Source
Adapted from my own handwritten and hand-typed notes from Year 1 to Year 3 of my EEE degree
Newton's method (Newton-Raphson) finds an approximate root of . Starting from a guess, it repeatedly follows the tangent line down to the -axis. A good starting guess often comes from the Intermediate Value Theorem.
Derivation
The tangent at is . Setting gives its -intercept, which becomes the next guess:
Steps
Work through it in this order:
- Write the problem as and find
- Pick near the root (use a sign change: and of opposite sign means a root in )
- Iterate , best done in a table
- Stop when successive values agree to the accuracy needed
It can fail if (horizontal tangent) or if is too far from the root.
Worked Example
Solve :
- and , so a root lies in ; take
| 0 | ||||
| 1 | ||||
| 2 | ||||
| 3 | ||||
| 4 |
Root .
Classic Uses
Reciprocal of without division: use .
Only multiplication and subtraction are needed, which suited early computers.
Square root of : use .
Example for with : , (true ).