Sharpness, Restart and Acceleration
نویسندگان
چکیده
The Łojasievicz inequality shows that sharpness bounds on the minimum of convex optimization problems hold almost generically. Here, we show that sharpness directly controls the performance of restart schemes. The constants quantifying sharpness are of course unobservable, but we show that optimal restart strategies are fairly robust, and searching for the best scheme only increases the complexity by a logarithmic factor compared to the optimal bound. Overall then, restart schemes generically accelerate accelerated methods.
منابع مشابه
Some new restart vectors for explicitly restarted Arnoldi method
The explicitly restarted Arnoldi method (ERAM) can be used to find some eigenvalues of large and sparse matrices. However, it has been shown that even this method may fail to converge. In this paper, we present two new methods to accelerate the convergence of ERAM algorithm. In these methods, we apply two strategies for the updated initial vector in each restart cycles. The implementation of th...
متن کاملStop and restart of granular clock in a vibrated compartmentalized bidisperse granular system.
This paper studies a bidisperse granular mixture consisting of two species of stainless steel spheres in a vertically vibrated compartmentalized container. The experiments show that with proper vibration acceleration, the granular clock stops when horizontal segregation of the large spheres residing in the far end from the barrier wall occurs. When the segregation is broken, the granular clock ...
متن کاملStudy of Noise Loudness and Sharpness Effects on Cognitive Performance on Male Students of Tehran University of Medical Sciences
Introduction: The noise could affect some aspects of human health, including the cognitive performance. In addition to sound pressure level and exposure time, the psychoacoustic features of noise may cause destructive effects on humans. A few recent studies have been conducted on effect of sound quality on cognitive performance. This study aims to find the noise loudness and sharpness levels as...
متن کاملRenegar’s Condition Number, Sharpness and Compressed Sensing Performance
We show that several quantities controlling compressed sensing performance also directly control algorithmic complexity. We describe linearly convergent restart schemes solving a broad range of compressed sensing problems using first-order methods. The key term controlling convergence measures the sharpness of the optimum and can be interpreted as a condition number, computed as the ratio betwe...
متن کاملComputational Complexity versus Statistical Performance on Sparse Recovery Problems
We show that several classical quantities controlling compressed sensing performance directly match classical parameters controlling algorithmic complexity. We first describe linearly convergent restart schemes on first-order methods solving a broad range of compressed sensing problems, where sharpness at the optimum controls convergence speed. We show that for sparse recovery problems, this sh...
متن کامل