Wolfe conditions
From Wikipedia, the free encyclopedia
In (unconstrained) optimization, the Wolfe conditions are a set of inequalities for performing inexact linesearch, especially in quasi-Newton methods. Inexact line searches provide an efficient way of computing an acceptable step length α that reduces the cost 'sufficiently', rather than minimizing the cost over
exactly.
Let
be a smooth objective function, and
be a given search direction. A step length αk is said to satisfy the Wolfe conditions if the following two inequalities hold.
- i)
, - ii)
,
with 0 < c1 < c2 < 1. Inequality i) is known as the Armijo rule and ii) as the curvature condition; i) ensures that αk decreases f 'sufficiently', and ii) ensures that the slope of the function
at αk is greater than c2 times that at α = 0.
The Wolfe conditions, however, can result in a value for the step length that is not close to a minimizer of φ. If we modify the curvature condition to the following,
- iia)

then i) and iia) together form the so-called strong Wolfe conditions, and force αk to lie close to a critical point of φ.
The Goldstein conditions are similar but more commonly used with Newton methods (as opposed to quasi-Newton methods).
[edit] References
- J. Nocedal and S. J. Wright, Numerical optimization. Springer Verlag, New York, NY, 1999.

