The equation (x)^(4) − 5(x)^(2) + 4 = 0 has degree four, but one structure repeats: (x)^(4) is the square of (x)^(2). Writing t = (x)^(2) produces a familiar quadratic. We must then return from t to every possible x, rather than answer a question about a different variable.

We need factoring and quadratic equations. We work over the real numbers; complex solutions are outside this lesson.

A substitution comes with a restriction

A biquadratic equation has the form a(x)^(4) + b(x)^(2) + c = 0, a ≠ 0. Substituting t = (x)^(2) gives a(t)^(2) + bt + c = 0, necessarily with t ≥ 0.

Here (t)^(2) − 5t + 4 = 0, or (t − 1)(t − 4) = 0. Thus t = 1 or 4. This is not yet the final answer:

(x)^(2) = 1 or (x)^(2) = 4.

The first gives x = ± 1, the second x = ± 2. All four fit. Check: when (x)^(2) = 1, the original expression is 1 − 5 + 4 = 0; when (x)^(2) = 4, it is 16 − 20 + 4 = 0.

x⁴ − 5x² + 4 = 0t = x² ≥ 0t² − 5t + 4 = 0t = 1t = 4x = −1, 1x = −2, 2
The substitution t=x² requires t ≥ 0. Both t values are allowed. Returning gives all four x values: −2, −1, 1, 2; the negative sign must not be omitted.

The substitution is not one-to-one: 1 and −1 give the same t = 1. That is why the return journey branches.

A root for t may give no real x

For (x)^(4) + 3(x)^(2) − 4 = 0, the same substitution gives (t − 1)(t + 4) = 0. The number t = − 4 correctly solves the quadratic in t, but cannot be the square of a real x. Reject it because t ≥ 0; only x = ± 1 remain.

We can also substitute an entire repeated expression, such as t = (x + 1)^(2). The restriction t ≥ 0 remains, but we must return to (x + 1)^(2) = t, not (x)^(2) = t. Equations in quadratic form, OpenStax §2.6.

Factoring offers another way to reduce degree

Let P(x) = (x)^(3) − 4(x)^(2) + x + 6. Trying x = 2 gives 8 − 16 + 2 + 6 = 0, suggesting the factor x − 2. To find the rest, divide the polynomial by this binomial, successively removing its leading term.

First take (x)^(2): subtracting (x − 2)(x)^(2) = (x)^(3) − 2(x)^(2) leaves − 2(x)^(2) + x + 6. Next take − 2x: subtract − 2(x)^(2) + 4x, leaving − 3x + 6. The last quotient term, −3, removes the rest. Hence:

P(x) = (x − 2)((x)^(2) − 2x − 3).

The quotient is (x)^(2) − 2x − 3; the remainder is 0. Multiply back to check all four coefficients of P. In general, division by a nonzero polynomial B gives an identity P = BQ + R, with R = 0 or the degree of R below that of B. Polynomial division, OpenStax §5.4.

The remainder theorem explains substitution

For divisor x − r, the remainder is constant: P(x) = (x − r)Q(x) + R. Substitute x = r; the product vanishes, so R = P(r). This is the polynomial remainder theorem, also called Bézout’s theorem on remainders. Thus r is a root if and only if x − r divides the polynomial exactly. Remainder and factor theorems, OpenStax §5.5.

Here (x)^(2) − 2x − 3 = (x − 3)(x + 1), so P(x) = 0 gives x = 2,3, − 1. We did not divide the equation by x − 2 and lose 2: we considered the zero of every factor. Success with the trial number 2 does not imply that trial will always find a root quickly.

Branch · A checkDivision works even without a known root.

When the remainder is nonzero

For P(x) = (x)^(2) + 1, we have P(x) = (x − 2)(x + 2) + 5. The remainder 5 equals P(2), so 2 is not a root. The identity holds even at x = 2, although the numerical fraction P(x)/(x − 2) is undefined there. Polynomial identities and quotient domains are different questions.

Degree reduction works when we preserve the link to the original problem: a substitution’s restriction and every return branch, or every factor of a zero product. A shorter intermediate expression is useful together with that link.