0

Consider the polynomial

$$2x^2-5x-3$$

We can easily factorise this polynomial (with a high-school trial and error approach) and get

$$(2x+1)(x-3)$$

However, we I'm interested in how one can approach the question more systematically, and search for integer coefficients of the following statement

$$(ax+b)(cx+d)$$

Equating the co-efficients we get the following set of simultaneous equations

  1. $ac=2$
  2. $ad+bc=-5$
  3. $bd=-3$

For $a, b, c, d \in \Bbb Z$

What sort of mathematics do we need to solve this system of equations (where we have fewer inequivalent equations than variables) and what would the working look like for this question?

Many thanks

  • 4
    You just need elementary facts on divisibility of integers. For instance, from $ac=2$ we get $a \in {\pm1 , \pm 2}$ because $2$ is prime. Similarly for $bd=-3$. – lhf Jul 21 '17 at 12:29
  • 1
    Writing $a, b, c, d \in \Bbb Z$ assumes that the roots are rational, which may fail to be true. For quadratic equations, there is no reason to bypass the so-simple explicit formulas, and the method of indeterminate coefficients will be much less systematic. –  Jul 21 '17 at 13:37
  • https://math.stackexchange.com/questions/1937742/different-ways-to-factor/1941684#1941684 – AlexSam Aug 02 '17 at 14:18

3 Answers3

0

In this specific case, you have, from your first equation: $$ac=2$$ and, since $a,c\in\mathbb{Z}$, and $2>0$ you have the following possible pairs: $$\begin{array}{|c|c|} \hline a & c\\ \hline 2 & 1\\ \hline 1 & 2\\ \hline -1 & -2\\ \hline -2 & -1\\ \hline \end{array}$$ In the same way, since $b,d\in\mathbb{Z}$, we have that: $$\begin{array}{|c|c|} \hline b & d\\ \hline 3 & -1\\ \hline 1 & -3\\ \hline -1 & 3\\ \hline -3 & 1\\ \hline \end{array}$$ Now, due to the second equation: $$ad+bc=-5$$ we can see that the solution is $(a,b,c,d)=(2,1,1,-3)$.

Now, in general, the problem of finding a factoriasation of $$p(x)=p_0+p_1x+p_2x^2,\ p_0,p_1,p_2\in\mathbb{Z}$$ in the form of: $$E(Ax+B)(Cx+D)$$ with integers coefficients, boils down to finding rational roots of the equation $f(x)=0$, since, if $\frac{a}{b}$ and $\frac{c}{d}$ are the two roots of $f(x)=0$, then: $$p(x)=p_2(x-\frac{a}{b})(x-\frac{c}{d})=\frac{p_2}{ad}(ax-b)(cx-d)$$ So, if $x=\frac{a}{b}$ is a rational root of $p(x)=0$ with $(a,b)=1$, we will have: $$p\left(\frac{a}{b}\right)=0\Rightarrow p_0+p_1\frac{a}{b}+p_2\frac{a^2}{b^2}=0\Rightarrow b^2p_0=a(p_1b+p_2a)\Rightarrow b^2p_0|a\overset{(a,b)=1}{\Rightarrow}p_0|a$$ So, our first result is that $p_0$ should be a divisor of $a$, if $\frac{a}{b}$ is a root of $p(x)=0$. Now, we also note that - in case $a\neq0$, since this is a trivial case, due to the fact that $p_0=0$ and $p(x)=x(p_1+p_2x)$: $$p_0+p_1\frac{a}{b}+p_2\frac{a^2}{b^2}=0\Leftrightarrow p_2+p_1\frac{b}{a}+p_0\frac{b^2}{a^2}=0$$ So, we can see that, as previously, $p_2|b$ is needed, as well.

In this way you can find several similar criteria.

Note: The conditions $p_0|a$ and $p_2|b$ are necessary conditions in order for $p$ to have rational roots, so it does not mean that $p$ has rational roots if $p_0|a$ and $p_2|b$, but that if $p$ has rational roots, then $p_0|a$ and $p_2|b$.

0

Consider the product \begin{align*} (2x + 1)(x - 3) & = 2x(x - 3) + 1(x - 3)\\ & = \color{blue}{2}x^2 \color{green}{- 6}x \color{green}{+ 1}x - \color{blue}{3}\\ & = \color{blue}{2}x^2 - \color{green}{5}x \color{blue}{- 3} \end{align*} Observe that the product of the two numbers whose sum is the linear coefficient is equal to the product of the quadratic and constant coefficients, that is, $$\color{blue}{(2)(-3)} = \color{green}{(-6)(1)}$$ In order to factor the polynomial, we wish to work backwards. \begin{align*} 2x^2 - 5x - 3 & = 2x^2 - 6x + x - 3 && \text{split the linear term}\\ & = 2x(x - 3) + 1(x - 3) && \text{factor by grouping}\\ & = (2x + 1)(x - 3) && \text{extract the common factor} \end{align*} To do so, we must figure out how to split the linear term. The key idea is that we must find two numbers with product $(2)(-3) = -6$ and sum $-5$.

In general, if we want to factor $ax^2 + bx + c$ with respect to the rational numbers, we must find two rational numbers with product $ac$ and sum $b$.

Suppose $ax^2 + bx + c$, with $a \neq 0$, is a quadratic polynomial with integer coefficients that has factorization $(rx + s)(tx + u)$. Then \begin{align*} ax^2 + bx + c & = (rx + s)(tx + u)\\ & = rx(tx + u) + s(tx + u)\\ & = \color{blue}{rt}x^2 + \color{green}{ru}x + \color{green}{st}x + \color{blue}{su}\\ & = \color{blue}{rt}x^2 + \color{green}{(ru + st)}x + \color{blue}{su} \end{align*} Matching coefficients suggests that \begin{align*} a & = rt\\ b & = ru + st\\ c & = su \end{align*} Moreover, observe that if this is true, then $$ac = (rt)(su) = rtsu = (ru)(st)$$ that is, the product of the two numbers whose sum is $b$ is equal to the product of $a$ and $c$.

Since $$ax^2 + bx + c = (rx + s)(tx + u) \tag{1}$$ is an algebraic identity, it is true for every real value of $x$. In particular, it is true when $x = 0$, $x = 1$, and $x = -1$.

Substituting $0$ for $x$ in equation 1 yields $$c = su$$

Substituting $1$ for $x$ in equation 1 yields \begin{align*} a + b + c & = (r + s)(t + u)\\ & = r(t + u) + s(t + u)\\ & = rt + ru + st + su \end{align*} Since $c = su$, we can cancel $c$ from the left hand side and $su$ from the right hand side of the equation to obtain $$a + b = rt + ru + st \tag{2}$$

Substituting $-1$ for $x$ in equation 1 yields \begin{align*} a - b + c & = (-r + s)(-t + u)\\ & = -r(-t + u) + s(-t + u)\\ & = rt - ru - st + su \end{align*} Since $c = su$, we can cancel $c$ from the left hand side and $su$ from the right hand side of the equation to obtain $$a - b = rt - ru - st \tag{3}$$

Adding equations 2 and 3 yields \begin{align*} 2a & = 2rt\\ a & = rt \end{align*} Subtracting equation 3 from equation 2 yields \begin{align*} 2b & = 2ru + 2st\\ b & = ru + st \end{align*} Hence, we have shown that $a = rt$, $b = ru + st$, and $c = su$. Therefore, we can factor $ax^2 + bx + c$ if we can find two integers with product $ac$ and sum $b$.

N. F. Taussig
  • 76,571
0

There is no shortcut to the resolution of polynomial equations. Whatever the approach, it will lead you back to the resolution of the original problem (be it directly or indirectly) and bring no simplification.

Anyway, you may be interested in the Rational Root theorem.

  • Any method that involves factoring integers (e.g. OP, or using the RRT= Rational Root Test) will be highly inefficient for computing rational roots when the coefficients are large. Instead one can use methods that isolate the real roots (e.g. Descartes rule of signs and Newton iteration) and then use RRT denominator bounds to refine the isolating interval enough to determine if the real root is rational, e.g. for efficient algorithms see Sagraloff and Mehlhorn: Computing Real Roots of Real Polynomials. 2015 – Bill Dubuque Jul 21 '17 at 15:07
  • Closely related algorithms are implemented in most all computer algebra systems (In fact I implement the one used in Macsyma). – Bill Dubuque Jan 29 '20 at 19:32