The Lie group and representation theory approach to special functions, and how they solve the ODEs arising in physics is absolutely amazing. I've given an example of its power below on Bessel's equation.

Kaufman's article describes algebraic methods for dealing with Hermite, Legendre & associated Legendre functions. Can we take the other special functions mentioned in this paper, obtainable as linear combinations of the conformal symmetries of the Laplacian (expressed as Lie algebra elements), and obtain their solution analogously to how Bessel is solved below? I believe it's something like a geometric interpretation of Weisner's method.

Bessel's equation seems to be saying: find a function in the plane such that when we shift it right, then shift it back left again, all locally (i.e. differentially) in polar coordinates, we get the same function:

enter image description here

(c.f. Killingbeck, Mathematical Technique's and Applications, sec. 8.21).

The idea is to take Bessel's equation, factor it, add an extra variable to make the factors parameter independent so that they become elements of a Lie algebra, identify the meaning of those factors, in this case notice the Lie algebra factors are translations in polar coordinates, and realize it's just a differential expression of a symmetry.

Bessel's equation arises from $L R v = v$ when you express $L$ & $R$ in polar coordinates. It makes sense to express them in polar coordinates since Bessel arises from separating the Laplacian assuming cylindrical symmetry, and the $LRv = v$ assumption (not $LRv = w$) is motivated by the symmetry of the Laplacian.

Using this idea we can, for some reason, actually solve Bessel's equation with a picture!:

enter image description here

We just want to shift $\mathcal{J}_n(r,\phi)$ in the x-direction using the operator $e^{a\tfrac{\partial}{\partial x}}$ expressed in polar coordinates: $e^{\tfrac{a}{2}(\mathcal{L}-\mathcal{R})}$ and realize it will be equal to $\mathcal{J}_n(r',\phi')$:

enter image description here

So the last line comes from dragging all this to the origin and putting it along the x-axis, here we see the geometric meaning of Bessel functions!

(I am not yet completely sure how we link this calculation to the matrix representation of the Euclidean Group $E_2$

$$ g(x,y,\theta) = \left( \begin{array}{ccc} \cos(\theta) & - \sin(\theta) & x \\ \sin(\theta) & \cos(\theta) & y \\ 0 & 0 & 1 \end{array} \right)$$

Sadly, vaguely we can see it represents planar motions, and this is all linked to representation theory, but it's not fully clear to me how exactly).

The question is: Is there a similar, easy, unified, geometric exposition for the rest of these types of equations, analogous to [or a better version of] this one, such as those in the following list: enter image description here

It'd be great to understand these other equations, their formulation and solution, with a geometric interpretation like this one. Understanding the intuition behind those groups seems to be the key. If you you feel so inspired, please post your versions of these problems here.


  1. Killingbeck, Mathematical Techniques and Applications, sec. 8.21
  2. Vilenkin, Representation of Lie Groups and Special Functions Vol. 1
  3. Vilenkin, Special Functions and Theory of Group Representations
  4. Miller, Lie Theory and Special Functions
  5. Kaufman, Special Functions of Mathematical Physics from the Viewpoint of Lie Algebra
  • 21
    $\begingroup$ This question is very well put together. Nice job. Don't have any immediate answers, though. But that's the mark of a real question. Plus One! $\endgroup$ Dec 11, 2014 at 20:22
  • 6
    $\begingroup$ You might enjoy Olver's Applications of Lie Groups to Differential Equations. It is very informative, and it becomes fairly intuitive once you get a handle on jets. $\endgroup$ Dec 11, 2014 at 20:22
  • 2
    $\begingroup$ It seems to me that the above has more to do with Operator Calculus than with Lie groups, though I know that the two are interconnected (in Quantum Mechanics at least). (+1) anyway. $\endgroup$ Feb 27, 2015 at 16:16
  • 1
    $\begingroup$ A little off the point: but I have construct a merging of Roman's Umbral calculus, Sheffer sequences, with Pascal Matrices that shows that the polynomials mentioned are all matrix similar via Jordan normal forms that gives an explicit conversion form that is basically a transform of the independent variables and scaling. It also extends a little beyond Sheffer sequences into Jacobi Polynomials and such. I haven't extended my intuition into the Lie algebra approach although they seem related. $\endgroup$
    – rrogers
    Apr 3, 2016 at 18:15
  • 1
    $\begingroup$ @G.Bergeron hi, have you found the time for this? It would be great if you can answer this $\endgroup$
    – Ooker
    Nov 24, 2017 at 10:00

1 Answer 1


The Lie method for differential equations occurs in chapter 29 of Sadri Hassani's Mathematical Physics text ( the advanced text) Springer Pub.

  • $\begingroup$ maybe best as a comment, but since you have no reputation I think this is OK. $\endgroup$ Nov 22, 2016 at 3:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .