1

How can I ind the values of $n\in \mathbb{N}$ that make the fraction $\frac{2n^{7}+1}{3n^{3}+2}$ reducible ?

I don't know any ideas or hints how I solve this question.

I think we must be writte $2n^{7}+1=k(3n^{3}+2)$ with $k≠1$

Bill Dubuque
  • 272,048
Ellen Ellen
  • 2,319
  • 3
    For the fraction to be reducible does not require it to be an integer. You need only that the GCD of the numerator and denominator be greater than 1. – SlipEternal Oct 09 '19 at 16:09
  • 1
    Hint : Divide the numerator by denominator and find out the remainder . – The Demonix _ Hermit Oct 09 '19 at 16:17
  • 1
    What is the context of this question? It can be solved using the resultant of the polynomials $2n^7+1$ and $3n^3+2$, but I don't know if that solution will be helpful to you. – Greg Martin Oct 09 '19 at 16:31
  • Experiment suggests that the solution is $n\equiv435\pmod{1163}$ – saulspatz Oct 09 '19 at 16:37
  • @saulspatz can you drop solution please! – Ellen Ellen Oct 09 '19 at 16:54
  • 2
    I don't have a solution. I just wrote a little python script to check $n$ up to $10000$ and that's what I found. It's easy to confirm that all such values work. I don't how to prove that that they are the only ones, and the only reason I have to believe they are is that I haven't found any others. – saulspatz Oct 09 '19 at 17:00
  • @saulspatz , could you please tell me the least '$n$' for which this expression is reducible? –  Oct 09 '19 at 17:34
  • 1
    What is the smallest positive $n$ congruent to $435\pmod{1163}$? – saulspatz Oct 09 '19 at 17:36
  • So you mean $435$?? –  Oct 09 '19 at 17:37
  • I greatly expanded my answer to highlight the close relationship with the Euclidean gcd algorithm . – Bill Dubuque Oct 10 '19 at 02:31

5 Answers5

4

I have a solution, but I'm sure there's a better way to do this. The greatest common divisor $g$ of of $2n^7+1$ and $3n^3+2$ must also divide $$3(2n^7+1)-2n^4(3n^3+2)=3-4n^4$$ then $g$ must also divide $$4n(3n^3+2)-3(4n^4-3)=8n+9$$ Then $g$ must divide $$ 3n^4(8n+9)-8(3n^3+2)=27n^2-16$$

Continuing in this manner, we eventually find that $g$ must divide $1163$ which is prime. So any solution satisfies $$3n^3+2\equiv0\pmod{1163}$$

The only solution to this is is $n\equiv435\pmod{1163}$ which I found with a python script, though I imagine there's a way to do it with a pencil.

It's easy to verify that also $2\cdot435^7+1\equiv0\pmod{1163}$, so the complete solution is $$n\equiv435\pmod{1163}.$$

EDIT

Daniel Wainfleet's answer shows the right way to find $435$.

saulspatz
  • 53,131
  • $435^7=3^7. 5^7.29^7$... $29^2=841\equiv -322\mod 1163.$ Compute $29^4\equiv (-322)^2\equiv X$ and $29^6\equiv -322 X\equiv Y$ and $29^7\equiv 29 Y.$ And so on. Another way might be to re-arrange & re-write the argument to have all "iff" so as to avoid major arithmetic but I haven't tried this...........+1 – DanielWainfleet Oct 09 '19 at 18:39
  • This is essentially a special case of the Euclidean algorithm - see my answer. – Bill Dubuque Oct 10 '19 at 01:42
3

Suppose $p$ is prime.

If $p$ divides $2n^7+1$ & $3n^3+2$

then $p$ divides $2(2n^7+1)-(3n^3+2)=n^3(4n^4-3)$

then $p$ divides $4n^4-3$ ( See Footnote )

then $p$ divides $2(4n^4-3)+3(3n^3+2)= n^3(8n+9)$

then $p$ divides $8n+9$ (See Footnote)

then $p$ divides $9(3n^3+2)-2(8n+9)=n(27n^2-16)$

then $p$ divides $27n^2-16$ (See Footnote)

then $p$ divides $9(27n^2-16)+16(8n+9)=n(243n+128)$

then $p$ divides $243n+128$ (See Footnote)

then $p$ divides $9(243n+128)-128(8n+9)=1163$

then $p=1163$ because $p$ and $1163$ are both prime

then $8n+9\equiv 0 \mod 1163$ so $n\equiv 435 \mod 1163$

So $\gcd (2n^7+1, 3n^3+2)>1\implies n\equiv 435 \mod 1163.$ And we may verify that $n\equiv 435 \mod 1163\implies 2n^7+1\equiv 3n^3+2\equiv 0 \mod 1163\implies \gcd (2n^7+1,3n^3+2)>1.$

Footnote. Suppose $A,B,C, D$ are integers with $A>0$ and $C>0,$ and $p$ divides $n^A(Bn^C+D).$ Since $p$ is prime, $p$ divides $n^A$ or $p$ divides $Bn^C+D.$ Now if prime $p$ divides $n^A$ then $p$ divides $n$ and hence $p$ divides $2n^7$, BUT if $p$ also divides $2n^7+1$ then the prime $p>1$ divides $(2n^7+1)-(2n^7)=1,$ which is absurd. So instead, $p$ must divide $Bn^C+D.$

  • Of course! $8n+9\equiv0\pmod{1163}$ I knew I was missing something easy. – saulspatz Oct 09 '19 at 17:57
  • Manual arithmetic error of mine. I fixed my A after seeing @saulpatz A, which appeared while I was typing & editing mine. – DanielWainfleet Oct 09 '19 at 18:19
  • This is essentially a special case of the Euclidean algorithm - see my answer. – Bill Dubuque Oct 10 '19 at 01:42
  • This is essentially a special case of the Euclidean algorithm - see my answer, where I also give a couple quick ways to compute the above-omitted congruence $, n\equiv -9/8\equiv 435\pmod{!1163}$ by trivial mental arithmetic – Bill Dubuque Oct 10 '19 at 02:38
2

Let $d = \gcd(2n^7+1, 3n^3+2)$. Then since $2n^7+1 \ | \ 2^3n^{21}+1$ and $3n^3 + 2 \ | \ 3^7n^{21}+2^7$, we must have $$d \ | \ 3^7(2^3n^{21}+1) - 2^3(3^7n^{21}+2^7) \quad\Rightarrow\quad d \ | \ 1163.$$

Since $1163$ is a prime, if the fraction is reducible, $1163 \ | \ 3n^3 + 2$. Since $1163 \equiv 2 \pmod 3$, $n^3 \equiv -2\cdot 3^{-1} \equiv 587 \pmod{1163}$ has one unique solution modulo $1163$.

Fermat's little theorem tells you that $n \equiv n^{1163} \equiv n^{2325} \equiv n^{3\times 775}\pmod{1163}$. So the answer should be $n \equiv 587^{775} \pmod{1163}$.

This is hard to solve by hand, so if you believe saulspatz's computation that $n \equiv 435 \pmod{1163}$, that is all the possible solutions.

Hw Chu
  • 5,761
1

This gcd is computable purely mechanically by a slight generalization of the Euclidean algorithm which allows us to scale by integers $\,c\,$ coprime to the gcd during the modular reduction step, i.e.

$$\bbox[8px,border:2px solid #c00]{(a,b)\, = \,(a,\,cb\bmod a)\ \ \ {\rm if}\ \ \ (a,c) = 1}\qquad\qquad $$

which is true since $\,(a,c)= 1\,\Rightarrow\, (a,\,cb\bmod a) = (a,cb) = (a,b)\ $ by Euclid. When computing the gcd of polynomials $\,f(n),g(n)$ with integer coef's, we can use such scalings to force the lead coef of the dividend to be divisible by the lead coef of the divisor, which enables the division to be performed with integer (vs. fraction) arithmetic. Let's do that in the example at hand (but you may find it helpful to first study the simpler examples here and here).

$\!\begin{align}{\rm{This\,\ yields}\!: \ \ \ }(3n^3\!+2,\,2n^7\!+1) &\,=\, (3n^3\!+2,\,\color{#0a0}{8n+9})\ \ \ {\rm by}\ [\![1]\!]\ \text{ below, $\,c = 3^2$}\\[.4em] &\,=\, (\color{#90f}{-1163},\ \ 8n+9)\ \ \ {\rm by}\, [\![2]\!]\ \text{ below, $\,c = 8^3$}\end{align}$ $\!\!\!\begin{align} &\bmod\:\! \color{#c00}{3n^3\!+2}^{\phantom{|^{|^|}}}\!\!\!\!\!:\,\ 3^2(2n^7\!+1)\equiv 2n(\color{#c00}{3n^3})^2\!+9\,\equiv\, \color{#0a0}{8n+9},\, \ {\rm by}\ \ \color{#c00}{3n^3\equiv -2}\,\qquad [\![1]\!]\\[.4em] &\bmod\:\! \color{#0a0}{8n+9}\!:\ \ 8^3(3n^3\!+2)\equiv 3(\color{#0a0}{8n})^3\!+ 2(8^3) \equiv \color{#90f}{-1163},\ \ {\rm by}\ \ \color{#0a0}{8n\,\equiv \,-9}\:\!\qquad[\![2]\!] \end{align}$

So the gcd $\!=\! (1163,8n\!+\!9)\! >\! 1\! \iff\!$ prime $p\! =\! 1163\mid 8n\!+\!9\!$ $\iff\! \bbox[5px,border:1px solid #0a0]{n\equiv \color{90f}{435}\pmod{\!p}}\,$ by

$\!\!\bmod 1163\!:\,\ n\equiv\dfrac{\!\!-9}8\!\equiv 3\left[\dfrac{-3}8\right] \equiv 3\left[\dfrac{1160}8\right]\equiv 3[145]\equiv 435.\ \ $ [See here for 5 more ways]

Remark $ $ Above $\,k \bmod a\,$ denotes some "simpler" $k'$ such that $\,k'\equiv k\pmod{\!a},\,$ not necessarily the least nonnegative such value. Here we are "simplifying" by reducing the degree in $\,n,\,$ i.e. essentially using the Euclidean algorithm for polynomials (the scaling by $\,c\,$ corresponds to a fraction-free form of the algorithm using the Nonmonic Division Algorithm).

Bill Dubuque
  • 272,048
1

The extended euclidean algorithm for gcd (of polynomials with rational coefficients) also tells us that $$ \left( 2 x^{7} + 1 \right) \left( { 1728 x^{2} - 1944 x + 2187 } \right) - \left( 3 x^{3} + 2 \right) \left( { 1152 x^{6} - 1296 x^{5} + 1458 x^{4} - 768 x^{3} + 864 x^{2} - 972 x + 512 } \right) = 1163 $$

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

$$ \left( 2 x^{7} + 1 \right) $$

$$ \left( 3 x^{3} + 2 \right) $$

$$ \left( 2 x^{7} + 1 \right) = \left( 3 x^{3} + 2 \right) \cdot \color{magenta}{ \left( \frac{ 6 x^{4} - 4 x }{ 9 } \right) } + \left( \frac{ 8 x + 9 }{ 9 } \right) $$ $$ \left( 3 x^{3} + 2 \right) = \left( \frac{ 8 x + 9 }{ 9 } \right) \cdot \color{magenta}{ \left( \frac{ 1728 x^{2} - 1944 x + 2187 }{ 512 } \right) } + \left( \frac{ -1163}{512 } \right) $$ $$ \left( \frac{ 8 x + 9 }{ 9 } \right) = \left( \frac{ -1163}{512 } \right) \cdot \color{magenta}{ \left( \frac{ - 4096 x - 4608 }{ 10467 } \right) } + \left( 0 \right) $$ $$ \frac{ 0}{1} $$ $$ \frac{ 1}{0} $$ $$ \color{magenta}{ \left( \frac{ 6 x^{4} - 4 x }{ 9 } \right) } \Longrightarrow \Longrightarrow \frac{ \left( \frac{ 6 x^{4} - 4 x }{ 9 } \right) }{ \left( 1 \right) } $$ $$ \color{magenta}{ \left( \frac{ 1728 x^{2} - 1944 x + 2187 }{ 512 } \right) } \Longrightarrow \Longrightarrow \frac{ \left( \frac{ 576 x^{6} - 648 x^{5} + 729 x^{4} - 384 x^{3} + 432 x^{2} - 486 x + 256 }{ 256 } \right) }{ \left( \frac{ 1728 x^{2} - 1944 x + 2187 }{ 512 } \right) } $$ $$ \color{magenta}{ \left( \frac{ - 4096 x - 4608 }{ 10467 } \right) } \Longrightarrow \Longrightarrow \frac{ \left( \frac{ - 1024 x^{7} - 512 }{ 1163 } \right) }{ \left( \frac{ - 1536 x^{3} - 1024 }{ 1163 } \right) } $$ $$ \left( 2 x^{7} + 1 \right) \left( \frac{ 1728 x^{2} - 1944 x + 2187 }{ 1163 } \right) - \left( 3 x^{3} + 2 \right) \left( \frac{ 1152 x^{6} - 1296 x^{5} + 1458 x^{4} - 768 x^{3} + 864 x^{2} - 972 x + 512 }{ 1163 } \right) = \left( 1 \right) $$

Will Jagy
  • 139,541
  • 1
    It's much easier to use modular arithmetic as I explain in my answer. None of the Bezout coef's are needed. As usual - they only serve to obfuscate the essence of the matter and make the computation much more arduous. – Bill Dubuque Oct 09 '19 at 18:52