Abstract:We consider the numerical irreducible decomposition of a positive dimensional solution set of a polynomial system into irreducible factors. Path tracking techniques computing loops around singularities connect points on the same irreducible components. The computation of a linear trace for each factor certifies the decomposition. This factorization method exhibits a good practical performance on solution sets of relative high degrees.
Using the same concepts of monodromy and linear trace, we present a new monodromy breakup algorithm. It shows a better performance than the old method which requires construction of permutations of witness points in order to break up the solution set. In contrast, the new algorithm assumes a finer approach allowing us to avoid tracking unnecessary homotopy paths.
As we designed the serial algorithm keeping in mind distributed computing, an additional advantage is that its parallel version can be easily built. Synchronization issues resulted in a performance loss of the straightforward parallel version of the old algorithm. Our parallel implementation of the new approach bypasses these issues, therefore, exhibiting a better performance, especially on solution sets of larger degree.
Distributed, Parallel, and Cluster Computing,Algebraic Geometry,Numerical Analysis
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: how to effectively decompose the positive - dimensional solution set of a polynomial system into irreducible components. Specifically, the author proposes a new parallel monodromy breakup algorithm, aiming to improve the performance of existing methods, especially when dealing with solution sets of larger degrees.
### Problem Background
Polynomial systems are becoming more and more common in the fields of science and engineering, and it is particularly important to simplify these systems. One way to understand the solution set of a polynomial is to factorize it, which is equivalent to finding the irreducible decomposition of its solution set. This is a core problem in numerical algebraic geometry.
### Limitations of Existing Methods
The traditional monodromy breakup algorithm decomposes the solution set by constructing permutations, which requires tracking unnecessary homotopy paths, resulting in low computational efficiency. In addition, synchronization problems will occur when directly parallelizing the old algorithm, thereby reducing the overall performance.
### Advantages of the New Algorithm
The newly proposed monodromy breakup algorithm has the following advantages:
1. **Avoid Unnecessary Path Tracking**: The new algorithm adopts a more refined method and avoids tracking unnecessary homotopy paths.
2. **Better Parallel Performance**: The new algorithm is designed with distributed computing in mind, so a parallel version can be easily constructed, and it bypasses the synchronization problems existing in the old algorithm, and is especially suitable for solution sets of larger degrees.
### Mathematical Description
For a polynomial system \( f(x)=(f_1(x),\ldots,f_m(x)) = 0\), where \( f_i\in\mathbb{C}[x]=\mathbb{C}[x_1,\ldots,x_n]\), its solution set \( V\) can be organized into pure - dimensional solution sets \( V = [V_0,V_1,\ldots,V_n]\), where \(\dim(V_k)=k\). A numerical representation of a pure - dimensional solution set \( V_k\) is a witness set, which consists of the following parts:
1. Polynomials \( f_i\) (\(i = 1,\ldots,m\));
2. \(k\) linear equations \( L(x)=(L_1(x),\ldots,L_k(x)) = 0\), describing \(k\) general hyperplane slices;
3. A list \(W\) of \(\deg(V_k)\) solutions of the solution system \( f(x)=L(x) = 0\).
By choosing general coefficients, the solution sets defined by these hyperplanes contain exactly \(\deg(V_k)\) isolated regular solutions.
### Working Principle of the New Algorithm
The new algorithm utilizes the concepts of monodromy and linear trace, and defines loops around singularities by constructing homotopy paths. The specific steps are as follows:
1. Initialize the witness set \(W_L\).
2. Generate two slices \(L'\) and \(L''\) parallel to \(L\).
3. Track \(d\) paths to the new slices.
4. Generate a new random \(\gamma\), and continue to track the paths.
5. Calculate the permutation and update the decomposition result.
6. Use the linear trace test to verify the decomposition result.
In this way, the new algorithm can complete the decomposition within a fewer number of loop times and shows better performance in a parallel computing environment.
### Experimental Results
The experimental results show that the new algorithm is more stable and efficient when dealing with solution sets of larger degrees. Compared with the old algorithm, the performance of the new algorithm on a single - processor is also improved, and it shows better parallel performance in a multi - processor environment.
### Conclusion
The newly proposed monodromy breakup algorithm is superior to the old algorithm in performance, especially when dealing with solution sets of larger degrees. This makes it more feasible and efficient to decompose the solution set of a polynomial system.