A convergence result for nonautonomous subgradient evolution equations and its application to the steepest descent exponential penalty trajectory in linear programming (Q5957477): Difference between revisions

From authors' abstract: The authors present a new result on the asymptotic behavior of nonautonomous subgradient evolution equations of the form \[ \dot u(t)\in - \partial \phi_{t}(u(t)), \] where \(\{\phi_{t}: t\geq 0\}\) is a family of closed proper convex functions. The result is used to study the flow generated by the family \[ \phi_{t}(x)=f(x,r(t)), \] where \(f(x,r):=c^{T}x+r\sum\exp [\frac {A_{i}x-b_{i}}{r}]\) is the exponential penalty approximation to the linear program min \(\{c^{T}x: Ax\leq b\},\) and \(r(t)\) is a positive function tending to \(0\) when \(t\to\infty\). The authors prove that the trajectory \(u(t)\) converges to an optimal solution \(u^{\infty}\) to the linear program and give conditions for the convergence of an associated dual trajectory \(\mu(t)\) toward an optimal solution to the dual program.