HeunG (General Heun Function)
What HeunG is
Section titled “What HeunG is”The general Heun equation is the most general second-order Fuchsian ordinary differential equation with four regular singular points. After a Möbius transformation, these singularities can be placed at
The function commonly called HeunG in computer algebra systems is the normalized Frobenius solution around corresponding to the exponent :
- it is analytic at ,
- it satisfies ,
- it depends on an accessory parameter that controls global analytic behavior and typically becomes an eigenvalue in boundary value problems.
1. Canonical general Heun equation and parameters
Section titled “1. Canonical general Heun equation and parameters”1.1 Canonical form
Section titled “1.1 Canonical form”The canonical (general) Heun equation is commonly written as
with complex parameters and with
The parameters are not independent: Fuchsian balance at infinity implies
1.2 Singularities and local exponents
Section titled “1.2 Singularities and local exponents”The equation is Fuchsian with regular singularities at
A convenient way to summarize local exponents is via the Riemann -symbol:
meaning that the pairs of characteristic exponents are
- at : ,
- at : ,
- at : ,
- at : .
1.3 What do and mean?
Section titled “1.3 What do aaa and qqq mean?”- locates the third finite singular point (in addition to and ). It is often called the singularity parameter.
- is the accessory parameter. Unlike (which fix exponent data), is not determined by local exponents; in applications it frequently plays the role of an eigenvalue fixed by global boundary/regularity conditions.
2. Definition of the general Heun function HeunG
Section titled “2. Definition of the general Heun function HeunG”2.1 Local (Frobenius) solution at the origin
Section titled “2.1 Local (Frobenius) solution at the origin”Among the two local Frobenius solutions about , the one corresponding to exponent is analytic at and can be normalized by setting . This normalized local solution is what most CAS call the general Heun function:
Definition (local):
HeunG(a,q,α,β,γ,δ;z)is the solution of the canonical Heun equation that is analytic at and satisfies
In the Ronveaux (1995) convention, this same object is often denoted
with understood implicitly through .
2.2 The second local solution at
Section titled “2.2 The second local solution at z=0z=0z=0”Generically (when ), the second local solution behaves like
When , the exponent difference is an integer and the second local solution may involve a term (“logarithmic case”). This is one of the main sources of special-function subtlety for Heun equations.
2.3 Quickstart in CAS (Maple and Wolfram Language)
Section titled “2.3 Quickstart in CAS (Maple and Wolfram Language)”Most researchers first use HeunG through a CAS. The two most common interfaces are:
Maple
HeunG(a, q, alpha, beta, gamma, delta, z);HeunGPrime(a, q, alpha, beta, gamma, delta, z); # derivative w.r.t. zWolfram Language / Mathematica
HeunG[a, q, α, β, γ, δ, z]HeunGPrime[a, q, α, β, γ, δ, z] (* derivative w.r.t. z *)Both systems use the same parameter order and the same canonical differential equation in §1, with the same normalization .
A good habit (especially when inheriting conventions from a paper) is to ask the CAS for its own definition:
- Maple:
FunctionAdvisor(definition, HeunG); - Wolfram Language: consult the built-in
HeunGdocumentation page.
3. Power-series expansion: coefficients and recurrence
Section titled “3. Power-series expansion: coefficients and recurrence”3.1 Series ansatz and normalization
Section titled “3.1 Series ansatz and normalization”For the exponent- local solution about , use the series
and impose the standard normalization
3.2 Three-term recurrence (Ronveaux 1995)
Section titled “3.2 Three-term recurrence (Ronveaux 1995)”Substituting the series into the Heun equation yields a three-term recurrence. The starting relation is
For ,
where
and .
Equivalently, the forward form is
with the convention .
3.3 Radius of convergence
Section titled “3.3 Radius of convergence”The power series about converges up to the nearest singularity. Since the finite singularities other than are at and ,
So:
- if , the disk of convergence is typically ,
- if , the disk is typically .
This matters in practice: when is close to or , the local series can have a very small domain of convergence.
4. The second local solution at
Section titled “4. The second local solution at z=0z=0z=0”A standard way to write the second independent local solution at (generic case) is
where is again a local Heun function with shifted parameters.
One explicit parameter shift is
and
So a local basis near is
This is the direct analog of the hypergeometric basis and .
5. Transformations
Section titled “5. Transformations”Heun theory is transformation-heavy. Two families are particularly important:
- Möbius transformations of (permuting ).
- Prefactor transformations of of the form to swap exponent choices.
Together these generate many equivalent local series representations (often summarized as “ series”), but they reduce to the expected eight local solutions: two around each singularity.
5.1 A simple scaling identity
Section titled “5.1 A simple scaling identity”One especially useful identity relates expansions at and :
This can be used to move inside or outside the unit disk to improve convergence regions.
5.2 An example where the accessory parameter transforms nontrivially
Section titled “5.2 An example where the accessory parameter transforms nontrivially”For the change of variables , one finds an identity of the form
The point to internalize is not the specific map, but the phenomenon: transforms in a genuinely non-obvious way, and it encodes global information.
6. Hypergeometric degenerations
Section titled “6. Hypergeometric degenerations”The general Heun equation can degenerate to the Gauss hypergeometric equation in special parameter limits. If denotes the normalized exponent- Frobenius solution at with , then three useful degenerations are:
- If and ,
- If and ,
- If and (equivalently ),
These limits are excellent sanity checks for both symbolic manipulations and numerics.
7. Global solutions and why becomes an eigenvalue
Section titled “7. Global solutions and why qqq becomes an eigenvalue”A local HeunG/ solution is defined near one singularity (e.g. near ). In contrast, the reference uses the term Heun function (denoted ) for a solution that is simultaneously a Frobenius solution about two singularities , i.e. analytic in a region containing both (with branch cuts chosen for single-valuedness).
In many physical problems you impose two endpoint conditions such as:
- regularity at and ,
- ingoing behavior at one singular point and normalizability at another,
- prescribed monodromy around a contour.
These global requirements typically hold only for a discrete set of values. That is why behaves as an eigenvalue.
A practical picture: connection matching
Section titled “A practical picture: connection matching”To enforce “regular at and ” you can:
-
Build a local basis near : and .
-
Build a local basis near (by a transformation or a dedicated local expansion).
-
Evaluate both bases and their derivatives at a matching point in an overlap region.
-
Solve for the connection coefficients and impose that the coefficient of the singular basis element vanishes.
This yields a scalar condition on , i.e. an eigenvalue equation.
8. Heun polynomials and series termination
Section titled “8. Heun polynomials and series termination”A Heun polynomial is a solution that is simultaneously Frobenius at three singularities and has the structured form
where is a polynomial and each is one of the two local exponents at the corresponding singularity.
8.1 The simplest termination mechanism
Section titled “8.1 The simplest termination mechanism”From the recurrence, if
then . If one chooses so that , the series terminates and produces a polynomial of degree (class I in the reference). The admissible values satisfy an algebraic equation of degree .
8.2 Eight classes of Heun polynomials
Section titled “8.2 Eight classes of Heun polynomials”The reference classifies Heun polynomials into eight classes I–VIII, determined by the choices of and corresponding constraints on .
A convenient summary is:
9. Branch cuts and multi-valuedness
Section titled “9. Branch cuts and multi-valuedness”Even though the local series defines an analytic function in its convergence disk, analytic continuation around singularities produces monodromy, so the global object is generally multi-valued.
9.1 What software does
Section titled “9.1 What software does”Computer algebra systems choose a principal branch, which is implemented by placing branch cuts that connect singular points to infinity.
In Wolfram Language, the principal branch of HeunG[a,q,α,β,γ,δ,z] has two branch cut discontinuities in the complex -plane: one running from to infinity and one running from to infinity in the direction of .
9.2 Researcher workflow
Section titled “9.2 Researcher workflow”- If you are doing numerics, always record:
- which branch you intend (path of continuation),
- where you expect discontinuities,
- how you validate (e.g. by checking the ODE numerically).
- If you are doing spectral/eigenvalue problems, make sure the boundary/regularity conditions are imposed on a consistent branch.
10. How to reduce a given ODE to HeunG form
Section titled “10. How to reduce a given ODE to HeunG form”Suppose you start with a second-order linear ODE
and you suspect it has four regular singular points.
A robust reduction procedure is:
Step 1: map the singular points to
Section titled “Step 1: map the singular points to {0,1,a,∞}\{0,1,a,\infty\}{0,1,a,∞}”If the finite regular singular points are at (distinct), choose a Möbius transformation
so that
for some computed .
Step 2: normalize exponents with a gauge factor
Section titled “Step 2: normalize exponents with a gauge factor”Write
and choose so that has the desired local exponents at (often one chooses the exponent at each finite singularity for the canonical Heun normalization).
Step 3: match coefficients to read off and
Section titled “Step 3: match coefficients to read off (γ,δ,ϵ)(\gamma,\delta,\epsilon)(γ,δ,ϵ) and (αβ,q)(\alpha\beta,q)(αβ,q)”After Steps 1–2, bring the equation to the canonical form
Then:
- the residues of the coefficient at give ,
- the numerator of the coefficient gives (coefficient of ) and (minus the constant term).
Finally, enforce the Fuchs relation to determine (or as a consistency check).
11. Common pitfalls
Section titled “11. Common pitfalls”- Forgetting the constraint (software often hides ).
- Using forbidden values or (singularities collide; you enter a confluent regime).
- Logarithmic cases (, similarly at or when or are integers): the naive Frobenius basis changes.
- Small radius of convergence when or : use transformations or re-expansion.
- Branch cut crossings in analytic continuation: results can jump between branches.
- Assuming symmetry in : the equation is symmetric under , but your chosen solution/branch and normalization may not make that symmetry manifest.
- Underestimating precision needs near singular points: high-precision arithmetic is often necessary.
References
Section titled “References”-
Primary convention source for this page:
A. Ronveaux (ed.), Heun’s Differential Equations, Oxford University Press (1995), Part A: “Heun’s equation” (canonical form, local solution , transformations, Heun functions , Heun polynomials ). -
Additional standard references you may see in the literature:
- A. F. Slavyanov and W. Lay, Special Functions: A Unified Theory Based on Singularities (useful for broader special-function context).
- R. S. Maier, work on the 192 solutions and transformation theory (useful for symmetry and connection problems).