Intuition The one core idea
A big ugly fraction of polynomials was built by adding together a few simple fractions over a common denominator — and adding is reversible, so we can un-add it back into those simple pieces. Every symbol below exists to answer one of two questions: "what were the simple pieces?" or "how do I integrate each simple piece?"
Before you can trust the partial fractions machine , you need every symbol it silently assumes. We build them one at a time, from nothing, each anchored to a picture.
A polynomial is a sum of terms, each a number times a whole-number power of x : for example x 2 − x − 2 or 3 x + 1 . The degree is the biggest power that appears. So x 2 − x − 2 has degree 2 ; 3 x + 1 has degree 1 ; a lone number like 7 has degree 0 .
Think of the degree as "how curvy the graph is allowed to be": degree 1 is a straight line, degree 2 is one bend (a parabola), higher degrees wiggle more.
Definition Rational function
A rational function is one polynomial divided by another: Q ( x ) P ( x ) . Here P (top, the numerator ) and Q (bottom, the denominator ) are both polynomials. The letters P and Q are just names — P for the particular top, Q for the quotient's bottom.
deg
deg P is shorthand for "the degree of P ". So deg ( 3 x + 1 ) = 1 . We write deg P < deg Q to say "the top is less curvy than the bottom".
Intuition Why "proper" matters
Look at the figure. When the top has smaller degree than the bottom (deg P < deg Q ), the fraction shrinks toward 0 as x grows huge — it is proper . When the top is as curvy or curvier (deg P ≥ deg Q ), the fraction grows without bound — it is improper , and no sum of shrinking proper pieces could ever equal it. That is why the recipe demands proper input first.
To factor a polynomial is to write it as a product of smaller polynomials, like turning x 2 − x − 2 into ( x − 2 ) ( x + 1 ) . Each piece in the product is a factor . (See Factoring polynomials over the reals .)
A root of a polynomial is a value of x that makes it equal 0 . The root of ( x − 2 ) is x = 2 , because plugging x = 2 gives 2 − 2 = 0 . Roots and factors are two views of the same thing: ( x − a ) is a factor exactly when x = a is a root.
Definition Linear vs. quadratic factor
A linear factor has degree 1 : ( x − a ) . Its graph crosses zero once (one root).
A quadratic factor has degree 2 : x 2 + b x + c . It may cross zero twice, touch once, or never.
Definition The discriminant
For x 2 + b x + c the discriminant is the number b 2 − 4 c . It answers one question: does this quadratic hit zero on the real number line?
b 2 − 4 c > 0 : two real roots → it factors into two linear pieces.
b 2 − 4 c = 0 : one repeated real root.
b 2 − 4 c < 0 : no real roots → it is irreducible (cannot be broken into real linear factors).
Worked example Discriminant in action
x 2 + 1 has b = 0 , c = 1 , so discriminant = 0 2 − 4 ( 1 ) = − 4 < 0 → irreducible (never touches the axis; look at the pink curve in the figure). But x 2 − 5 x + 6 has 25 − 24 = 1 > 0 → it factors as ( x − 2 ) ( x − 3 ) , so it is not allowed to keep its quadratic template.
Definition Repeated factor and the exponent
k
( x − a ) k means the factor ( x − a ) multiplied by itself k times. The small raised number k is an exponent : ( x − 1 ) 2 = ( x − 1 ) ( x − 1 ) . Graphically, a repeated root is where the curve touches the axis and turns back instead of crossing straight through.
The parent note gives repeated factors a whole stack of fractions (x − a A 1 + ( x − a ) 2 A 2 + … ). Section 6 below explains why one term is never enough.
Definition Unknown numerators
The letters A , B , C , … are placeholders for numbers we don't yet know . Partial fractions writes the answer as a template full of these unknowns, then solves for them. Subscripts like A 1 , A 2 just number several unknowns of the same "family" (the powers of one repeated factor). A linear top B x + C means "some unknown number B times x , plus some unknown number C ".
∑
j = 1 ∑ m is shorthand for "add these up, letting j run 1 , 2 , … , m ". So ∑ j = 1 3 ( x − a ) j A j = x − a A 1 + ( x − a ) 2 A 2 + ( x − a ) 3 A 3 . It is just a compact way to write a long sum.
Once a fraction is split, each piece integrates using ONE of three standard results. Here is why each tool and not another .
Definition The integral sign and
d x
∫ f ( x ) d x asks: "what function has f ( x ) as its rate of change?" The d x names x as the variable we integrate against. The trailing + C is the arbitrary constant (any constant has zero rate of change, so it's invisible to differentiation).
Definition Substitution — the glue tool
Integration by substitution lets us rename a chunk of the integrand as a single letter u , integrate in u , then rename back. The parent uses it when it writes ∫ x 2 + 1 x d x = 2 1 ln ( x 2 + 1 ) : set u = x 2 + 1 , so the x d x on top becomes 2 1 d u , turning the whole thing into the log integral above.
Definition Completing the square
Completing the square rewrites x 2 + b x + c as ( x + 2 b ) 2 + d 2 — a perfect square plus a leftover constant. Why? The arctan tool needs the exact shape ( stuff ) 2 + ( constant ) 2 . Completing the square forces any irreducible quadratic into that shape so Tool 3 can fire.
Intuition Numerator degree is one less than denominator
Over a common denominator, a single fraction's top can only reach a degree one below its bottom. A linear bottom ( x − a ) (degree 1) allows only a constant top A (degree 0). A quadratic bottom (degree 2) allows a linear top B x + C (degree 1). That single rule generates every template in the parent note — and it's why a repeated factor needs all lower powers to recover the missing freedom.
Discriminant tells irreducible
Test yourself — each line reveals the answer.
What does deg P < deg Q mean in words, and why must it hold before splitting? The numerator is less curvy than the denominator (proper); only then can shrinking simple pieces sum to it.
When is x 2 + b x + c irreducible? When the discriminant b 2 − 4 c < 0 (no real roots — the parabola never touches the axis).
What is the discriminant of x 2 − 5 x + 6 , and what does it tell you? 25 − 24 = 1 > 0 , so it factors into ( x − 2 ) ( x − 3 ) — treat as two linear pieces, not a quadratic.
Why does a quadratic factor get a numerator B x + C and not just A ? A top can reach degree one below its bottom; a degree-2 bottom allows a degree-1 top.
Which integral gives a log, and why? ∫ x − a 1 d x = ln ∣ x − a ∣ , because ln is defined as the function whose rate of change is x 1 .
Why does ∫ ( x − a ) − 2 d x give − ( x − a ) − 1 and not a log? The log only comes from exponent − 1 ; every other power uses the ordinary power rule.
What shape must a quadratic be forced into before arctan works, and how? ( x + 2 b ) 2 + d 2 , via completing the square, to match square + const 1 .
What does arctan answer? "Which angle has this tangent?" — it converts a square + const 1 piece into an angle.
What does the subscript in A 1 , A 2 record? Which power of a repeated factor each unknown belongs to.