Algorithms for 2-Solvable Difference Equations

Our paper “Solving Third Order Linear Difference Equations in Terms of Second Order Equations” gave two algorithms for solving difference equations in terms of lower order equations: an algorithm for absolute factorization, and an algorithm for solving third order equations in terms of second order. Here we improve the efficiency for absolute factorization, and extend the other algorithm to order four.

💡 Research Summary

This paper, “Algorithms for 2-Solvable Difference Equations,” presents significant advancements in solving linear difference equations (recurrence relations) by reducing them to lower-order equations. It builds upon the authors’ prior work, which provided algorithms for solving third-order equations in terms of second-order ones. The dual focus of this paper is enhancing the efficiency of an existing algorithm for “absolute factorization” and extending the reduction framework to handle fourth-order difference operators.

The core problem is finding “2-expressible” solutions for an operator L of order n, meaning solutions that can be expressed using solutions of second-order equations. The authors establish a complete theoretical classification: an irreducible L of order 3 or 4 is 2-solvable if and only if it falls into one of four cases: (1) L is reducible in the standard difference operator ring D; (2) L is not “absolutely irreducible”; (3) the D-module D/DL is isomorphic to a tensor product of two 2-dimensional modules; (4) L is gauge equivalent to a symmetric product of a second-order operator.

The first major contribution is a substantial efficiency improvement for the absolute factorization algorithm (Case 2). The existing method checks for reducibility of the “section operator” L(p) for primes p dividing n. The bottleneck is factoring L(p), which involves searching for right-factors by examining a potentially large set of candidate types (determinants). The key innovation (Theorem 3.6) leverages determinant formulas for symmetric and exterior powers of modules (developed in Section 2). It proves that the determinant of any d-th order right-factor R of L(p) must be gauge equivalent to a specific power of det(L). This provides a powerful necessary condition that filters out a vast number of invalid candidate combinations before the costly polynomial solution computation begins. An explicit example (Example 3.4) demonstrates how this optimization can reduce the number of cases from 1791 to a much smaller set, drastically cutting computation time.

The second major contribution is the development of new algorithms to handle Cases 3 and 4 for fourth-order operators, thereby fully extending the 2-solvable framework to order four. For Case 3 (tensor product), the algorithm computes the exterior square ∧²(L) and looks for its third-order factors that are themselves reducible to order two using the prior third-order algorithm. For Case 4 (gauge equivalence to a symmetric square), the algorithm analyzes the symmetric square LⓈ². Depending on whether its order is 7 or 10, the method deduces the data of a second-order operator L2 and a rational function r such that L is gauge equivalent to L2 Ⓢ (τ - r).

Underpinning these algorithms is a robust mathematical foundation using difference Galois theory and module theory. The implementations of these algorithms are publicly available, validating their practicality. Ultimately, this work expands the toolbox for computer algebra systems, enabling them to find a broader class of closed-form solutions for difference equations beyond the well-known Liouvillian (1-expressible) solutions.