極大值函數(shù)方程的梯度類算法研究
發(fā)布時間:2020-08-13 02:25
【摘要】:本文研究了一類特殊極大值函數(shù)非光滑方程問題,首先利用絕對值函數(shù)的光滑函數(shù)與極大值函數(shù)的光滑函數(shù)對提出的非光滑問題進行磨光處理,隨后給出了解決此類問題的光滑化梯度類算法,并且對所給出的算法進行了全局收斂性分析。本文具體結構為:第一章在基于絕對值函數(shù)與極大值函數(shù)的光滑函數(shù)的情況下,研究了一類特殊極大值函數(shù)方程問題的光滑譜共軛梯度法,在一般的假設條件下,給出了方法的全局收斂性分析,最后的數(shù)值結果表明算法的有效性。第二章利用絕對值函數(shù)和極大值函數(shù)的光滑函數(shù)對提出的非光滑方程問題進行轉化,根據(jù)轉化后的問題,給出了光滑Barzilai-Borwein調比共軛梯度法并且給出了全局收斂性分析和相應的數(shù)值實驗。第三章提出了一種光滑保守DPRP共軛梯度法來對特殊極大值函數(shù)方程問題進行了求解,然后給出了方法的全局收斂性分析,最后的數(shù)值結果表明了算法的有效性。
【學位授予單位】:青島大學
【學位級別】:碩士
【學位授予年份】:2018
【分類號】:O241.8
【圖文】:
圖 1.1 目標函數(shù)minΦ x ,p 的變化曲線2 考慮問題 1-(1),其中 A與 b 由 Matlab 程序隨機生成四維矩陣,元之間的整數(shù)。隨機矩陣 A與 向量b 分別為(8* (4))(8* (1,4))'.A round randb round rand 體數(shù)值實驗結果見表 1.2 和圖 1.2.
0x*xΦ .0000,0.0000,0.0000, 0.0000]T[0.8927,-1.1464,0.2584,0.0942]T1.86.8147,0.9058,0.1270,0.9134]T[0.8927,-1.1464,0.2584,0.0942]T2.74.6324,0.0975,0.2785,0.5469]T[0.8927,-1.1464,0.2584,0.0942]T2.94.9575,0.9649,0.1576,0.9706]T[0.8927,-1.1464,0.2584,0.0942]T3.15.9572,0.4854,0.8003,0.1419]T[0.8927,-1.1464,0.2584,0.0942]T1.26.6557,0.0357,0.8491,0.9340]T[0.8927,-1.1464,0.2584,0.0942]T2.50.2769,0.0462,0.0971,0.8235]T[0.8927,-1.1464,0.2584,0.0942]T1.85
目標函數(shù)minkpx隨迭代次數(shù)k的變化曲線
本文編號:2791379
【學位授予單位】:青島大學
【學位級別】:碩士
【學位授予年份】:2018
【分類號】:O241.8
【圖文】:
圖 1.1 目標函數(shù)minΦ x ,p 的變化曲線2 考慮問題 1-(1),其中 A與 b 由 Matlab 程序隨機生成四維矩陣,元之間的整數(shù)。隨機矩陣 A與 向量b 分別為(8* (4))(8* (1,4))'.A round randb round rand 體數(shù)值實驗結果見表 1.2 和圖 1.2.
0x*xΦ .0000,0.0000,0.0000, 0.0000]T[0.8927,-1.1464,0.2584,0.0942]T1.86.8147,0.9058,0.1270,0.9134]T[0.8927,-1.1464,0.2584,0.0942]T2.74.6324,0.0975,0.2785,0.5469]T[0.8927,-1.1464,0.2584,0.0942]T2.94.9575,0.9649,0.1576,0.9706]T[0.8927,-1.1464,0.2584,0.0942]T3.15.9572,0.4854,0.8003,0.1419]T[0.8927,-1.1464,0.2584,0.0942]T1.26.6557,0.0357,0.8491,0.9340]T[0.8927,-1.1464,0.2584,0.0942]T2.50.2769,0.0462,0.0971,0.8235]T[0.8927,-1.1464,0.2584,0.0942]T1.85
目標函數(shù)minkpx隨迭代次數(shù)k的變化曲線
【參考文獻】
相關期刊論文 前6條
1 DAI YuHong;KOU CaiXia;;A Barzilai-Borwein conjugate gradient method[J];Science China(Mathematics);2016年08期
2 李敏;屈愛平;;一種充分下降的PRP共軛梯度法的全局收斂性[J];高等學校計算數(shù)學學報;2013年02期
3 張麗麗;李建宇;李興斯;;極大值函數(shù)的一類光滑逼近函數(shù)的性質研究[J];數(shù)學的實踐與認識;2008年24期
4 喻高航;關履泰;;具有充分下降性的修正PRP算法及其收斂性[J];中山大學學報(自然科學版);2006年04期
5 袁玉波,嚴杰,徐成賢;多項式光滑的支撐向量機[J];計算機學報;2005年01期
6 李興斯;非線性極大極小問題的一個有效解法[J];科學通報;1991年19期
本文編號:2791379
本文鏈接:http://www.sikaile.net/kejilunwen/sousuoyinqinglunwen/2791379.html
最近更新
教材專著