[Paper Review] On Second Solutions to Second-Order Difference Equations

William C. Parke, L. C. Maximon|arXiv (Cornell University)|Jan 18, 2016

Nonlinear Waves and Solitons10 references3 citations

TL;DR

This paper develops explicit, closed-form expressions for second independent solutions to linear second-order difference equations using d'Alembert's reduction of order and an iterative method, with a focus on confluent hypergeometric-type recurrences. It derives polynomial second solutions when standard methods fail due to degeneracy, confirming equivalence to known solutions via the extended Cauchy-integral method and series matching.

ABSTRACT

We investigate and derive second solutions to linear homogeneous second-order difference equations using a variety of methods, in each case going beyond the purely formal solution and giving explicit expressions for the second solution. We present a new implementation of d'Alembert's reduction of order method, applying it to linear second-order recursion equations. Further, we introduce an iterative method to obtain a general solution, giving two linearly independent polynomial solutions to the recurrence relation. In the case of a particular confluent hypergeometric function for which the standard second solution is not independent of the first, i.e. the solutions are degenerate, we use the corresponding differential equation and apply the extended Cauchy-integral method to find a polynomial second solution for the difference equation. We show that the standard d'Alembert method also generates this polynomial solution.

Motivation & Objective

• To provide explicit, non-formal expressions for second independent solutions to linear second-order difference equations when one solution is known.
• To address the case where the standard second solution becomes dependent (degenerate) for confluent hypergeometric functions with negative integer first parameters.
• To develop a new implementation of d'Alembert's reduction of order for recurrence relations.
• To derive polynomial second solutions via iterative methods and verify consistency with known solutions using the extended Cauchy-integral method.
• To confirm equivalence between the derived solution and the standard DLMF solution for confluent hypergeometric equations with integer parameters.

Proposed method

• Applies d'Alembert's reduction of order to difference equations by assuming a second solution of the form $ y_n^{(2)} = w_n f_n $, where $ f_n $ is the known first solution.
• Derives a first-order recurrence for $ u_n = \Delta w_n $, leading to a closed-form expression for $ w_n $ via summation of products involving coefficients $ a_n, c_n $ and the first solution $ f_n $.
• Uses an iterative method to generate two linearly independent polynomial solutions for the recurrence relation, particularly effective when the first solution is a polynomial.
• Applies the extended Cauchy-integral method to the corresponding differential equation to derive a second solution, which is then adapted to the difference equation framework.
• Compares the derived solution with the standard DLMF solution for $ {}_1F_1(-N; n+1; x) $, matching logarithmic and power-series terms to confirm identity.
• Employs series expansion and Gauss's hypergeometric identity to verify term-by-term agreement between the new solution and the known solution, establishing $ \Psi_{PM} = \Psi_{DL} $.

Experimental results

Research questions

• RQ1 How can d'Alembert's reduction of order be systematically adapted to second-order linear difference equations to yield explicit second solutions?
• RQ2 What is the form of the second solution when the standard second solution degenerates (e.g., for confluent hypergeometric functions with negative integer parameters)?
• RQ3 Can an iterative method generate two linearly independent polynomial solutions for a given second-order recurrence relation?
• RQ4 How does the extended Cauchy-integral method applied to the differential equation counterpart yield a valid second solution for the difference equation?
• RQ5 To what extent does the derived second solution match the known solution in the Digital Library of Mathematical Functions (DLMF)?

Key findings

• The paper derives a closed-form expression for the second solution of a second-order linear difference equation: $ y_n^{(2)} = f_1 f_n \sum_{k=1}^{n-1} \frac{1}{f_k f_{k+1}} \prod_{l=0}^{k-1} \frac{c_l}{a_l} $, valid when the first solution $ f_n $ is known.
• For the confluent hypergeometric recurrence with $ a = -N $, $ b = n+1 $, the method produces a polynomial second solution that matches the DLMF solution $ \Psi_{DL}(N,n,x) $, confirming $ \Psi_{PM}(N,n,x) = \Psi_{DL}(N,n,x) $.
• The derived solution contains logarithmic terms matching the DLMF solution, including $ \ln x $ and Euler-Mascheroni constant $ \gamma $, verified via series expansion.
• The polynomial part of the solution is explicitly given as $ P(N,n,x) $, involving finite sums of factorials and powers of $ x $, and matches the DLMF's first term in the series expansion.
• The method confirms that even in degenerate cases (e.g., $ a = -N $), a second independent polynomial solution exists and can be constructed via reduction of order.
• The equivalence between the new solution and the DLMF solution is established by matching both the logarithmic and power-series components term-by-term, with $ c_1 = 1 $, $ c_2 = 0 $ in the solution basis.