partial-fraction decomposition
When you solve an IVP by Laplace, the algebra hands you F(s) as a single ratio of polynomials — one big, complicated fraction. That fraction usually matches no row of your table. Partial-fraction decomposition is the algebraic workhorse that breaks the big fraction back into a sum of small, simple fractions, each of which you can recognize and invert. It is the main tool of inversion.
The idea is to run the addition-of-fractions process in reverse. You factor the denominator into its simplest pieces — linear factors like (s - 2) and irreducible quadratics like (s^2 + 4) — and then write the big fraction as a sum of unknown-numerator fractions over those pieces. A distinct linear factor (s - a) gets a term A/(s - a); a repeated factor (s - a)^2 needs both A/(s - a) and B/(s - a)^2; an irreducible quadratic (s^2 + b^2) gets a term (Cs + D)/(s^2 + b^2). You then solve for the unknown constants A, B, C, D by clearing denominators and matching coefficients (or by plugging in convenient values of s). The result is a sum where every piece is a table entry in disguise.
The reason this matters so much is that every piece it produces inverts on sight: A/(s - a) is A e^(at); a quadratic piece is a sine/cosine (possibly shifted); a repeated factor brings in a t. So decomposition is the bridge from 'the algebra gave me an unrecognizable F(s)' to 'I can read off f(t) line by line.' The shapes of the denominator's factors are not arbitrary either — they are exactly the roots of the characteristic equation, so partial fractions in the s-domain mirror the case analysis (real, repeated, complex roots) you would do by hand.
Decompose 1/((s - 1)(s + 2)) = A/(s - 1) + B/(s + 2). Clearing denominators: 1 = A(s + 2) + B(s - 1). Set s = 1: 1 = 3A, so A = 1/3. Set s = -2: 1 = -3B, so B = -1/3. Hence the inverse is (1/3)e^t - (1/3)e^(-2t).
Break one ugly fraction into simple ones, then invert each on sight.
The decomposition only works directly when the numerator's degree is lower than the denominator's. If it is not, do polynomial long division first to split off a polynomial part (which inverts to delta-like terms) before decomposing the proper remainder.