Numerical Computation

Newton-Raphson Method: Formula, Steps and When It Fails

Numerical Computation Study guides

Newton-Raphson finds roots fast using the tangent line: x(n+1) = x(n) - f(x(n)) / f'(x(n)). Learn the steps, quadratic convergence and the common failure cases.

Newton-Raphson solves f(x) = 0 using the tangent line. Start from a guess x0. Draw the tangent to the curve at x0 and take the point where it crosses the x-axis as the next guess. Algebraically, x(n+1) = x(n) - f(x(n)) / f'(x(n)). Repeat until successive values agree to the accuracy you need, for example |x(n+1) - x(n)| below a tolerance.

Near a simple root the method converges quadratically: the number of correct digits roughly doubles with every step, much faster than bisection. The price is that it needs the derivative and a good starting guess. It can fail when f'(x(n)) is zero or very small (the tangent is flat and the next guess flies far away), when the guess is near a local maximum or minimum, when the iteration cycles between points, or when the root is repeated, where convergence drops to linear.

A useful special case is computing a square root. To find the square root of a, solve f(x) = x^2 - a = 0. Then f'(x) = 2x and the formula becomes x(n+1) = (x(n) + a / x(n)) / 2, which is the classic Babylonian method. The secant method replaces the derivative by a slope through the last two points, which avoids computing f' at the cost of slightly slower convergence.

Key points

  • x(n+1) = x(n) - f(x(n)) / f'(x(n)).
  • Quadratic convergence near a simple root.
  • Fails if f'(x) is zero or the guess is poor.
  • Needs a derivative, bisection does not.
  • Square root of a: x(n+1) = (x(n) + a / x(n)) / 2.

Worked example

Solve x^2 - 2 = 0 starting at x0 = 1.5. f(x) = x^2 - 2 and f'(x) = 2x. x1 = 1.5 - (0.25) / 3 = 1.416667. x2 = 1.416667 - (0.006944) / 2.833333 = 1.414216. x3 = 1.414216 - (0.000006) / 2.828431 = 1.414214. After three steps the answer is correct to six decimals.

Common questions

Why is Newton-Raphson faster than bisection?

It uses the slope of the function, which gives quadratic convergence near a simple root, while bisection only halves the interval each step.

When does Newton-Raphson fail?

When the derivative is zero or tiny at an iterate, when the starting guess is far from the root, or when the iteration cycles.

Related topics

BASMAGLY is an independent student project, not an official Delta University site. Check lectures and your instructor for the final word.