Large systems of random linear equations with nonnegative solutions: Characterizing the solvable and the unsolvable phase.

Large systems of random linear equations with nonnegative solutions: Characterizing the solvable and the unsolvable phase.
复制标题

具有非负解的大型随机线性方程组:表征可解相和不可解相

DOI:
10.1103/physreve.101.062119
复制
发表时间:
2020
期刊:
Physical review. E
影响因子:
--
通讯作者:
A. Engel
A. Engel
中科院分区:
--
文献类型:
--
作者:
S. Landmann;A. Engel

文献摘要

参考文献

被引文献

相似文献

大型线性方程组在科学中无处不在。通常情况下,例如,当考虑种群动力学或化学网络时,解必须是非负的。最近,研究表明,大型随机线性方程组呈现出从以概率1的概率存在非负解的阶段到通常可能找不到这样的解的阶段的急剧转变。结合Farkas引理和复制法确定了两相分离的临界线。在这里,我们表明,相同的方法仍然是可行的,以表征这两个阶段远离临界。为此,我们解析地确定了系统在不可解阶段的剩余范数和在可解阶段解的稳健性的一个适当的度量。我们的结果与数值模拟结果非常吻合。
Large systems of linear equations are ubiquitous in science. Quite often, e.g., when considering population dynamics or chemical networks, the solutions must be nonnegative. Recently, it has been shown that large systems of random linear equations exhibit a sharp transition from a phase, where a nonnegative solution exists with probability one, to one where typically no such solution may be found. The critical line separating the two phases was determined by combining Farkas' lemma with the replica method. Here we show that the same methods remain viable to characterize the two phases away from criticality. To this end we analytically determine the residual norm of the system in the unsolvable phase and a suitable measure of robustness of solutions in the solvable one. Our results are in very good agreement with numerical simulations.
可以从示例中学习的系统:感知器均匀收敛边界的复制计算。
DOI: 10.1103/physrevlett.71.1772
发表时间: 1993
影响因子: 8.6
作者:
Engel;Van den Broeck C
通讯作者: Van den Broeck C
感知器模型中的复制对称不稳定性
DOI: 10.1088/0305-4470/27/17/033
发表时间: 1994
期刊: Journal of Physics A
影响因子: --
作者:
M. Bouten;B. Derrida
通讯作者: B. Derrida
德格鲁特
DOI: --
发表时间: --
期刊:
影响因子: --
作者:
通讯作者: --
DOI: 10.1103/physrevlett.81.4999
发表时间: 1998-11-30
影响因子: 8.6
作者:
Berg, J;Engel, A
通讯作者: Engel, A
饱和以上吸引子神经网络的学习和检索
DOI: 10.1088/0305-4470/24/3/030
发表时间: 1991
期刊: Journal of Physics A
影响因子: --
作者:
M. Griniasty;H. Gutfreund
通讯作者: H. Gutfreund