5
$\begingroup$

I need to obtain an analytical solution to an equation of the following form: $$ (x-a)(x-b)(x-c)=d(x-e)(x-f), $$ where $a$, $b$, $c$, $d$, $e$, and $f$ are known numbers and $x$ is the variable.

Of course, the equation can be reduced to a "simple" equation for the roots of a 3rd-order polynomial, and its solution is provided by the Cardano formula (and other similar methods), but the result is too complex...

So the question arises: does any simpler method exist that makes use of the known roots of the "constituent polynomials" (left-hand and right-hand side of my equation)?

P.S. It is possible to try the Cardano method with the polynomial coefficients expressed through the roots $a$...$c$,$e$,$f$ and then to simplify the resulting gargantuan formulae using a CAS (computer algebra system) hoping that the CAS will cope with it... But I am not so sure that the CAS is powerful enough in the dark art of formulae simplification.

$\endgroup$
  • 1
    $\begingroup$ Suppose you want to find the roots of $x^3-x^2+4=0.$ It does not get get any easier (or harder) if you rewrite as $(x-0)(x-0)(x-0)=(x+2)(x-2).$ $\endgroup$ – Aaron Meyerowitz Apr 17 '16 at 22:27
9
$\begingroup$

As you point out, this is no harder than solving a general cubic. It turns out that it is also no easier.

For any choice of the $6$ parameters $a,b,c,d,e,f$ your desired equation can be multiplied out to a cubic $x^3+Ax^2+Bx+C$ which can be solved by Cardano's formula. You hope for a nicer solution perhaps based on the form. However one might expect that with $6$ parameters one can easily get $A,B,C$ to be anything desired. This turns out to be the case.

Here is one way assuming all parameters constrained to be real. The complex case is as easy.

Pick $a=b=0$ and $c=-(A+d)$ where $d$ is $+1$ or $-1$ according as $C \ge 0$ or $C \lt 0.$ Then $e$ and $f$ are the (real) roots $$\frac{Bd \pm \sqrt{B^2+4|C|}}{2}$$ of the quadratic equation. $$x^2-\frac{B}{d}x-\frac{C}{d}.$$

$\endgroup$
3
$\begingroup$

While a simple closed-form solution isn't available, you could try series expansions in the parameter $d$, if $d$ is small. Thus suppose you want the root that is $a$ at $d=0$. For simplicity, rewrite the equation as $$ X (X-B)(X-C) = d(X-E)(X-F)$$ where $X=x-a$, $B = b-a$, $C=c-a$, $E=e-a$, $F=f-a$. Then this root is $$\eqalign{ X &= {\frac {EF}{BC}}d+ \left( -{\frac {{E}^{2}F}{{B}^{2}{C}^{2}}}-{\frac {E{F}^{2}}{{B}^{2}{C}^{2}}}+{\frac {{E}^{2}{F}^{2}}{{B}^{2}{C}^{3}}}+{ \frac {{E}^{2}{F}^{2}}{{B}^{3}{C}^{2}}} \right) {d}^{2}\cr+ &\left( { \frac {{E}^{3}F}{{B}^{3}{C}^{3}}}+3\,{\frac {{E}^{2}{F}^{2}}{{B}^{3}{C }^{3}}}+{\frac {E{F}^{3}}{{B}^{3}{C}^{3}}}-3\,{\frac {{E}^{3}{F}^{2}}{ {B}^{3}{C}^{4}}}-3\,{\frac {{E}^{2}{F}^{3}}{{B}^{3}{C}^{4}}}+2\,{ \frac {{E}^{3}{F}^{3}}{{B}^{3}{C}^{5}}}-3\,{\frac {{E}^{3}{F}^{2}}{{B} ^{4}{C}^{3}}}-3\,{\frac {{E}^{2}{F}^{3}}{{B}^{4}{C}^{3}}}+3\,{\frac {{ E}^{3}{F}^{3}}{{B}^{4}{C}^{4}}}+2\,{\frac {{E}^{3}{F}^{3}}{{B}^{5}{C}^ {3}}} \right) {d}^{3}\cr \cr+&\ldots} $$ This (and as many terms as you want) can be obtained from the Lagrange–Bürmann formula.

$\endgroup$
1
$\begingroup$

The short answer is no. -- Take e.g. the case $a = b = c = 0$ and $d = e = f = 1$. Then the polynomial whose roots you need to find is $x^3-x^2+2x-1$, which has Galois group ${\rm S}_3$.

The same works for higher degrees as well, e.g. in order to solve the equation $$ (x-0)(x-0)(x-0)(x-0)(x-0) \ = \ (x-1)(x-1), $$ you need to find roots of the polynomial $x^5-x^2+2x-1$ whose Galois group is not solvable and whose roots hence cannot be written as radicals.

$\endgroup$
0
$\begingroup$

Perhaps in some cases, you can apply Rouché's Theorem

$\endgroup$

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.