论文标题
半决赛编程的更快的内点方法
A Faster Interior Point Method for Semidefinite Programming
论文作者
论文摘要
半决赛程序(SDP)是一类基本的优化问题,具有近似算法,量子复杂性,健壮的学习,算法圆形和对抗性深度学习的最新应用。本文提出了一种更快的内点方法,以求解具有可变尺寸$ n \ times n $和$ m $约束的通用SDP \ begin \ begin {align*} \ widetilde {o}(\ sqrt {n} {n}矩阵乘法和$ε$的相对精度。在$ m \ geq n $的主要情况下,我们的运行时的表现优于先前最快的SDP求解器,该求解器基于江,李,李,歌曲和Wong [JLSW20]的切割平面方法。 Our algorithm's runtime can be naturally interpreted as follows: $\widetilde{O}(\sqrt{n} \log (1/ε))$ is the number of iterations needed for our interior point method, $mn^2$ is the input size, and $m^ω+ n^ω$ is the time to invert the Hessian and slack matrix in each iteration.这些构成了进一步改善用于解决通用SDP的内点方法的运行时的自然障碍。
Semidefinite programs (SDPs) are a fundamental class of optimization problems with important recent applications in approximation algorithms, quantum complexity, robust learning, algorithmic rounding, and adversarial deep learning. This paper presents a faster interior point method to solve generic SDPs with variable size $n \times n$ and $m$ constraints in time \begin{align*} \widetilde{O}(\sqrt{n}( mn^2 + m^ω+ n^ω) \log(1 / ε) ), \end{align*} where $ω$ is the exponent of matrix multiplication and $ε$ is the relative accuracy. In the predominant case of $m \geq n$, our runtime outperforms that of the previous fastest SDP solver, which is based on the cutting plane method of Jiang, Lee, Song, and Wong [JLSW20]. Our algorithm's runtime can be naturally interpreted as follows: $\widetilde{O}(\sqrt{n} \log (1/ε))$ is the number of iterations needed for our interior point method, $mn^2$ is the input size, and $m^ω+ n^ω$ is the time to invert the Hessian and slack matrix in each iteration. These constitute natural barriers to further improving the runtime of interior point methods for solving generic SDPs.