Fig. 3
From: Estimating the optimal linear combination of predictors using spherically constrained optimization
Exploratory moves on the surface of unit sphere by SCOR. a shows an initial point \((\beta _1,\beta _2,\beta _3)\) (green) on the surface of the unit sphere and a point explored \((\beta _1+t_2,\beta _2+s,\beta _3+t_2)\) (red) after making exploratory move with step-size s for the \(2^{nd}\) coordinate, where \(t_2\) denotes the adjustment step-size. b shows exploratory moves in a typical iteration of SCOR: starting from the current solution point \((0.289,-0.816,0.5)\) (green), the points explored after making exploratory moves with step-size \(s=0.03\) (blue) and \(s=0.06\) (red)