^3-5=q^3$ (I believe that the only rational points on this curve are $(p,q)=(1,-1)$ and $(p,q)=(2,3)$). Personally, I don't think there is a polynomial $f(x)\in\mathbb{Q}[x]$ of degree at most $3$ other than the trivial $f(x)=a^3(x-b)^3$ and the constant polynomials such that $f$ takes perfect-cube values at more than $5$ consecutive integer points.

$\endgroup$
  • $\begingroup$ What approach you took? It is really nice! $\endgroup$ – Bhaskar Vashishth Jul 15 '15 at 14:43
  • 1
    $\begingroup$ I used Lagrange polynomials, and tried to adjust the values of $f$ so that the coefficients are integers. In fact, for any $s_0,s_1,s_2,s_3\in\mathbb{Z}$, there exists a cubic $f(x)\in\mathbb{Q}[x]$ such that $f(i)=s_i^3$ for $i=0,1,2,3$. If you want $f$ to have integer coefficients, you need $2\mid s_0-s_2$, $2\mid s_1-s_3$, and $3\mid s_0-s_3$. $\endgroup$ – Batominovski Jul 15 '15 at 14:47
  • $\begingroup$ Yup, just recalled lagrange interpolation on wiki, this wont work for five points or more. Thanks. It works fine for four points though. :) $\endgroup$ – Bhaskar Vashishth Jul 15 '15 at 14:57
0
$\begingroup$

Engineers and numerical analysts know how to fit a polynomial curve passing through, e.g., the points $(1,10^3), (2, -7^3), (3, 11^3), (4,7^3)$ etc.

Given a million and one points there exists a polynomial of degree at most one million passing through them. If the given points all have integer co-ordinates, the polynomial will have rational coefficients. (Lagrange Interpolation Theorem). Can also be regarded a version of Chinese Remainder theorem.

$\endgroup$
  • $\begingroup$ The OP is looking for cubics only. $\endgroup$ – JP McCarthy Jul 15 '15 at 16:12

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.