Let $i_1,i_2$ and $j_1,j_2$ be non-zero real roots of $ax^2+bx+c$ and $px^2+qx+r$ respectively, where a,p $\neq$0. If the system of equations $ i_1y+i_2z=0$ and $j_1y+j_2z=0$ has a non-trivial solution, then prove that$ b^2 pr= q^2 ac$.

I proved it as follows -

Let $i_1 = \frac{-b + \sqrt{b^2 -4ac}}{2a}$ and $i_2=\frac{-b - \sqrt{b^2 -4ac}}{2a}$

$j_1 = \frac{-q + \sqrt{q^2 -4pr}}{2p}$ and $j_2=\frac{-q - \sqrt{q^2 -4pr}}{2p}$

$ i_1y+i_2z=0$ can be re-arranged to $$\frac{b}{2a} (y+z) = \frac{\sqrt{b^2 -4ac}}{2a}(y-z)$$ $$b(y+z)=\sqrt{b^2 -4ac}(y-z) -------------1$$ Similarly, $$q(y+z)=\sqrt{q^2 -4pr}(y-z) -------------2$$ can be obtained.

I can't think of a way to say that y$\neq$ z , but, $\frac{[1]}{[2]}$ will give the required relationship.

Is my method correct and is there a way to solve this using matrices?

I know the systems of equations can be written as -

$$\begin{pmatrix} a & b \\ p & q \end{pmatrix} \begin{pmatrix} x^2\\ x \end{pmatrix} = \begin{pmatrix} -c\\ -r \end{pmatrix}$$ and $$\begin{pmatrix} i_1 & i_2 \\ j_1 & j_2 \end{pmatrix} \begin{pmatrix} y\\ z \end{pmatrix} = \begin{pmatrix} 0\\ 0 \end{pmatrix}$$ but I do not know how to continue this method. Any help is appreciated.

  • $\begingroup$ $y$ is allowed to be equal to $z$ but in that case $i_1 = -i_2$ and $j_1 = -j_2$ and so you must have that $b =0$ and $q=0$ so it is trivially true that $b^2pr = q^2ac$, since they're both equal to zero. $\endgroup$ – DavidButlerUofA Jul 9 '14 at 0:45

You can use a fact about matrices, namely that if $A$ is square and $A{\bf x}={\bf0}$ has a non-trivial solution then $\det A=0$. This gives $$i_1j_2=i_2j_1\ .$$ Now substitute in formulae for the roots and simplify: $$\bigl(-b+\sqrt{b^2-4ac}\bigr) \bigl(-q+\sqrt{\smash{q^2-4pr}\vphantom{b^2}}\bigr) =\bigl(-b-\sqrt{b^2-4ac}\bigr) \bigl(-q-\sqrt{\smash{q^2-4pr}\vphantom{b^2}}\bigr)$$ so $$-b\sqrt{\smash{q^2-4pr}\vphantom{b^2}}-q\sqrt{b^2-4ac} =b\sqrt{\smash{q^2-4pr}\vphantom{b^2}}+q\sqrt{b^2-4ac}\ .$$ Hence $$2b\sqrt{\smash{q^2-4pr}\vphantom{b^2}}=-2q\sqrt{b^2-4ac}\ ,\tag{$*$}$$ and squaring both sides completes the proof.

Note that if we arranged the roots differently in our first equation, then $(*)$ would become $$2b\sqrt{\smash{q^2-4pr}\vphantom{b^2}}=2q\sqrt{b^2-4ac}\ ,$$ but since we are going to square both sides, it doesn't change the result.

In your solution, you should consider two more cases. Note that $y,z$ form a non-trivial solution of the linear equations, so they are not both zero.

  • If $y=z$ then $y\ne-z$, so $(1)$ and $(2)$ give $b=0$ and $q=0$.
  • If $y=-z$ then $(1)$ and $(2)$ give $b^2=4ac$ and $q^2=4pr$.

In both cases the result follows easily.

| cite | improve this answer | |

I don't know about matrices, but here is an alternative proof:

Since we know the roots of the polynomials, we can factorise to get: $ax^2 + bx + c = a(x-i_1)(x-i_2)$ and $px^2 + qx + c = p(x-j_1)(x-j_2)$.

From this, we can find that $a(i_1 + i_2) = -b$, $ai_1 i_2 = c$, $p(j_1 + j_2) = -q$ and $pj_1 j_2 = r$.

Now consider the system $i_1 y + i_2 x = 0, j_1 y + j_2 z =0$, which has a nontrivial solution.

Suppose $i_1 = i_2 = 0$. Then the first equation is actually $ax^2$ so both $b$ and $c$ are zero. Hence $b^2pr = q^2ac$ since they are both zero.

Now suppose $i_1$ and $i_2$ are not both zero. In this case, a nontrivial solution only happens when $(j_1, j_2)$ is a multiple of $(i_1,i_2)$. So let $j_1 = k i_1$ and $j_2 = k i_2$ for some $k$.

Using the equations above: $$ p(j_1 j_2) = r\\ p(k i_1 k i_2) = r\\ k^2 p i_1 i_2 = r\\ k^2 p a i_1 i_2 = ar\\ k^2 p c = ar $$ and also: $$ p(j_1 + j_2 ) = -q\\ p(ki_1 + ki_2) = -q\\ kp(i_1 + i_2) = -q\\ kpa(i_1 +i_2) = -qa\\ kp(-b) = -qa\\ kpb = qa\\ k^2p^2b^2 = q^2 a^2\\ k^2p^2b^2c = q^2 a^2c\\ k^2 pc b^2p = q^2 a^2c\\ ar b^2p = q^2 a^2c\\ b^2pr = q^2 ac $$

(Note I have specifically avoided dividing by k, since I don't know if it is equal to zero.)

| cite | improve this answer | |

The relations between the roots and the coefficients are:





The condition for the system to have nontrivial solutions is that the corresponding matrix is noninvertible i.e. it's determinant is zero:


We have:




Hence, you have to show that:


under the above condition:


| cite | improve this answer | |

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.