In the unconstrained minimization problem, the Wolfe conditions are a set of inequalities for performing inexact line search, especially in quasi-Newton methods, first published by Philip Wolfe in 1969.[1][2]

In these methods the idea is to find for some smooth . Each step often involves approximately solving the subproblem where is the current best guess, is a search direction, and is the step length.

The inexact line searches provide an efficient way of computing an acceptable step length that reduces the objective function 'sufficiently', rather than minimizing the objective function over exactly. A line search algorithm can use Wolfe conditions as a requirement for any guessed , before finding a new search direction .

Armijo rule and curvature

edit

A step length   is said to satisfy the Wolfe conditions, restricted to the direction  , if the following two inequalities hold:

  1.  
  2.  

with  . (In examining condition (ii), recall that to ensure that   is a descent direction, we have  , as in the case of gradient descent, where  , or Newton–Raphson, where   with   positive definite.)

  is usually chosen to be quite small while   is much larger; Nocedal and Wright give example values of   and   for Newton or quasi-Newton methods and   for the nonlinear conjugate gradient method.[3] Inequality i) is known as the Armijo rule[4] and ii) as the curvature condition; i) ensures that the step length   decreases   'sufficiently', and ii) ensures that the slope has been reduced sufficiently. Conditions i) and ii) can be interpreted as respectively providing an upper and lower bound on the admissible step length values.

Strong Wolfe condition on curvature

edit

Denote a univariate function   restricted to the direction   as  . The Wolfe conditions can result in a value for the step length that is not close to a minimizer of  . If we modify the curvature condition to the following,

  1.  

then i) and iii) together form the so-called strong Wolfe conditions, and force   to lie close to a critical point of  .

Rationale

edit

The principal reason for imposing the Wolfe conditions in an optimization algorithm where   is to ensure convergence of the gradient to zero. In particular, if the cosine of the angle between   and the gradient,   is bounded away from zero and the i) and ii) conditions hold, then  .

An additional motivation, in the case of a quasi-Newton method, is that if  , where the matrix   is updated by the BFGS or DFP formula, then if   is positive definite ii) implies   is also positive definite.

Comments

edit

Wolfe's conditions are more complicated than Armijo's condition, and a gradient descent algorithm based on Armijo's condition has a better theoretical guarantee than one based on Wolfe conditions (see the sections on "Upper bound for learning rates" and "Theoretical guarantee" in the Backtracking line search article).

See also

edit

References

edit
  1. ^ Wolfe, P. (1969). "Convergence Conditions for Ascent Methods". SIAM Review. 11 (2): 226–235. doi:10.1137/1011036. JSTOR 2028111.
  2. ^ Wolfe, P. (1971). "Convergence Conditions for Ascent Methods. II: Some Corrections". SIAM Review. 13 (2): 185–188. doi:10.1137/1013035. JSTOR 2028821.
  3. ^ Nocedal, Jorge; Wright, Stephen (1999). Numerical Optimization. p. 38.
  4. ^ Armijo, Larry (1966). "Minimization of functions having Lipschitz continuous first partial derivatives". Pacific J. Math. 16 (1): 1–3. doi:10.2140/pjm.1966.16.1.

Further reading

edit