8

Question:If $α$ and $β$ be the roots of $ax^2+2bx+c=0$ and $α+δ$, $β+δ$ be those of $Ax^2+2Bx+C=0$, prove that, $\frac{b^2-ac}{a^2}=\frac{B^2-AC}{A^2}$.

My Attempt: Finding the sum of roots and product of roots for both the equations we get,

$α+β=\frac{-2b}{a}$


$αβ=\frac{c}{a}$


$α+δ+β+δ=\frac{-2B}{A}$

⇒ $α+β+2δ =\frac{-2B}{A}$


$(α+δ)(β+δ)=\frac{C}{A}$

⇒ $αβ+αδ+βδ+δ^2=\frac{C}{A}$

⇒$\frac{c}{a}+αδ+βδ+δ^2=\frac{C}{A}$ ⇒ $αδ+βδ+δ^2=\frac{Ca-cA}{Aa}$


$(α+β)^2=\frac{4b^2}{a^2}$

⇒ $α^2+β^2+2αβ=\frac{4b^2}{a^2}$

$α^2+β^2+\frac{2c}{a}=\frac{4b^2}{a^2}$

⇒ $α^2+β^2=\frac{4b^2-2ac}{a^2}$ -(1)


$(α+β+2δ)^2 =\frac{4B^2}{A^2}$

⇒ $α^2+β^2+(2δ)^2+2(αβ+2βδ+2αδ)=\frac{4B^2}{A^2}$

⇒$α^2+β^2+4δ^2+2αβ+4βδ+4αδ=\frac{4B^2}{A^2}$

⇒$α^2+β^2+2αβ+4(δ^2+βδ+αδ)=\frac{4B^2}{A^2}$

⇒$α^2+β^2+\frac{2c}{a}+4(\frac{Ca-cA}{Aa})=\frac{4B^2}{A^2}$

⇒$α^2+β^2=\frac{4aB^2-2A^2 c-4Aac+4cA^2}{A^2a}$ -(2)


From (1) and (2) we get,

$\frac{4b^2-2ac}{a^2}=\frac{4aB^2-2A^2 c-4Aac+4cA^2}{A^2a}$

My problem: I tried simplifying it further but could not reach the required result. A continuation of my method would be more appreciated compared to other methods.

MrAP
  • 3,003

5 Answers5

20

Instead doing all the hard work you did, you can notice that the difference of roots $(\vert x_1-x_2\vert )$ is same for both the equations. Hence :

$$|\alpha-\beta|=|(\alpha+\delta)-(\beta+\delta)|=\sqrt{(\alpha+\beta)^2-4\alpha \beta}=\sqrt{(\alpha+\delta +\beta+ \delta)^2-4(\alpha+\delta)( \beta+\delta)}$$

$$\implies \sqrt{\left(\frac {-2b}{a} \right)^2 -4\left(\frac ca \right)}= \sqrt{\left(\frac {-2B}{A} \right)^2 -4\left(\frac CA\right)}$$

$$\implies \frac{b^2-ac}{a^2}=\frac{B^2-AC}{A^2}$$

Jaideep Khare
  • 19,293
8

Alternative hint: we can assume WLOG that $\,a=A=1\,$, since both the roots and the equality to be proved are homogeneous in the respective coefficients.

Then, if $\alpha, \beta$ are the roots of $x^2+2bx+c=0$, the polynomial with roots $\alpha+\delta, \beta+\delta$ is:

$$(x-\delta)^2+2b (x-\delta)+c=0 \;\;\iff\;\; x^2 + 2(b-\delta)x+\delta^2-2b\delta+c=0\,$$

Identifying coefficients gives $B=b-\delta$ and $C=\delta^2-2b\delta+c\,$, then:

$$\require{cancel} B^2 - C=(b^2-\bcancel{2 b\delta}+\cancel{\delta^2})-(\cancel{\delta^2}-\bcancel{2b\delta}+c) = b^2 - c $$


[ EDIT ]  To answer OP's edit:

A continuation of my method would be more appreciated compared to other methods.

There is an error/typo in formula (2). Once corrected (in red below): $$\require{cancel} \frac{4b^2-2ac}{a^\bcancel{2}}=\frac{4aB^2-2A^2 c-4Aa\color{red}{C}+4cA^2}{A^2 \bcancel{a}} $$

$$ 4b^2A^2-\bcancel{2acA^2}=4a^2B^2-\bcancel{2acA^2}-4a^2AC+4acA^2 $$

$$ \bcancel{4}A^2(b^2-ac) = \bcancel{4}a^2(B^2-AC) $$

$$\frac{b^2-ac}{a^2} =\frac{B^2-AC}{A^2} $$

MrAP
  • 3,003
dxiv
  • 76,497
4

WLOG, $A=a=1$ (otherwise you can divide the trinomials by their leading coefficient).

Then

$$(x+\delta)^2+2B(x+\delta)+C=x^2+2(\delta+B)x+\delta^2+2B\delta+C=x^2+2bx+c,$$ and

$$b^2-c=(\delta+B)^2-(\delta^2+2B\delta+C)=B^2-C.$$

  • Why is $Ax^2+2(Aδ+B)x+Aδ^2+2Bδ+C=ax^2+2bx+c$? – MrAP Jun 18 '17 at 23:14
  • @MrAP: because two quadratic polynomials that have the same roots have proportional coefficients. And when $a=A$, the coefficients are equal. –  Jun 18 '17 at 23:17
2

Here is the LHS: $$\frac{b^2-ac}{a^2} = \left(\frac{b}{a}\right)^2-\frac {c}{a}$$ $$=\left(\frac {-(\alpha+\beta)} {2}\right)^2-\alpha\beta$$ $$=\frac {{\alpha}^2+{\beta}^2+2\alpha\beta}{4}-\frac {4\alpha\beta}{4}$$ $$=\frac{{\alpha}^2+{\beta}^2-2\alpha\beta} {4}$$ $$=\left(\frac{\alpha-\beta} {2}\right)^2$$ And here is the RHS: $$\frac{B^2-AC}{A^2}=\left(\frac {B}{A}\right)^2-\frac{C}{A}$$ $$=\left(\frac{-(\alpha+\beta+2\delta)}{2}\right)^2-(\alpha+\delta) (\beta+\delta)$$ $$=\frac {{\alpha}^2+{\beta}^2+4{\delta}^2+2\alpha\beta+4\alpha\delta+4\beta\delta} {4}-\frac {4\alpha\beta+4\alpha\delta+4\beta\delta} {4}$$ $$=\frac {{\alpha}^2+{\beta}^2-2\alpha\beta} {4}$$ $$=\left(\frac {\alpha-\beta} {2}\right)^2$$ Clearly, LHS = RHS. QED.

  • Nice (+1) but you still need to conclude the proof. Something like: by an entirely similar calculation, the RHS is $\frac{B^2-AC}{A^2}=\cdots=\left(\frac{(\alpha+\delta)-(\beta+\delta)} {2}\right)^2=\left(\frac{\alpha-\beta} {2}\right)^2,$, which proves that the RHS=LHS. P.S. And now you can delete the first posted draft. – dxiv Jun 19 '17 at 05:36
  • Thanks, I'm a bit slow in typing MathJax. – Myungjin Hyun Jun 19 '17 at 05:43
  • Ignore other proofs on this page by me, they are incomplete. – Myungjin Hyun Jun 19 '17 at 05:54
  • By the way, MathJax seems to be rather like a programming language. Does anyone know it's history? I'm curious! – Myungjin Hyun Jun 19 '17 at 05:58
  • It would be more apt to compare it to MathML. – Myungjin Hyun Jun 19 '17 at 05:59
  • If you want to say something to someone, then you can notify by using @username. If you don't tag someone, they will never know that you wanted to say something to them. For example @JaideepKhare will notify me. – Jaideep Khare Jun 19 '17 at 07:25
  • Thanks, @JaideepKhare, I'll bear that in mind. – Myungjin Hyun Jun 19 '17 at 16:56
0

By the formula for the roots of a quadratic equation, the squared difference between them is $$\left(\frac{\pm\sqrt{b^2-ac}}{a}\right)^2=\frac{b^2-ac}{a^2}$$ (factor $4$ omitted) and is invariant by translation.

  • Very similar argument than JaideepKhare, but requires no development. –  Jun 19 '17 at 08:41