x₂ for x² − 2
x₀ = 1, x₁ = 2: x₂ = 2 − 2(2 − 1)/(2 − (−1)) = 4/3
Use the chord through (1, −1) and (2, 2).
The secant method replaces the derivative in Newton's method by the slope of the line through the last two points: .
This page is easier if you are comfortable with:
For with , : , , so . Next, , , , converging to . Each step uses only function values, which helps when the derivative is expensive or unavailable (derivatives from first principles give the slope idea).
The order of convergence is (the golden ratio, Binet's formula): faster than bisection (the bisection error bound) but slower than Newton's method (Newton's quadratic convergence) per step, though it needs only one new evaluation per step. The method can fail if the two points give the same function value (division by zero) and does not keep a bracket.
It is the root-finding analogue of linear interpolation (Lagrange interpolation) and a close relative of regula falsi. Compare also Newton–Raphson and fixed-point iteration.
The secant method uses straight-line extrapolation through the last two points to find a root without derivatives.
The line crosses zero at 1.333, closer to the root 1.414 than either start.
These examples are worked by hand, step by step, so you can compare your own working.
Find for , , .
Find from , .
Find from , .
A few situations come up again and again. Knowing them saves time and helps you spot answers that cannot be right.
x₀ = 1, x₁ = 2: x₂ = 2 − 2(2 − 1)/(2 − (−1)) = 4/3
Use the chord through (1, −1) and (2, 2).
≈ 1.4
Next iterate.
√2 ≈ 1.4142
Converging.
xₙ₊₁ = xₙ − f(xₙ)(xₙ − xₙ₋₁)/(f(xₙ) − f(xₙ₋₁))
No derivative needed.
The secant method replaces the derivative in Newton’s method by the gradient of the chord: xₙ₊₁ = xₙ − f(xₙ)(xₙ − xₙ₋₁)/(f(xₙ) − f(xₙ₋₁)). Its order of convergence is about 1.618.
Solve x² − 2 = 0 with x₀ = 1, x₁ = 2. Find x₂.
Find x₃ from x₁ = 2, x₂ = 4/3.
Find the error of x₃ = 1.4 against √2 to 4 decimal places.
f(x) = x − cos x, x₀ = 0, x₁ = 1. Find x₂ to 4 decimal places.
f′(xₙ) in the denominator ✗
a chord gradient ✓
Secant avoids derivatives.
f(xₙ₋₁) − f(xₙ) ✗
f(xₙ) − f(xₙ₋₁) ✓
Keep the same order top and bottom.
bracketing required ✗
not required ✓
Secant need not bracket.
f(xₙ) = f(xₙ₋₁) ✗
choose different starting points ✓
A flat chord has no root.
Try each one on paper, then open the answer. Use the solver above to see the full working.
-3
5
1.75
-0.04
1.618
0
Unlimited fresh questions with instant marking. Choose a level, work it out on paper, then type your answer.
Solve: …
Newton’s method with the derivative replaced by a chord gradient.
About 1.618.
No.
It avoids computing f′.
Newton’s method.
bisection.
More in Numerical methods and optimisation and the formula sheets.