Simple Unconstrained Optimization Method

Requires a Wolfram Notebook System
Interact on desktop, mobile and cloud with the free Wolfram Player or other Wolfram Language products.
This Demonstration shows the simple one-dimensional unconstrained maximization algorithm for finding the (local) maximum of the function near the starting value
. It is a nongradient method and it requires only the starting value
and the initial step value
.
Contributed by: Marko Petkovic (April 2013)
Open content licensed under CC BY-NC-SA
Snapshots
Details
Assume that the function has a single local maximum
on the interval
. For the given real numbers
and
, construct the sequences
and
in the following way:
.
It can be proved that the sequence converges to
. The method is simple and easy to implement. It does not require the differentiability of the function nor any additional condition on
and
. This method is also known as simplex method I and has a linear convergence.
Permanent Citation