We announce two breakthrough results concerning important questions in the Theory of Computational Complexity. In this expository paper, a systematic and comprehensive geometric characterization of the Subset Sum Problem is presented. We show the existence of a universal geometric structure, comprised of a family of non-decreasing paths in the Cartesian plane, that captures any instance of the problem of size $n$. Inspired by the geometric structure, we provide an unconditional, deterministic and polynomial time algorithm, albeit with fairly high complexity, thereby showing that $\mathcal{P} = \mathcal{NP}$. Furthermore, our algorithm also outputs the number of solutions to the problem in polynomial time, thus leading to $\mathcal{FP} = \mathcal{\#P}$. As a bonus, one important consequence of our results, out of many, is that the quantum-polynomial class $\mathcal{BQP} \subseteq \mathcal{P}$. Not only this, but we show that when multiple solutions exist, they can be placed in certain equivalence classes based on geometric attributes, and be compactly represented by a polynomial sized directed acyclic graph. We show that the Subset Sum Problem has two aspects, namely a combinatorial aspect and a relational aspect, and that it is the latter which is the primary determiner of complexity. We reveal a surprising connection between the size of the elements and their number, and the precise way in which they affect the complexity. In particular, we show that for all instances of the Subset Sum Problem, the complexity is independent of the size of elements, once the difference between consecutive elements exceeds $\lceil{7\log{}n}\rceil$ bits in size. We provide some numerical examples to illustrate the algorithm, and also show how it can be used to estimate some difficult combinatorial quantities such as the number of restricted partitions.
翻译:我们宣布两项关于计算复杂性理论重要问题的突破性成果。在这篇阐述性论文中,我们系统而全面地给出了子集和问题的几何刻画。我们证明存在一种通用几何结构——由笛卡尔平面上一族非递减路径构成——能够捕捉任意规模为$n$的问题实例。受此几何结构启发,我们提出了一种无条件的确定性多项式时间算法(尽管复杂度较高),从而证明$\mathcal{P} = \mathcal{NP}$。此外,该算法还能在多项式时间内输出问题的解的数量,进而推导出$\mathcal{FP} = \mathcal{\#P}$。作为额外成果,我们的研究还得出一个重要推论(众多推论之一):量子多项式类$\mathcal{BQP} \subseteq \mathcal{P}$。不仅如此,我们证明当存在多个解时,这些解可根据几何属性归入特定等价类,并能通过多项式规模的有向无环图进行紧凑表示。我们揭示子集和问题具有组合性与关系性两个层面,而后者才是复杂性的主要决定因素。我们发现了元素大小与数量之间令人惊讶的关联性,以及它们影响复杂性的精确方式。特别地,我们证明对于所有子集和问题实例,当相邻元素差值超过$\lceil{7\log{}n}\rceil$比特时,其复杂度与元素大小无关。我们提供了若干数值示例以说明算法流程,并展示如何利用该算法估算受限划分数等困难组合量。