Difference between revisions of "Cohn's criterion"
(Cohn) |
(Cohn's Criterion and Proof) |
||
Line 1: | Line 1: | ||
− | Let <math>p</math> be a prime number, and <math>b\geq 2</math> an integer. If <math>\overline{p_np_{n-1}\ | + | Let <math>p</math> be a prime number, and <math>b\geq 2</math> an integer. If <math>\overline{p_np_{n-1}\cdots p_1p_0}</math> is the base-<math>b</math> representation of <math>p</math>, and <math>0\leq p_i<b</math>, then |
<cmath>f(x)=p_nx^n+p_{n-1}x^{n-1}+\cdots+P_1x+p_0</cmath> | <cmath>f(x)=p_nx^n+p_{n-1}x^{n-1}+\cdots+P_1x+p_0</cmath> | ||
is irreducible. | is irreducible. | ||
Line 12: | Line 12: | ||
This means <math>g(z)>0</math> if <math>|z|\geq \frac{1+\sqrt{1+4H}}{2}</math>, so <math>|\phi|<\frac{1+\sqrt{1+4H}}{2}</math>. | This means <math>g(z)>0</math> if <math>|z|\geq \frac{1+\sqrt{1+4H}}{2}</math>, so <math>|\phi|<\frac{1+\sqrt{1+4H}}{2}</math>. | ||
− | If <math>b\geq 3</math>, this implies <math>|b-\phi|>1</math> if <math>b\geq 3</math> and <math>f(\phi)=0</math>. Let <math>f(x)=g(x)h(x)</math>. Since <math>f(b)=p</math>, one of <math>|g(b)|</math> and <math>h(b)</math> is 1. WLOG, assume <math>g(b)=1</math>. Let <math>\phi_1, phi_2,\cdots,\phi_r</math> be the roots of <math>g(x)</math>. This means that <math>|g(b)|=|b-\phi_1||b-\phi_2|\cdots|b-\phi_r|>1</math>. Therefore, <math>f(x) is irreducible. | + | If <math>b\geq 3</math>, this implies <math>|b-\phi|>1</math> if <math>b\geq 3</math> and <math>f(\phi)=0</math>. Let <math>f(x)=g(x)h(x)</math>. Since <math>f(b)=p</math>, one of <math>|g(b)|</math> and <math>h(b)</math> is 1. WLOG, assume <math>g(b)=1</math>. Let <math>\phi_1, phi_2,\cdots,\phi_r</math> be the roots of <math>g(x)</math>. This means that <math>|g(b)|=|b-\phi_1||b-\phi_2|\cdots|b-\phi_r|>1</math>. Therefore, <math>f(x)</math> is irreducible. |
− | If < | + | If <math>b=2</math>, we will need to prove another lemma: |
− | All of the zeroes of < | + | All of the zeroes of <math>f(x)</math> satisfy Re <math>z>\frac{3}{2}</math>. |
− | Proof: If < | + | Proof: If <math>n=1</math>, then the two polynomials are <math>x</math> and <math>x\pm 1</math>, both of which satisfy our constraint. For <math>n=2</math>, we get the polynomials <math>x^2</math>, <math>x^2\pm x</math>, <math>x^2\pm 1</math>, and <math>x^2\pm x\pm 1</math>, all of which satisfy the constraint. If <math>n\geq 3</math>, |
<cmath>|\frac{f(z)}{z^n}|\geq |1+\frac{a_{n-1}}{z}+\frac{a_{n-2}}{z^2}|-(\frac{1}{|z|^3}+\cdots+\frac{1}{|z|^n})>|1+\frac{a_{n-1}}{z}+\frac{a_{n-2}}{z^2}|-\frac{1}{|z|^2(|z|-1)}</cmath> | <cmath>|\frac{f(z)}{z^n}|\geq |1+\frac{a_{n-1}}{z}+\frac{a_{n-2}}{z^2}|-(\frac{1}{|z|^3}+\cdots+\frac{1}{|z|^n})>|1+\frac{a_{n-1}}{z}+\frac{a_{n-2}}{z^2}|-\frac{1}{|z|^2(|z|-1)}</cmath> | ||
− | If Re < | + | If Re <math>z\geq 0</math>, we have Re <math>\frac{1}{z^2}\geq 0</math>, and then |
<cmath>|\frac{f(z)}{z^n}|>1-\frac{1}{|z|^2(|z|-1)}</cmath> | <cmath>|\frac{f(z)}{z^n}|>1-\frac{1}{|z|^2(|z|-1)}</cmath> | ||
− | For < | + | For <math>|z|\geq \frac{3}{2}</math>, then <math>|z|^2(|z|-1)\geq(\frac{3}{2})^2(\frac{1}{2})=\frac{9}{8}>1</math>. Therefore, <math>z</math> is not a root of <math>f(x)</math>. |
However, if Re <math>z<0</math>, we have from our first lemma, that <math>|z|<\frac{1+\sqrt{5}}{2}</math>, so Re <math>z<\frac{1+\sqrt{5}}{2\sqrt{2}}<\frac{3}{2}</math>. Thus we have proved the lemma. | However, if Re <math>z<0</math>, we have from our first lemma, that <math>|z|<\frac{1+\sqrt{5}}{2}</math>, so Re <math>z<\frac{1+\sqrt{5}}{2\sqrt{2}}<\frac{3}{2}</math>. Thus we have proved the lemma. | ||
To finish the proof, let <math>f(x)=g(x)h(x)</math>. Since <math>f(2)=p</math>, one of <math>g(2)</math> and <math>h(2)</math> is 1. WLOG, assume <math>g(2)=1</math>. By our lemma, <math>|z-2|>|z-1|</math>. Thus, if <math>\phi_1, \phi_2,\cdots,\phi_r</math> are the roots of <math>g(x)</math>, then <math>|g(2)|=|2-\phi_1||2-\phi_2|\cdots|2-\phi_n|>|1-\phi_1||1-\phi_2|\cdots|1-\phi_n|=|g(1)|\leq 1</math>. This is a contradiction, so <math>f(x)</math> is irreducible. | To finish the proof, let <math>f(x)=g(x)h(x)</math>. Since <math>f(2)=p</math>, one of <math>g(2)</math> and <math>h(2)</math> is 1. WLOG, assume <math>g(2)=1</math>. By our lemma, <math>|z-2|>|z-1|</math>. Thus, if <math>\phi_1, \phi_2,\cdots,\phi_r</math> are the roots of <math>g(x)</math>, then <math>|g(2)|=|2-\phi_1||2-\phi_2|\cdots|2-\phi_n|>|1-\phi_1||1-\phi_2|\cdots|1-\phi_n|=|g(1)|\leq 1</math>. This is a contradiction, so <math>f(x)</math> is irreducible. |
Revision as of 15:19, 14 August 2018
Let be a prime number, and an integer. If is the base- representation of , and , then is irreducible.
Proof
The following proof is due to M. Ram Murty.
We start off with a lemma. Let . Suppose , , and . Then, any complex root of , , has a non positive real part or satisfies .
Proof: If and Re , note that: This means if , so .
If , this implies if and . Let . Since , one of and is 1. WLOG, assume . Let be the roots of . This means that . Therefore, is irreducible.
If , we will need to prove another lemma:
All of the zeroes of satisfy Re .
Proof: If , then the two polynomials are and , both of which satisfy our constraint. For , we get the polynomials , , , and , all of which satisfy the constraint. If ,
If Re , we have Re , and then For , then . Therefore, is not a root of .
However, if Re , we have from our first lemma, that , so Re . Thus we have proved the lemma.
To finish the proof, let . Since , one of and is 1. WLOG, assume . By our lemma, . Thus, if are the roots of , then . This is a contradiction, so is irreducible.