Constant Stepsize Algorithms
Vivek S. Borkar · Texts and readings in mathematics · 2008
In many practical circumstances, it is more convenient to use a small constant stepsize a ( n ) = a ∈ (0, 1) rather than the decreasing stepsize considered thus far. One such situation is when the algorithm is ‘hard-wired’ and decreasing stepsize may mean additional overheads. Another important scenario is when the algorithm is expected to operate in a slowly varying environment (e.g., in tracking applications) where it is important that the timescale of the algorithm remain reasonably faster than the timescale on which the environment is changing, for otherwise it would never adapt.