Problem Statement:-
Find all the values of $a$ for which the equation $$x^4+(a-1)x^3+x^2+(a-1)x+1=0$$ possess at least two negative roots.
I know that there has been a post regarding this same problem here, but I have a different issue with the problem than that posted in the OP in the above post.
So, I don't have an issue of finding a solution and coincidentally I had the same approach as the one posted by RicardoCruz.
But still if you have a solution than you are more than welcome to post one, well come on who doesn't want some good reputation change ;P.
My Attempt at a solution:-
For those who wanna skip to my main issue may start to read from the next quote.
We can see very clearly that $x=0$ isn't a root of the equation $$x^4 +(a-1)x^3 +x^2 +(a-1)x+1=0$$ hence it is safe to divide the equation throughout by $x^2$ for arriving at the condition which the problem wants.
So, we get $$\left(x^2+\dfrac{1}{x^2}+2\right)+(a-1)(x+\dfrac{1}{x})-1=0\\ \implies \left(x+\dfrac{1}{x}\right)^2+(a-1)(x+\dfrac{1}{x})-1=0$$.
Now, let $z=x+\dfrac{1}{x}$ then the equation becomes $$z^2+(a-1)z-1=0\\ \implies z=\dfrac{-(a-1)\pm\sqrt{\left(a-1\right)^2+4}}{2}$$
Now, its necessary to note that $\sqrt{\left(a-1\right)^2+4}\gt(a-1)$, hence for the roots to be negative $$\left(x+\dfrac{1}{x}\right)=\dfrac{-(a-1)-\sqrt{\left(a-1\right)^2+4}}{2}$$
As, we know that $\text{A.M. $\ge$ G.M.}$ for non negative real numbers, and as we are dealing with negative roots of the equation, so we will be applying $\text{A.M. $\ge$ G.M.}$ to $-x,-\dfrac{1}{x}$, we get $$\dfrac{-\left(x+\dfrac{1}{x}\right)}{2}\ge1\implies x+\dfrac{1}{x}\le-2\\ \implies \left(x+\dfrac{1}{x}\right)=\dfrac{-(a-1)-\sqrt{\left(a-1\right)^2+4}}{2}\le -2$$
But as the roots are supposed to be distinct hence, we have to exclude the case of equality, because it occurs when $x=\dfrac{1}{x}=-1$, which results in $x+\dfrac{1}{x}=-2\implies (x-1)^2=0$, hence $x$ has a repeated root. So, we get
$$\dfrac{-(a-1)-\sqrt{\left(a-1\right)^2+4}}{2}\lt -2\implies \sqrt{\left(a-1\right)^2+4}\gt 5-a$$
Now, this is where I am having problem, that's right solving the "inequality". I always had thought of this before but never tried to test it before, so bear with it and follow me on to a trip to solve inequalities in a weird way.
I did a case study for the inequality, which is as follows:-
Case-1:-
If $(5-a)\gt 0\implies a\lt 5$, then there is no problem in squaring the inequality as both sides are positive. $$\left(a-1\right)^2+4\gt a^2+25-10a\implies a\gt \dfrac{5}{2}$$ $$\therefore a\in\left(\dfrac{5}{2},5\right)$$
Case-2:-
If $5-a\lt0$, then we might be faced with a problem cause, there might be a case where $|5-a|\gt\sqrt{\left(a-1\right)^2+4}$, which would make the inequality $\sqrt{\left(a-1\right)^2+4}\gt 5-a$ on squaring change the sign of inequality to be in accordance with the condition $|5-a|\gt\sqrt{\left(a-1\right)^2+4}$.
So, on solving the inequality we get,
$$\sqrt{\left(a-1\right)^2+4}\gt 5-a\implies (a-1)^2+4\lt(5-a)^2$$ Now, why did I change the sign of inequality was due to the fact that I was finding the interval where(as per the condition I stated before) $|5-a|\gt\sqrt{\left(a-1\right)^2+4}$. So, just like $5\gt-6$ but on squaring the inequality we get $25\lt36$. So that's what I did.
This results in $a\lt\dfrac{5}{2}\cup a\gt5\implies a\in\emptyset$
So, on combining both the cases we get $a\in\left(\dfrac{5}{2},5\right)$
But from RobertCruz's solution and the books answer it seems that $ a\in\left(\dfrac{5}{2},\infty\right)$ is the correct solution, so what am I doing wrong here.
case 2: if 5−a < 0Then the inequality holds for all eligible $a$ since the LHS > 0 > RHS. The eligible $a$ are determined by $5-a \lt 0 \iff a \gt 5$. – dxiv Nov 21 '16 at 07:33