Abstract:The mixed volume counts the roots of generic sparse polynomial systems. Mixed cells are used to provide starting systems for homotopy algorithms that can find all those roots, and track no unnecessary path. Up to now, algorithms for that task were of enumerative type, with no general non- exponential complexity bound. A geometric algorithm is introduced in this paper. Its complexity is bounded in the average and probability-one settings in terms of some geometric invariants: quermassintegrals associated to the tuple of convex hulls of the support of each polynomial. Besides the complexity bounds, numerical results are reported. Those are consistent with an output- sensitive running time for each benchmark family where data is available. For some of those families, an asymptotic running time gain over the best code available at this time was noticed.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the problem of computing the mixed volume and all mixed cells. Specifically, the goal of the paper is to develop a new geometric algorithm to efficiently compute these values, and the complexity of this algorithm can be bounded by some geometric invariants (such as quermassintegrals related to the convex hull of the support of each polynomial) in the average case and with probability 1.
### Background and Problem Description
The mixed volume is used to calculate the number of roots of a sparse polynomial system. Mixed cells are used to provide the starting system for homotopy algorithms, which can find all roots without tracing unnecessary paths. Before this, most algorithms for this task were of the enumeration type and did not have general non - exponential complexity bounds.
### Main Contributions of the Paper
This paper introduces a new geometric algorithm whose complexity can be bounded by geometric invariants in the average case and with probability 1. These invariants include quermassintegrals related to the convex hull of the support of each polynomial. In addition to the complexity bounds, the paper also reports numerical results, which show that the algorithm has output - sensitive running time in some benchmark families and that, for some families, the asymptotic running time of the algorithm is better than that of the best existing code.
### Key Formulas
The definition of the mixed volume \( V(A_1,\ldots,A_n) \) is as follows:
\[ V(A_1,\ldots,A_n)=\frac{1}{n!}\left.\frac{\partial^n}{\partial t_1\cdots\partial t_n}\text{Vol}(t_1A_1+\cdots +t_nA_n)\right|_{t = 0} \]
where \( t_1,\ldots,t_n\geq0 \), and the derivative is evaluated at \( t = 0 \). This generalizes the ordinary volume:
\[ V(A,A,\ldots,A)=\text{Vol}(A) \]
### Main Algorithm and Complexity Analysis
The main algorithm AllMixedCellsFull in the paper is stated on page 21. This algorithm generates all lower mixed cells with a time complexity of \( O(T + T') \) arithmetic operations, where:
\[ T=\left(\sum_{i = 2}^n v_i\right)\left(n^{2n}\sum_{i = 1}^n E_i+\log\sum_{i = 2}^n v_i\right) \]
\[ T'=\left(\max V_i\right)\left(n^{2n}\sum_{i = 1}^n|A_i|+\log\max_{i = 1,\ldots,n}V_i\right) \]
Here \( v_i \) is a random variable that satisfies the following two bounds:
1. With probability 1,
\[ v_i\leq n!V(A_1,\ldots,A_{i - 1},A,B_n,\ldots,B_n) \]
2. If \( \bar{v}_i \) is the average value of \( v_i \), then
\[ \bar{v}_i\leq\frac{n!}{2^{n - i}}V(A_1,\ldots,A_{i - 1},A,B_n,\ldots,B_n) \]
### Experimental Results and Conclusions
The paper provides practical implementation and experimental results to verify the performance of the algorithm. Experiments show that the algorithm performs well in multiple benchmark tests, especially having good scalability in a multi - core parallel environment.
In conclusion, this paper solves the problem of computing the mixed volume and mixed cells by introducing a new geometric algorithm and provides theoretical and experimental support to prove its effectiveness and superiority.