文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Improved low-density subset sum algorithms
侵权投诉

Improved low-density subset sum algorithms

The general subset sum problem is NP-complete. However, there are two algorithms, one due to Brickell and the other to Lagarias and Odlyzko, which in polynomial time solve almost all subset sum problems of sufficiently low density. Both methods rely on bas

section derives the 0:6463::: bound using simpler techniques due to Frieze 7]. Our presentation di ers from that of 7] in a few technical details. Let A be a positive integer and let a;:::; an be random integers with 0< ai A for 1 i n. Let e= (e;:::; en) 2 f0; 1gn, e 6= (0; 0;:::; 0) depending only on n, be xed and let 1 1

s=

X

n

i=1

eiai;

t=

X

n

i=1

ai:

We may assume that s t=n, since if s< t=n any ai t=n cannot be in the subset, and may be removed from consideration. Similarly, s (1? (1=n))t, otherwise any ai (1? (1=n))t may be removed from consideration. Thus, 1 t s n? 1 t: (3) We recall the Lagarias-Odlyzko attack on low-density subset sum problems. De ne the vectors b1;:::; bn 1 as follows: +

n

n

. . . bn= (0; 0;:::; 1; Nan); bn 1= (0; 0;:::; 0; Ns); +

b1= (1; 0;:::; 0; Na ); b2= (0; 1;:::; 0; Na ); 1 2

where N is a positive integer which will be chosen later. Let L be the lattice spanned by the vectors b1;:::; bn 1 (i.e. L= f n zibi: zi 2 Z for 1 i n+ 1g). i Notice that the solution vector^= (e;:::; en; 0) is in L. Following the proof in e 7] we are interested in vectors x= (x; x;:::; xn ) which satisfy:^ P++1=1 1 1 2+1

kxk k^k;^ e x 2 L;^ x 62 f0;^;?^g:^ e e We may assume that X

(4)

n

i=1

ei

1 2

n; P=1 1 2 1 2

(5)

(i.e. the subset contains at most one-half of the ai's). If n ei> n, we may replace s i 0= (0;:::; 0; N (t? s)), and^ by^0= (1? e; 1? e;:::; 1? en; 0). by t? s, bn 1 by bn 1 e e ++

上一页第11页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email