TY - GEN
T1 - Prediction-Correction Dual Ascent for Time-Varying Convex Programs
AU - Simonetto, Andrea
N1 - Publisher Copyright:
© 2018 AACC.
PY - 2018/8/9
Y1 - 2018/8/9
N2 - We develop a prediction-correction dual ascent algorithm that tracks the optimal primal-dual pair of linearly constrained time-varying convex programs. This is especially useful, e.g., for time-varying distributed optimization problems. We prove that our discrete time algorithm has better asymptotical error bound than state-of-the-art methods, which only correct their approximate primal-dual pair without predicting how this changes with time. In numerical simulations, we show that the improvement in accuracy still holds even when computational considerations are taken into account, in almost all cases.
AB - We develop a prediction-correction dual ascent algorithm that tracks the optimal primal-dual pair of linearly constrained time-varying convex programs. This is especially useful, e.g., for time-varying distributed optimization problems. We prove that our discrete time algorithm has better asymptotical error bound than state-of-the-art methods, which only correct their approximate primal-dual pair without predicting how this changes with time. In numerical simulations, we show that the improvement in accuracy still holds even when computational considerations are taken into account, in almost all cases.
U2 - 10.23919/ACC.2018.8431821
DO - 10.23919/ACC.2018.8431821
M3 - Conference contribution
AN - SCOPUS:85052569943
SN - 9781538654286
T3 - Proceedings of the American Control Conference
SP - 4508
EP - 4513
BT - 2018 Annual American Control Conference, ACC 2018
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2018 Annual American Control Conference, ACC 2018
Y2 - 27 June 2018 through 29 June 2018
ER -