自適應(yīng)擬牛頓子空間跟蹤算法分析
Analysis of adaptive quasi-newton algorithm for eigensubspace tracking
-
摘要: 該文對(duì)自適應(yīng)擬牛頓子空間跟蹤算法進(jìn)行了詳細(xì)分析，證明了其本質(zhì)上類似于冪法，但一般來(lái)說(shuō)暫態(tài)性能劣于冪法。給出了在相鄰數(shù)據(jù)之間只做一次迭代時(shí)特征值、特征向量估計(jì)的誤差，計(jì)算機(jī)仿真結(jié)果證明了該文分析的正確性。
-
關(guān)鍵詞:
- 子空間跟蹤; 擬牛頓法; 特征值; 特征向量
Abstract: In this paper, an adaptive quasi-Newton algorithm for eigensubspace estimation is analysed and proved. This algorithm is similar to power method, but its transient performance is poor than power method. Assuming one iteration to be done between snapshots, the error between estimated and true eigenpairs is obtained. The simulation results show the correct of the analysis in this paper. -
Yang Bin, Projection approximation subspace tracking, IEEE Trans. On SP, 1995, (43)1, 95-107.[2]G. Mathew, V. U. Reddy, Orthogonal eigensubspace estimation using neural networkings, IEEETrans. On SP, 1994, (42)7, 1803-1811.[3]G. Mathew, V. U. Reddy, S. Dasgupta, Adaptive estimation of eigensubspace, IEEE Trans. on SP, 1995, (43)2,401-411.[4]Z.J. Kang, et al, An adaptive quasi-Newton algorithm for eigensubspace estimation, IEEE. Trans.on SP, 2000, (48)12, 3328-3333.[5]P. Strobach, Square-root QR inverse iteration for tracking the minor subspace, IEEE Trans. on SP, 2000, (48)11, 2994-2999.[6]F. Chatelin, Spectral Approximation of Linear Operators, New York, Academic Press Inc, 1983,1-84. -
計(jì)量
- 文章訪問數(shù): 2296
- HTML全文瀏覽量: 118
- PDF下載量: 723
- 被引次數(shù): 0
下載:
