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

P

bound by reducing a polynomial number of bases with di erent bn 1 vectors. However, for small dimensions it might be possible to improve the bound, even though any such advantage will disappear as n grows. In cases where the subset sum problem (Equation 1) to be solved is known to have ei small (as occurs in some knapsack cryptosystems, such as the Chor-Rivest one 4], which has still not been broken), it is possible to again improve on the results of 11] by our approach. For example, if we know that + X

n

we can replace the vector bn 1 in the basis of L by +

i=1

ei= n;

b00 1= (;;:::;; Ns); n +

and then the lattice L will contain a vector of length n (1? ), and our analysis shows that in this case it then becomes possible to solve most problems with even smaller weights ai. However, it appears that there are choices for parameters in the Chor-Rivest knapsack that would resist even this attack. When we consider the L1 or sup-norm, then we nd that the vector^0 has norm 1/2. Since there are at most 2n non-zero e 0 of norm 1=2, we can solve almost all subset sum problems of any vectors in L density< 1 if we have a lattice oracle for the sup-norm. Formally, we may make the following proposition: Proposition. 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 be arbitrary, and let s= n eiai. If the density d< 1, then the subset sum problem de ned by a;:::; an i and s may\almost always" be solved in polynomial time, given the existence of a sup-norm lattice oracle. 1 P 1=1 1

q

k(x;:::; xn)k1= maxn jxj j; j 1 1

The general sup-norm shortest vector problem is known to be NP-complete 6]; the complexity of the square-norm shortest vector proble

m is an open problem. That a sup-norm lattice oracle yields a better density bound than a square-norm lattice oracle suggests that the shortest vector problem for the sup-norm might be harder than for the square-norm. Sections 3 and 4 presented theoretical results that assume the availability of an e cient method for nding the shortest non-zero vector in a lattice. When one uses

上一页第4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