I came across this problem which asks to transform $x^2-x-2$ to $x^2-x-1$ ,if possible, using the following rules:

Given a quadratic equation $ax^2+bx+c$ you can :

1)Interchange $a$ and $c$

2)Replace $x$ by $x+t$ where $t$ is a real number.

My approach:

I wrote the equations as

${(x-\frac12)}^2-\frac94$ and ${(x-\frac12)}^2-\frac54$

Replacing $x$ by $x+t$ in first equation ${(x-\frac12+t)}^2-\frac94$ And setting this equal to ${(x-\frac12)}^2-\frac54$

Simplifying and solving gives a quadratic with variable $t$ Which on applying quadratic formula gives $$t=\frac{1-2x\pm \sqrt{4x^2-4x+5}}{2}$$

So, is this correct? If wrong, could somebody help me in the right direction?

  • $\begingroup$ Seems to me that according to the rules, $t$ must have a specific, fixed value. $\endgroup$
    – amd
    May 30 '17 at 6:27
  • 1
    $\begingroup$ Are you sure that it’s possible? What effect do the two allowable transformations have on a parabola? $\endgroup$
    – amd
    May 30 '17 at 6:30
  • $\begingroup$ I am not completely sure ...but if $t$ has a fixed value ...then I believe it maybe impossible.. $\endgroup$
    – user35508
    May 30 '17 at 6:32

We start with a quadratic $ax^2+bx+c$ and use one of the steps to transform it to $a'x^2+b'x+c'$. In the first case, we have $a'=c,b'=b,c'=a$, in the second case, we have $a'=a, b'=b+2at, c'=c+bt+t^2$. Note that in the first case, we have $b'^2-4a'c'=b^2-4ac$ and in the second case $b'^2-4a'c'=b^2+4abt+4a^2t^2-4a(c+bt+t^2)=b^2-4ac$. In other words, neithre of the allowed steps changes the discriminat of our quadratic. Hence by starting from $x^2-x-2$ with discriminant $9$, we cannot reach $x^2-x-1$ with discriminant $5$.


Let $f(x)=ax^2+bx+c$. The 2 operations than can be done is equivalent to converting $f(x)$ to the following respectively:

  1. $x^2f(\frac{1}{x})$
  2. $f(x+t)$

Thus, if $\alpha$ and $\beta$ are the roots of $f(x)$, the $1^{st}$ transformation gives a polynomial with roots $\frac{1}{\alpha}$ and $\frac{1}{\beta}$ and the $2^{nd}$ transformation gives a polynomial with roots $\alpha - t$ and $\beta -t$.

Now, the general method to solve such problems is to find something which remains invariant. Now, if we look at the roots of $x^2-x-2$ ($2$ and $-1$) and the roots of $x^2-x-1$
($\frac{1+\sqrt{5}}{2}$ and $\frac{1-\sqrt{5}}{2}$), we can see that we start with a polynomial with rational roots and end up with a polynomial with irrational roots which is not possible as long as $t$ is rational.

But this is not possible if $t$ is irrational. If $t$ is irrational, then we can instead show that the discriminant remains invariant in each transformation, which is what Hagen does.

  • 2
    $\begingroup$ "(...) each transformation preserve the rationality of the roots (...)" What about the second transformation for say $t=\pi$? $\endgroup$
    – Marc Paul
    May 30 '17 at 11:17
  • 1
    $\begingroup$ @Servaes But that creates a new problem, because the property of rational difference between the roots is not preserved under the first transformation (e.g. if $\alpha = \pi$, $\beta = \pi +1$, then $\alpha - \beta$ is rational and $1/\alpha - 1/\beta$ is not). $\endgroup$
    – Marc Paul
    May 30 '17 at 12:49

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.