OakRoadSystems Articles Math Polynomial Solving

Solving Polynomial Equations

revised 3 Jan 2013 Copyright 20022013 by Stan Brown, Oak Road Systems Summary: In algebra you spend lots of time solving polynomial equations or factoring polynomials (which is the same thing). It would be easy to get lost in all the techniques, but this paper ties them all together in a coherent whole. The Master Plan Factor = Root Exact or Approximate? Step by Step Cubic and Quartic Formulas Step 1. Standard Form and Simplify Step 2. How Many Roots? Descartes Rule of Signs Complex Roots Irrational Roots Multiple Roots Step 3. Quadratic Factors Step 4. Find One Factor or Root Monomial Factors Special Products Rational Roots Graphical Clues Boundaries on Roots Step 5. Divide by Your Factor Step 6. Numerical Methods Complete Example Whats New Copying: Youre welcome to print copies of this page for your own use, and to link from your own Web pages to this page. But please dont make any electronic copies and publish them on your Web page or elsewhere.

Contents:

The Master Plan

Factor = Root
Make sure you arent confused by the terminology. All of these are the same: Solving a polynomial equation p(x) = 0 Finding roots of a polynomial equation p(x) = 0 Finding zeroes of a polynomial function p(x) Factoring a polynomial function p(x) Theres a factor for every root, and vice versa. (xr) is a factor if and only if r is a root. This is the Factor Theorem: nding the roots or nding the factors is essentially the same thing. (The main dierence is how you treat a constant factor.)

Exact or Approximate?
Most often when we talk about solving an equation or factoring a polynomial, we mean an exact (or analytic) solution. The other type, approximate (or numeric) solution, is always possible and sometimes is the only possibility. When you can nd it, an exact solution is be er. You can always nd a numerical approximation to an exact solution, but going the other way is much more dicult. This page spends most of its time on methods for exact solutions, but also tells you what to do when analytic methods fail.

Step by Step
How do you nd the factors or zeroes of a polynomial (or the roots of a polynomial equation)? Basically, you whi le. Every time you chip a factor or root o the polynomial, youre left with a polynomial that is one degree simpler. Use that new reduced polynomial to nd the remaining factors or roots. At any stage in the procedure, if you get to a cubic or quartic equation (degree 3 or 4), you have a choice of continuing with factoring or using the cubic or quartic formulas. These formulas are a lot of work, so most people prefer to keep factoring. Follow this procedure step by step: 1. If solving an equation, put it in standard form with 0 on one side and simplify. [ details ] 2. Know how many roots to expect. [ details ] 3. If youre down to a linear or quadratic equation (degree 1 or 2), solve by inspection or the quadratic formula. [ details ] Then go to step 7. 4. Find one rational factor or root. This is the hard part, but there are lots of techniques to help you. [ details ] If you can nd a factor or root, continue with step 5 below; if you cant, go to step 6. 5. Divide by your factor. This leaves you with a new reduced polynomial whose degree is 1 less. [ details ] For the rest of the problem, youll work with the reduced polynomial and not the original. Continue at step 3. 6. If you cant nd a factor or root, turn to numerical methods. [ details ] Then go to step 7. 7. If this was an equation to solve, write down the roots. If it was a polynomial to factor, write it in factored form, including any constant factors you took out in step 1. This is an example of an algorithm, a set of steps that will lead to a desired result in a nite number of operations. Its an iterative strategy, because the middle steps are repeated as long as necessary.

Cubic and Quartic Formulas

The methods given herend a rational root and use synthetic divisionare the easiest. But if you cant nd a rational root, there are special methods for cubic equations [URL h p://mathworld.wolfram.com /CubicFormula.html accessed 2013-01-03] (degree 3) and quartic equations [URL h p://mathworld.wolfram.com /QuarticEquation.html accessed 2013-01-03] (degree 4), both at Mathworld. An alternative approach is provided by Dick Nickalls in PDF for cubic [URL h p://www.nickalls.org/dick/papers/maths/cubic1993.pdf accessed 2013-01-03] and quartic [URL h p://www.nickalls.org/dick/papers/maths/quartic2009.pdf accessed 2013-01-03] equations.

Step 1. Standard Form and Simplify

This is an easy stepeasy to overlook, unfortunately. If you have a polynomial equation, put all terms on one side and 0 on the other. And whether its a factoring problem or an equation to solve, put your polynomial in standard form, from highest to lowest power. For instance, you cannot solve this equation in this form: x + 6x + 12x = 8 You must change it to this form: x + 6x + 12x + 8 = 0 Also make sure you have simplied, by factoring out any common factors. This may include factoring out a 1 so that the highest power has a positive coecient. Example: to factor 7 6x 15x 2x begin by pu ing it in standard form: 2x 15x 6x + 7 and then factor out the 1 (2x + 15x + 6x 7) or (1)(2x + 15x + 6x 7) If youre solving an equation, you can throw away any common constant factor. But if youre factoring a polynomial, you must keep the common factor. Example: To solve 8x + 16x + 8 = 0, you can divide left and right by the common factor 8. The equation x + 2x + 1 = 0 has the same roots as the original equation. Example: To factor 8x + 16x + 8 , you recognize the common factor of 8 and rewrite the polynomial as 8(x + 2x + 1), which is identical to the original polynomial. (While its true that you will focus your further factoring eorts on x + 2x + 1, it would be an error to write that the original polynomial equals x + 2x + 1.) Your common factor may be a fraction, because you must factor out any fractions so that the polynomial has integer coecients. Example: To solve (1/3)x + (3/4)x (1/2)x + 5/6 = 0, you recognize the common factor of 1/12 and divide both sides by 1/12. This is exactly the same as recognizing and multiplying by the lowest common denominator of 12. Either way, you get 4x + 9x 6x + 10 = 0, which has the same roots as the original equation. Example: To factor (1/3)x + (3/4)x (1/2)x + 5/6, you recognize the common factor of 1/12 (or the lowest common denominator of 12) and factor out 1/12. You get (1/12)(4x + 9x 6x + 10), which is identical to the original polynomial.

Step 2. How Many Roots?

A polynomial of degree n will have n roots, some of which may be multiple roots. How do you know this is true? The Fundamental Theorem of Algebra tells you that the polynomial has at least one root. The Factor Theorem tells you that if r is a root then (xr) is a factor. But if you divide a polynomial of degree n by a factor (xr), whose degree is 1, you get a polynomial of degree n1. Repeatedly applying the Fundamental Theorem and Factor Theorem gives you n roots and n factors.

Descartes Rule of Signs

Descartes Rule of Signs can tell you how many positive and how many negative real zeroes the polynomial has. This is a big labor-saving device, especially when youre deciding which possible rational roots to pursue. To apply Descartes Rule of Signs, you need to understand the term variation in sign. When the polynomial is arranged in standard form, a variation in sign occurs when the sign of a coecient is dierent from the sign of the preceding coecient. (A zero coecient is ignored.) For example, p(x) = x5 2x3 + 2x2 3x + 12 has four variations in sign.

Descartes Rule of Signs: The number of positive roots of p(x)=0 is either equal to the number of variations in sign of p(x), or less than that by an even number. The number of negative roots of p(x)=0 is either equal to the number of variations in sign of p(x), or less than that by an even number. Example: Consider p(x) above. Since it has four variations in sign, there must be either four positive roots, two positive roots, or no positive roots. Now form p(x), by replacing x with (x) in the above: p(x) =(x)5 2(x)3 + 2(x)2 3(x) + 12 p(x) = x5 + 2x3 + 2x2 + 3x + 12 p(x) has one variation in sign, and therefore the original p(x) has one negative root. Since you know that p(x) must have a negative root, but it may or may not have any positive roots, you would look rst for negative roots. p(x) is a fthdegree polynomial, and therefore it must have ve zeros. Since x is not a factor, you know that x=0 is not a zero of the polynomial. (For a polynomial with real coecients, like this one, complex roots occur in pairs.) Therefore there are three possibilities: number of zeroes that are positive rst possibility second possibility third possibility 4 2 0 negative 1 1 1 complex not real 0 2 4

Complex Roots
If a polynomial has real coecients, then either all roots are real or there are an even number of non-real complex roots, in conjugate pairs. For example, if 5+2i is a zero of a polynomial with real coecients, then 52i must also be a zero of that polynomial. It is equally true that if (x52i) is a factor then (x5+2i) is also a factor. Why is this true? Because when you have a factor with an imaginary part and multiply it by its complex conjugate you get a real result: (x52i)(x5+2i) = x10x+254i = x10x+29 If (x52i) was a factor but (x5+2i) was not, then the polynomial would end up with imaginaries in its coecients, no ma er what the other factors might be. If the polynomial has only real coecients, then any complex roots must occur in conjugate pairs.

Irrational Roots
For similar reasons, if the polynomial has rational coecients then the irrational roots involving square roots occur (if at all) in conjugate pairs. If (x2+3) is a factor of a polynomial with rational coecients, then (x23) must also be a factor. (To see why, remember how you rationalize a binomial denominator; or just check what happens when you multiply those two factors.) As Je Beckman pointed out (20 June 2006), this is emphatically not true for odd roots. For instance, x2 = 0 has three roots, 2^(1/3) and two complex roots. Its an interesting problem whether irrationals involving even roots of order 4 must also occur in conjugate pairs. I dont have an immediate answer. Im working on a proof, as I have time.

Multiple Roots

When a given factor (xr) occurs m times in a polynomial, r is called a multiple root or a root of multiplicity m If the multiplicity m is an even number, the graph touches the x axis at x=r but does not cross it. If the multiplicity m is an odd number, the graph crosses the x axis at x=r. If the multiplicity is 3, 5, 7, and so on, the graph is horizontal at the point where it crosses the axis. Examples: Compare these two polynomials and their graphs: f(x) = (x1)(x4)2 = x3 9x2 + 24x 16 g(x) = (x1)3(x4)2 = x5 11x4 + 43x3 73x2 + 56x 16 These polynomials have the same zeroes, but the root 1 occurs with dierent multiplicities. Look at the graphs:

Both polynomials have zeroes at 1 and 4 only. f(x) has degree 3, which means three roots. You see from the factors that 1 is a root of multiplicity 1 and 4 is a root of multiplicity 2. Therefore the graph crosses the axis at x=1 (but is not horizontal there) and touches at x=4 without crossing. By contrast, g(x) has degree 5. (g(x) = f(x) times (x-1) 2.) Of the ve roots, 1 occurs with multiplicity 3: the graph crosses the axis at x=1 and is horizontal there; 4 occurs with multiplicity 2, and the graph touches the axis at x=4 without crossing.

Step 3. Quadratic Factors

When you have quadratic factors (Ax+Bx+C), it may or may not be possible to factor them further. Sometimes you can just see the factors, as with xx6 = (x+2)(x3). Other times its not so obvious whether the quadratic can be factored. Thats when the quadratic formula (shown at right) is your friend. For example, suppose you have a factor of 12xx35. Can that be factored further? By trial and error youd have to try a lot of combinations! Instead, use the fact that factors correspond to roots, and apply the formula to nd the roots of 12xx35 = 0, like this: x = [ (1) [1 4(12)(35)] ] / 2(12) x = [ 1 1681 ] / 24 1681 = 41, and therefore x = [ 1 41 ] / 24 x = 42/24 or 40/24 x = 7/4 or 5/3 If 7/4 and 5/3 are roots, then (x7/4) and (x+5/3) are factors. Therefore 12xx35 = (4x7)(3x+5) What about x5x+7? This one looks like its prime, but how can you be sure? Again, apply the formula: x = [ (5) [25 4(1)(7)] ] / 2(1) x = [ 5 (3) ] / 2 What you do with that depends on the original problem. If it was to factor over the reals, then x5x+7 is prime. But if that factor was part of an equation and you were supposed to nd all complex roots, you have two of them: x = 5/2 + ((3)/2)i, x = 5/2 ((3)/2)i Since the original equation had real coecients, these complex roots occur in a conjugate pair.

Step 4. Find One Factor or Root

This step is the heart of factoring a polynomial or solving a polynomial equation. There are a lot of techniques that can help you to nd a factor. Sometimes you can nd factors by inspection (see the rst two sections that follow). This provides a great shortcut, so check for easy factors before starting more strenuous methods.

Monomial Factors
Always start by looking for any monomial factors you can see. For instance, if your function is f(x) = 4x6 + 12x5 + 12x4 + 4x3 you should immediately factor it as f(x) = 4x3(x3 + 3x2 + 3x + 1) Ge ing the 4 out of there simplies the remaining numbers, the x3 gives you a root of x = 0 (with multiplicity 3), and now you have only a cubic polynomial (degree 3) instead of a sextic (degree 6). I n fact, you should now recognize that cubic as a special product, the perfect cube (x+1)3. When you factor out a common variable factor, be sure you remember it at the end when youre listing the factor or roots. x+3x+3x+1 = 0 has certain roots, but x(x+3x+3x+1) = 0 has those same roots and also a root at x=0.

Special Products
Be alert for applications of the Special Products. If you can apply them, your task becomes much easier. The Special Products are perfect square (2 forms): A 2AB + B = (A B) sum of squares: A + B cannot be factored on the reals, in general (for exceptional cases see Factoring the Sum of Squares) dierence of squares: A B = (A + B)(A B) perfect cube (2 forms): A 3AB + 3AB B = (A B) sum of cubes: A + B = (A + B)(A AB + B) dierence of cubes: A B = (A B)(A + AB + B) The expressions for the sum or dierence of two cubes look as though they ought to factor further, but they dont. AAB+B is prime over the reals. Consider p(x) = 27x 64 You should recognize this as p(x) = (3x) 4 You know how to factor the dierence of two cubes: p(x) = (3x4)(9x+12x+16) Bingo! As soon as you get down to a quadratic, you can apply the Quadratic Formula and youre done. Heres another example: q(x) = x6 + 16x3 + 64 This is just a perfect square trinomial, but in x3 instead of x. You factor it exactly the same way: q(x) = (x3)2 + 2(8)(x3) + 82 q(x) = (x3 + 8)2 And you can easily factor (x3+8)2 as (x+2)2(x22x+4)2.

Rational Roots

Assuming youve already factored out the easy monomial factors and special products, what do you do if youve still got a polynomial of degree 3 or higher? The answer is the Rational Root Test. It can show you some candidate roots when you dont see how to factor the polynomial, as follows. Consider a polynomial in standard form, wri en from highest degree to lowest and with only integer coecients: f(x) = anxn + ... + ao The Rational Root Theorem tells you that if the polynomial has a rational zero then it must be a fraction p/q, where p is a factor of the trailing constant ao and q is a factor of the leading coecient an. Example: p(x) = 2x4 11x3 6x2 + 64x + 32 The factors of the leading coecient (2) are 2 and 1. The factors of the constant term (32) are 1, 2, 4, 8, 16, and 32. Therefore the possible rational zeroes are 1, 2, 4, 8, 16, or 32 divided by 2 or 1: 1/2, 1/1, 2/2, 2/1, 4/2, 4/1, 8/2, 8/1, 16/2, 16/1, 32/2, 32/1 reduced: , 1, 2, 4, 8, 16, 32 What do we mean by saying this is a list of all the possible rational roots? We mean that no other rational number, like or 32/7, can be a root of p(x) = 0. Caution: Dont make the Rational Root Test out to be more than it is. It doesnt say those rational numbers are roots, just that no other rational numbers can be roots. And it doesnt tell you anything about whether some irrational or even complex roots exist. The Rational Root Test is only a starting point. Suppose you have a polynomial with non-integer coecients. Are you stuck? No, you can factor out the least common denominator (LCD) and get a polynomial with integer coecients that way. Example: (1/2)x (3/2)x + (2/3)x 1/2 The LCD is 1/6. Factoring out 1/6 gives the polynomial (1/6)(3x 9x + 4x 3) The two forms are equivalent, and therefore they have the same roots. But you cant apply the Rational Root Test to the rst form, only to the second. The test tells you that the only possible rational roots are 1/3, 1, 3. Once youve identied the possible rational zeroes, how can you screen them? The brute-force method would be to take each possible value and substitute it for x in the polynomial: if the result is zero then that number is a root. But theres a be er way. Use Synthetic Division to see if each candidate makes the polynomial equal zero. This is be er for three reasons. First, its computationally easier, because you dont have to compute higher powers of numbers. Second, at the same time it tells you whether a given number is a root, it produces the reduced polynomial that youll use to nd the remaining roots. Finally, the results of synthetic division may give you an upper or lower bound even if the number youre testing turns out not to be a root. Sometimes Descartes Rule of Signs can help you screen the possible rational roots further. For example, the Rational Root Test tells you that if q(x) = 2x4 + 13x3 + 20x2 + 28x + 8 has any rational roots, they must come from the list , 1, 2, 4, 8. But dont just start o substituting or synthetic dividing. Since there are no sign changes, there are no positive roots. Are there any negative roots? q(x) = 2x4 13x3 + 20x2 28x + 8 has four sign changes. Therefore there could be as many as four negative roots. (There could also be two negative roots, or none.) Theres no guarantee that any of the roots are rational, but any root that is rational must come from the list , 1, 2, 4, 8. (If you have a graphing calculator, you can pre-screen the rational roots by graphing the polynomial and seeing where it seems to cross the x axis. But you still need to verify the root algebraically, to see that f(x) is exactly 0 there, not just nearly 0.) Remember, the Rational Root Test guarantees to nd all rational roots. But it will completely miss real roots that are not rational, like the roots of x2=0, which are 2, or the roots of x+4=0, which are 2i. Finally, remember that the Rational Root Test works only if all coecients are integers. Look again at this function, which is graphed at right: p(x) = 2x4 11x3 6x2 + 64x + 32 The Rational Root Theorem tells you that the only possible rational zeroes are , 1, 2, 4, 8, 16, 32. But suppose you factor out the 2 (as I once did in class), writing the equivalent function

p(x) = 2(x4 (11/2)x3 3x2 + 32x + 16) This function is the same as the earlier one, but you can no longer apply the Rational Root Test because the coecients are not integers. In fact is a zero of p(x), but it did not show up when I (illegally) applied the Rational Root Test to the second form. My mistake was forge ing that the Rational Root Theorem applies only when all coecients of the polynomial are integers.

Graphical Clues
By graphing the functioneither by hand or with a graphing calculatoryou can get a sense of where the roots are, approximately, and how many real roots exist. Example: If the Rational Root Test tells you that 2 are possible rational roots, you can look at the graph to see if it crosses (or touches) the x axis at 2 or 2. If so, use synthetic division to verify that the suspected root actually is a root. Yes, you always need to checkfrom the graph you can never be sure whether the intercept is at your possible rational root or just near it.

Boundaries on Roots
Some techniques dont tell you the specic value of a root, but rather that a root exists between two values or that all roots are less than a certain number of greater than a certain number. This helps narrow down your search.

Intermediate Value Theorem

This theorem tells you that if the graph of a polynomial is above the x axis for one value of x and below the x axis for another value of x, it must cross the x axis somewhere between. (If you can graph the function, the crossings will usually be be obvious.) Example: p(x) = 3x + 4x 20x 32 The rational roots (if any) must come from the list 1/3, 2/3, 1, 4/3, 2, 8/3, 4, 16/3, 8, 32/3, 16, 32. Naturally youll look at the integers rst, because the arithmetic is easier. Trying synthetic division, you nd p(1) = 45, p(2) = 22, and p(4) = 144. Since p(2) and p(4) have opposite signs, you know that the graph crosses the axis between x=2 and x=4, so there is at least one root between those numbers. In other words, either 8/3 is a root, or there root(s) between 2 and 4 are irrational. (In fact, synthetic division reveals that 8/3 is a root.) The Intermediate Value Theorem can tell you where there is a root, but it cant tell you where there is no root. For example, consider q(x) = 4x 16x + 15 q(1) and q(3) are both positive, but that doesnt tell you whether the graph might touch or cross the axis between. (It actually crosses the axis twice, at x = 3/2 and x = 5/2.)

Upper and Lower Bounds

One side eect of synthetic division is that even if the number youre testing turns out as not a root, it may tell you that all the roots are smaller or larger than that number: If you do synthetic division by a positive number a, and every number in the bo om row is positive or zero, then a is an upper bound for the roots, meaning that all the real roots are a. If you do synthetic division by a negative number b, and the numbers in the bo om row alternate sign, then b is a lower bound for the roots, meaning that all the real roots are b. What if the bo om row contains zeroes? A more complete statement is that alternating nonnegative

and nonpositive signs, after synthetic division by a negative number, show a lower bound on the root. The next two examples clarify that. (By the way, the rule for lower bounds follows from the rule for upper bounds. Lower limits on roots of p(x) equal upper limits on roots of p(x), and dividing by (x+r) is the same as dividing by (xr).) Example: q(x) = x3 + 2x2 3x 4 Using the Rational Root Test, you identify the only possible rational roots as 4, 2, and 1. You decide to try 2 as a possible root, and you test it with synthetic division:
-2 | 1 2 -3 -4 | -2 0 6 |-----------------1 0 -3 2

2 is not a root of the equation f(x)=0. The third row shows alternating signs, and you were dividing by a negative number; however, that zero mucks things up. Recall that you have a lower bound only if the signs in the bo om row alternate nonpositive and nonnegative. The 1 is positive (nonnegative), and the 0 can count as nonpositive, but the 3 doesnt qualify as nonnegative. The alternation is broken, and you do not know whether there are roots smaller than 2. (In fact, graphical or numerical methods would show a root around 2.5.) Therefore you need to try the lower possible rational root, 4:
-4 | 1 2 -3 -4 | -4 8 -20 |-----------------1 -2 5 -24

Here the signs do alternate; therefore you know there are no roots below 4. (The remainder 24 shows you that 4 itself isnt a root.) Heres another example: r(x) = x + 3x 3 The Rational Root Test tells you that the possible rational roots are 1 and 3. With synthetic division for 3:
-3 | 1 3 0 -3 | -3 0 0 |-----------------1 0 0 -3

3 is not a root, but the signs do alternate here, since the rst 0 counts as nonpositive and the second as nonnegative. Therefore 3 is a lower bound to the roots, meaning that the equation has no real roots lower than 3.

Coefficients and Roots

There is an interesting relationship between the coecients of a polynomial and its zeroes. I mention it last because it is more suited to forming a polynomial that has zeroes with desired properties, rather than nding zeroes of an existing polynomial. However, if you know all the roots of a polynomial but one or two, you can easily use this technique to nd the remaining root. Consider the polynomial f(x) = anxn + an1xn1 + an2xn2 + ... + a2x2 + a1x + ao The following relationships exist: an1an = sum of all the roots +an2an = sum of the products of roots taken two at a time an3an = sum of the products of roots taken three at a time and so forth, until (1)na0an = product of all the roots Example: f(x) = x3 6x2 7x 8 has degree 3, and therefore at most three real zeroes. If we write the real zeroes as r1, r2, r3, then the sum of the roots is r1+r2+r3 = (6) = 6; the sum of the products of roots taken two at a time is r1r2+r1r3+r2r3 = 7, and the product of the roots is r1r2r3 = (1)3(8) = 8. Example: Given that the polynomial g(x) = x5 11x4 + 43x3 73x2 + 56x 16 has a triple root at x=1, nd the other two roots. Solution: Let the other two roots be c and d. Then you know that the sum of the all roots is 1+1+1+c+d =

(11) = 11, or c+d = 8. You also know that the product of all the roots is 111cd = (1)5(16) = 16, or cd = 16. c+d = 8, cd = 16; therefore c = d = 4, so the remaining roots are a double root at x=4.

More Coefficients and Roots

There are several further theorems about the relationship between coecients and roots. Wikipedias article Properties of Polynomial Roots [URL h p://en.wikipedia.org/wiki/Properties_of_polynomial_roots accessed 2013-01-03] gives a good though somewhat terse summary.

Step 5. Divide by Your Factor

Remember that r is a root if and only if xr is a factor; this is the Factor Theorem. So if you want to check whether r is a root, you can divide the polynomial by xr and see whether it comes out even (remainder of 0). Elizabeth Stapel has a nice example of dividing polynomials [URL h p://www.purplemath.com/modules/polydiv.htm accessed 2013-01-03] by long division. But its easier and faster to do synthetic division. If your synthetic division is a li le rusty, you might want to look at Dr. Maths short Synthetic Division tutorial [URL h p://mathforum.org/library/drmath/view/53056.html accessed 2013-01-03]; if you need a longer tutorial, Elizabeth Stapels Synthetic Division [URL h p://www.purplemath.com/modules/synthdiv.htm accessed 2013-01-03] is excellent. (Dr. Math also has a page on why Synthetic Division works [URL h p://mathforum.org/library/drmath/view/56406.html accessed 2013-01-03].) Synthetic division also has some side benets. If your suspected root actually is a root, synthetic division gives you the reduced polynomial. And sometimes you also luck out and synthetic division shows you an upper or lower bound on the roots. You can use synthetic division when youre dividing by a binomial of the form xr for a constant r. If youre dividing by x3, youre testing whether 3 is a root and you synthetic divide by 3 (not 3). If youre dividing by x+11, youre testing whether 11 is a root and you synthetic divide by 11 (not 11). Example: p(x) = 4x4 35x2 9 You suspect that x3 might be a factor, and you test it by synthetic division, like this:
3 | 4 0 -35 0 -9 | 12 36 3 9 |-------------------4 12 1 3 0

Since the remainder is 0, you know that 3 is a root of p(x) = 0, and x3 is a factor of p(x). But you know more. Since 3 is positive and the bo om row of the synthetic division is all positive or zero, you know that all the roots of p(x) = 0 must be 3. And you also know that p(x) = (x3)(4x3 + 12x2 + x + 3) 3 2 4x + 12x + x + 3 is the reduced polynomial. All of its factors are also factors of the original p(x), but its degree is one lower, so its easier to work with.

Step 6. Numerical Methods

When your equation has no more rational roots (or your polynomial has no more rational factors), you can turn to numerical methods to nd the approximate value of irrational roots: The Wikipedia article Root-nding title=accessed 2013-01-03 class=showURL>Algorithm has a decent summary, with pointers to specic methods. Many graphing calculators have a Solve or Root or Zero command to help you nd approximate roots. For instance, on the TI-83 or TI-84, you graph the function and then select [2nd] [Calc] [zero].

Complete Example
Solve for all complex roots: 4x + 15x 36 = 0 Step 1. The equation is already in standard form, with only zero on one side, and powers of x from highest to lowest. There are no common factors. Step 2. Since the equation has degree 3, there will be 3 roots. There is one variation in sign, and from Descartes Rule of Signs you know there must be one positive root. Examine the polynomial with x replacing x: 4x 15x 36 There are no variations in sign, which means there are no negative roots. The other two roots must therefore be complex conjugates. Steps 3 and 4. The possible rational roots are unfortunately rather numerous: any of 1, 2, 3, 4, 6, 9, 12, 18, 36 divided by any of 4, 2, 1. (Only positive roots are listed because you have already determined that there are no negative roots for this equation.) You decide to try 1 rst:
1 | 4 0 15 -36 | 4 4 19 |----------------4 4 19 -17

1 is not a root, so you test 2:

2 | 4 0 15 -36 | 8 16 62 |----------------4 8 31 26

Alas, 2 is not a root either. But notice that f(1) = 17 and f(2) = 26. They have opposite signs, which means that the graph crosses the x axis between x=1 and x=2, and a root is between 1 and 2. (In this case its the only root, since you have determined that there is one positive root and there are no negative roots.) The only possible rational root between 1 and 2 is 3/2, and therefore either 3/2 is a root or the root is irrational. You try 3/2 by synthetic division:
3/2 | 4 0 15 -36 | 6 9 36 |----------------4 6 24 0

Hooray! 3/2 is a root. The reduced polynomial is 4x + 6x + 24. In other words, (4x + 15x 36) (x3/2) = 4x + 6x + 24 The reduced polynomial has degree 2, so there is no need for more trial and error, and you continue to step 5. Step 5. Now you must solve 4x + 6x + 24 = 0 First divide out the common factor of 2: 2x + 3x + 12 = 0 Its no use trying to factor that quadratic, because you determined using Descartes Rule of Signs that there are no more real roots. So you use the quadratic formula: x = [ 3 [9 4(2)(12)] ] / 2(2) x = [ 3 (87) ] / 4 x = 3/4 ((87)/4)i Step 6. Remember that you found a root in an earlier step! The full list of roots is 3/2, 3/4 + ((87)/4)i, 3/4 ((87)/4) i

Whats New

3 Jan 2013: Add the sections Coecients and Roots and More Coecients and Roots. (intervening changes suppressed) 15 Feb 2002: First publication. this page: h p://oakroadsystems.com/math/polysol.htm