Bisection
Requires a sign-changing bracket. Convergence is steady and easy to bound because every successful step halves the interval.
Root finding · CMNA Laboratory
Bisection narrows a bracket. Newton follows a derivative. Secant estimates that derivative from recent function values. The answer may be the same; the path almost never is.
Requires a sign-changing bracket. Convergence is steady and easy to bound because every successful step halves the interval.
Uses the derivative to jump toward the root. Local convergence can be excellent, but the initial value and derivative matter.
Replaces the exact derivative with a slope estimated from nearby points, trading some of Newton's speed for less information.
Compare convergence
The graph uses log10 |f(x)| so that several orders of magnitude of residual fit on one axis. Lower is better.
Iteration count is not the whole cost of a numerical method. Newton asks for a derivative evaluation at every step. Bisection asks only for function values but needs a valid bracket. The secant method avoids an explicit derivative, but the estimated slope can become poor or collapse.
The laboratory therefore treats the graph as a way to inspect behavior, not a scoreboard declaring one method universally superior. Numerical analysis is mostly the art of knowing which assumptions you can afford.