[Paper Review] Quasi-Exactly-Solvable Differential Equations

Alexander V. Turbiner|ArXiv.org|Sep 12, 1994

Advanced Topics in Algebra20 references21 citations

TL;DR

This paper establishes a comprehensive classification of quasi-exactly-solvable differential and finite-difference operators by showing they must arise as polynomial elements in the universal enveloping algebra of specific Lie algebras. The key result is that such operators possess finite-dimensional invariant subspaces spanned by polynomials, with classification achieved via $sl_2(\mathbb{R})$, $sl_2(\mathbb{R})_q$, $osp(2,2)$, and $gl_2(\mathbb{R})_K$ in one-dimensional and matrix settings.

ABSTRACT

A general classification of linear differential and finite-difference operators possessing a finite-dimensional invariant subspace with a polynomial basis is given. The main result is that any operator with the above property must have a representation as a polynomial element of the universal enveloping algebra of the algebra of differential (difference) operators in finite-dimensional representation. In one-dimensional case a classification is given by algebras $sl_2({\bold R})$ (for differential operators in ${\bold R}$) and $sl_2({\bold R})_q$ (for finite-difference operators in ${\bold R}$), $osp(2,2)$ (operators in one real and one Grassmann variable, or equivalently, $2 imes 2$ matrix operators in ${\bold R}$) and $gl_2 ({\bold R})_K$ ( for the operators containing the differential operators and the parity operator). A classification of linear operators possessing infinitely many finite-dimensional invariant subspaces with a basis in polynomials is presented.

Motivation & Objective

• To classify linear differential and finite-difference operators that possess a finite-dimensional invariant subspace with a polynomial basis.
• To establish a general framework linking Lie algebra representations to quasi-exactly-solvable operators in one-dimensional and matrix settings.
• To extend the concept of quasi-exact solvability to include operators with parity and finite-difference structures.
• To provide a complete classification of such operators using the universal enveloping algebra of relevant Lie algebras.
• To clarify the distinction between quasi-exactly-solvable and exactly-solvable operators through grading and Casimir invariants.

Proposed method

• Represent operators as polynomial elements in the universal enveloping algebra of Lie algebras such as $sl_2(\mathbb{R})$, $sl_2(\mathbb{R})_q$, $osp(2,2)$, and $gl_2(\mathbb{R})_K$.
• Use finite-dimensional representations of these Lie algebras to construct invariant subspaces spanned by polynomials.
• Introduce a grading on generators to classify operators by degree and identify exactly-solvable cases (no positive grading terms).
• Apply gauge transformations and variable changes to standardize polynomial spaces to $\mathcal{F}_n(x) = \langle 1, x, \dots, x^n \rangle$.
• Derive explicit forms of second-order operators via quadratic polynomials in generators, with parameters constrained by Casimir invariants.
• Use the condition of commuting with the parity operator $K$ to identify operators with definite-parity eigenfunctions.

Experimental results

Research questions

• RQ1 Which linear differential and finite-difference operators possess a finite-dimensional invariant subspace with a polynomial basis?
• RQ2 How can such operators be systematically classified using Lie algebra representations?
• RQ3 What distinguishes quasi-exactly-solvable operators from exactly-solvable ones in terms of algebraic structure and grading?
• RQ4 How do the inclusion of parity operators and finite-difference structures extend the framework of quasi-exact solvability?
• RQ5 What role do Casimir operators play in characterizing the finite-dimensional representations of the underlying Lie algebras?

Key findings

• Any linear differential or finite-difference operator with a finite-dimensional invariant subspace of polynomials must be representable as a polynomial element in the universal enveloping algebra of a Lie algebra.
• In one dimension, the relevant Lie algebras are $sl_2(\mathbb{R})$ for differential operators, $sl_2(\mathbb{R})_q$ for finite-difference operators, $osp(2,2)$ for matrix or Grassmann-variable operators, and $gl_2(\mathbb{R})_K$ for operators including the parity operator.
• The representation space of dimension $n+1$ is preserved for generic $\nu$, with Casimir invariants $C_1 = n + \nu$ and $C_2 = \frac{1}{4}[(n+\nu)(n+\nu+2) - \nu^2]$.
• A general second-order quasi-exactly-solvable operator has 18 free parameters, while the subset with definite-parity eigenfunctions has 10 parameters.
• Exactly-solvable operators are characterized by the absence of positive-grading terms in the polynomial expression, reducing the number of free parameters to 12 for second-order operators.
• The eigenvalue problem for a quasi-exactly-solvable operator of order $k$ has exactly $n+1$ polynomial eigenfunctions of degree $\leq n$ if and only if the operator is quasi-exactly-solvable, and infinitely many if and only if it is exactly-solvable.