Secant and Regula Falsi Methods: Root Finding Guide

The secant method and regula falsi (false position) get Newton-like speed without any derivative. Both draw a straight secant line through two points on the curve and use its x-axis crossing as the next guess. The secant method keeps the last two guesses; regula falsi keeps a bisection-style bracket around the root and takes the chord point across it.

These methods matter because derivatives are often unavailable or expensive — the function may come from a simulation, a measurement, or code that cannot be differentiated. The secant method gives nearly Newton-like speed from function values alone, while regula falsi adds the safety of keeping the root trapped.

Key Concepts

• Secant line: the straight line through (x_n−1, f(x_n−1)) and (x_n, f(x_n)), an approximation of the tangent.
• Secant method: uses the two most recent points — needs two starting guesses, no bracket, and can wander like Newton.
• Regula falsi: keeps f(a) × f(b) < 0 at every step, so it never loses the root — but one endpoint can get stuck, slowing convergence.
• Order of convergence: about 1.618 (the golden ratio) near a simple root — superlinear, faster than bisection, slower than Newton.

Formulas

• Secant iteration: x_n+1 = x_n − f(x_n) · (x_n − x_n−1) / (f(x_n) − f(x_n−1)). Requires f(x_n) ≠ f(x_n−1).
• Regula falsi point: c = b − f(b) · (b − a) / (f(b) − f(a)), then replace a or b with c to keep opposite signs, exactly like a bisection update but with the chord point instead of the midpoint.
• Assumptions: f is continuous, the two starting values are distinct, and for regula falsi the starting interval must bracket a root.

Worked Example

Solve f(x) = x³ − x − 2 = 0 by the secant method from x₀ = 1, x₁ = 2, with f(1) = −2, f(2) = 4. Step 1: x₂ = 2 − 4·(2 − 1)/(4 − (−2)) = 2 − 4/6 = 1.333333, and f(1.333333) = 2.370370 − 1.333333 − 2 = −0.962963. Step 2: x₃ = 1.333333 − (−0.962963)(1.333333 − 2)/((−0.962963) − 4) = 1.333333 + 0.129356 = 1.462689, with f(1.462689) ≈ −0.333, homing in on the true root 1.52138. Regula falsi from the bracket [1, 2] gives the same first point 1.333333; since f there is negative like f(1), the bracket becomes [1.333333, 2], and the next point is c = 2 − 4(2 − 1.333333)/(4 + 0.962963) = 1.462726 — the same value, now safely bracketed.

Common Misconceptions

• "The secant method always stays inside the starting interval." Unlike regula falsi, the secant method keeps no bracket, so guesses can jump far outside — it trades safety for speed.
• "Regula falsi is always faster than bisection." Usually yes, but a stuck endpoint slows it to linear crawling. Modified versions (like the Illinois method) fix this by halving the stuck function value.
• "Two starting guesses must bracket the root for the secant method." No — any two distinct starting points work, though points near the root converge faster.

Real-World Applications

• Finance: computing the internal rate of return (IRR), where the cash-flow function has no simple derivative.
• Chemical engineering: solving equilibrium equations from experimental data.
• Structural engineering: finding load levels at which a structure buckles, using simulation output only.
• Computer graphics: ray-surface intersection with procedurally defined surfaces.

The secant method replaces Newton's tangent with a chord through the last two points for fast superlinear convergence with no derivative; regula falsi puts the same chord inside a bisection-style bracket for guaranteed safety. Both shine when derivatives are unavailable — the choice is speed versus certainty.