Accidental Symmetry

An accidental symmetry is a symmetry or degeneracy that is not required by the obvious symmetry originally used to describe the problem. The phrase is common but potentially misleading: “accidental” does not mean uncaused. It means “not explained by the manifest symmetry under discussion.”

The standard example is the ideal nonrelativistic hydrogen atom. Rotational symmetry explains degeneracy among different $m$ values at fixed $\ell$, but it does not explain why states with different $\ell$ can have the same energy. The extra Coulomb degeneracy is accidental from the viewpoint of ordinary spatial rotations, and it points to a larger hidden structure.

Manifest Symmetry Versus Observed Degeneracy

Suppose a Hamiltonian commutes with a visible symmetry group $G$. Then Hilbert space decomposes into symmetry sectors or irreducible representations. Symmetry can force degeneracy when a multiplet has dimension greater than one.

But spectra can show more degeneracy than $G$ alone requires. A repeated energy value may come from:

• a larger hidden symmetry not yet identified;
• a special functional form of the Hamiltonian;
• a fine-tuned parameter value;
• integrability or separability;
• an approximation that has more symmetry than the real system;
• a numerical coincidence in a small model.

A coincidence at one parameter value is better called an accidental degeneracy unless an operator algebra or transformation group explains the pattern. Any degenerate eigenspace formally permits unitary changes of basis that commute with the Hamiltonian. A useful hidden symmetry does more: explicit operators connect states and explain patterns across the spectrum or a meaningful family of models.

The Perturbation Test

A useful diagnostic is to ask what happens under perturbations that preserve the manifest symmetry.

Let a degenerate subspace $\mathcal D$ have projector $P$. Add a small perturbation $V$ that preserves all visible symmetries under discussion. The first-order splitting inside $\mathcal D$ is controlled by

$$PVP.$$

If symmetry forces $PVP$ to be proportional to the identity inside the whole degenerate subspace, the degeneracy is protected by that symmetry at first order. If a symmetry-preserving perturbation is allowed to have different eigenvalues inside $\mathcal D$, the degeneracy is not protected by the manifest symmetry.

Schur’s lemma forces a scalar within a single irreducible multiplet. It does not force the same scalar on different irreducible multiplets, and repeated copies of one representation can mix. Thus the representation content of $\mathcal D$, not its dimension alone, determines the protection. This test is not a full proof of absence of hidden structure. It is a practical way to decide whether the symmetry already named is enough to explain the degeneracy.

Hydrogen: Rotations Are Not Enough

For a generic central potential,

$$H = \frac{\mathbf P^2}{2\mu} + V(r),$$

rotational symmetry implies

$$[H,L_i]=0.$$

It therefore explains the $2\ell+1$ degeneracy among the states

$$m=-\ell,-\ell+1,\ldots,\ell$$

inside one angular-momentum multiplet. It does not generally imply degeneracy between different $\ell$ sectors.

The Coulomb potential is special:

$$V(r) = - \frac{e^2}{4\pi\epsilon_0 r}.$$

Its bound-state energies depend only on

$$n=n_r+\ell+1,$$

so the energy can be written as

$$E_n = - \frac{\mu e^4}{2(4\pi\epsilon_0)^2\hbar^2} \frac{1}{n^2}.$$

For fixed $n$, all allowed $\ell=0,1,\ldots,n-1$ are degenerate in the ideal spinless model. The total spatial degeneracy is

$$\sum_{\ell=0}^{n-1}(2\ell+1) = n^2.$$

The $m$ degeneracy is rotational. The additional degeneracy across different $\ell$ values is the famous accidental Coulomb degeneracy. The full derivation is in Degeneracy of the Hydrogen Atom.

Hidden Symmetry Interpretation

The hydrogen degeneracy is not a random numerical accident. The Coulomb problem has additional conserved quantities related to the Laplace–Runge–Lenz vector. In the bound-state sector, these can be organized into a larger symmetry algebra, often described as an $\mathrm{SO}(4)$ structure. The operator-algebra viewpoint is the subject of Hidden Symmetry.

From the viewpoint of ordinary rotations, the degeneracy is accidental. From the viewpoint of the enlarged algebra, it is symmetry-organized.

This is why terminology must be contextual. A degeneracy can be “accidental” relative to the first symmetry you noticed and “explained” after a hidden symmetry is found.

Isotropic Oscillator Example

The multidimensional isotropic harmonic oscillator has a similar lesson. In $d$ dimensions,

$$H = \sum_{i=1}^{d} \left( \frac{P_i^2}{2m} + \frac12m\omega^2X_i^2 \right).$$

The Cartesian number-state energies are

$$E_N = \hbar\omega \left( N+\frac d2 \right), \qquad N=n_1+\cdots+n_d.$$

For fixed $N$, many different tuples $(n_1,\ldots,n_d)$ have the same energy. The number of such tuples is

$$g_N^{(d)} = \binom{N+d-1}{d-1}.$$

Rotational symmetry explains part of the structure, but the full oscillator degeneracy is tied to the equal frequencies and the oscillator ladder algebra. The spectrum-generating side of this algebra is discussed in Dynamical Symmetry. If the frequencies are changed to unequal values,

$$H = \sum_i \left( \frac{P_i^2}{2m} + \frac12m\omega_i^2X_i^2 \right),$$

the isotropic degeneracy is generally reduced. Commensurate unequal frequencies are an instructive exception: with $\omega_y=2\omega_x$, the states $(n_x,n_y)=(2,0)$ and $(0,1)$ still have the same energy. A higher-order ladder operator connects them, as shown in hidden symmetry. Equal frequencies produce the larger bilinear algebra; rational frequency ratios can leave a smaller structure.

Fine-Tuned Accidents

Some degeneracies are less profound. Consider a Hamiltonian depending on a parameter $\lambda$ with two energy branches $E_1(\lambda)$ and $E_2(\lambda)$. If

$$E_1(\lambda_0) = E_2(\lambda_0)$$

at one value $\lambda_0$, but no symmetry relates the two states, a generic perturbation will split the crossing. This is an accidental degeneracy in the narrow sense.

Such coincidences are not useless. They can mark level crossings, avoided crossings after symmetry-breaking perturbations, or parameter values where an effective description changes character. But one should not infer a symmetry merely from equality of two numbers.

Relation to Approximate Symmetry

Accidental degeneracies often appear in idealized models. Real corrections may lift them:

• fine structure splits parts of the ideal hydrogen spectrum;
• Lamb-shift physics splits levels degenerate in simpler models;
• anharmonic terms split oscillator degeneracies;
• crystal fields split atomic or molecular multiplets;
• external fields split degeneracies by selecting directions.

If the corrections are small, the ideal accidental symmetry can remain an approximate symmetry over a useful regime. The labels remain good until the degeneracy lifting or mixing becomes experimentally or dynamically important.

How the Terms Fit Together

These terms answer different questions and can overlap. The ideal Coulomb symmetry is exact, hidden in the coordinate description, and responsible for degeneracies called accidental relative to ordinary rotations.

An exact symmetry is a stated invariance of the Hamiltonian or observable structure. A hidden symmetry is an exact structure that is not manifest in the first variables or geometric picture. An emergent symmetry appears in an effective description, often at low energies or long distances, even if it is absent microscopically.

Accidental symmetry is a diagnostic label used when the obvious symmetry does not account for a spectral pattern. Sometimes it is a clue to hidden symmetry. Sometimes it is a clue to fine tuning. The next step is always to test what operators commute with $H$ and what perturbations preserve or lift the degeneracy.

Common Mistakes

• Inferring a symmetry solely from an observed degeneracy.
• Calling the hydrogen $\ell$ degeneracy rotational.
• Treating “accidental” as meaning “unexplained forever.”
• Forgetting that hidden symmetry may become visible in a different operator basis.
• Assuming a degeneracy survives all small perturbations because it occurs in a famous ideal model.
• Confusing approximate equality of levels with exact degeneracy.
• Ignoring the subspace matrix $PVP$ when deciding whether a degeneracy is protected.

Exercises

1. Separate the hydrogen degeneracies.

For $n=3$, list the allowed $\ell$ values and the number of $m$ states in each. Which part is explained by rotations?

Solution

For $n=3$, the allowed values are

$$\ell=0,1,2.$$

The corresponding $m$ degeneracies are

$$2\ell+1 = 1,\ 3,\ 5.$$

Rotational symmetry explains the $3$ states inside the $\ell=1$ multiplet and the $5$ states inside the $\ell=2$ multiplet. It does not explain why the $\ell=0$, $\ell=1$, and $\ell=2$ sectors have the same Coulomb energy. That extra equality is the special Coulomb degeneracy.

2. Count oscillator degeneracy.

For a two-dimensional isotropic oscillator, how many Cartesian number states have total $N=3$?

Solution

The states satisfy

$$n_x+n_y=3,$$

with nonnegative integers. The possibilities are

$$(3,0),\quad (2,1),\quad (1,2),\quad (0,3).$$

There are four states, in agreement with

$$g_N^{(2)} = \binom{N+1}{1} = N+1 = 4.$$

3. Test a twofold accidental degeneracy.

Let a degenerate subspace be spanned by $\lvert1\rangle,\lvert2\rangle$. A symmetry-preserving perturbation is allowed to have

$$PVP = \begin{pmatrix} a&0\\ 0&b \end{pmatrix}.$$

What condition preserves the degeneracy at first order?

Solution

The first-order shifts are the eigenvalues of $PVP$, namely $a$ and $b$. The degeneracy is preserved at first order only if

$$a=b.$$

If the manifest symmetry allows $a\ne b$, then that symmetry does not protect the twofold degeneracy.

4. Split the Coulomb degeneracy without breaking rotations.

Add a weak central perturbation $V=\lambda r^2$ to hydrogen. Set $a=\hbar^2/(\mu\kappa)$ and $x=r/a$. The normalized radial functions for the $n=2$ states are

$$R_{20}(r)=\frac{(2-x)e^{-x/2}}{2\sqrt2\,a^{3/2}}, \qquad R_{21}(r)=\frac{xe^{-x/2}}{2\sqrt6\,a^{3/2}}.$$

Calculate the first-order shifts using $\int_0^\infty x^n e^{-x}\,dx=n!$. Which degeneracy survives?

Solution

The angular functions are normalized, so $\langle r^2\rangle=\int_0^\infty r^4|R(r)|^2\,dr$. Hence

$$\begin{aligned} \langle r^2\rangle_{2s} &=\frac{a^2}{8}(4\cdot4!-4\cdot5!+6!)=42a^2,\\ \langle r^2\rangle_{2p} &=\frac{a^2}{24}6!=30a^2. \end{aligned}$$

A central perturbation has no matrix element between different $\ell$ values, so these are already the eigenvalues of the perturbation in the degenerate shell. The shifts are $42\lambda a^2$ and $30\lambda a^2$, separated by $12\lambda a^2$. All three $2p$ magnetic sublevels retain the same shift. Rotations protect their $m$ degeneracy, but not the equality between $2s$ and $2p$.

References

• M. Bander and C. Itzykson, “Group theory and the hydrogen atom (I)”, Reviews of Modern Physics 38, 330–345, 1966.
• H. F. Jones, Groups, Representations and Physics, 2nd ed., CRC Press, 1998.
• L. D. Landau and E. M. Lifshitz, Quantum Mechanics: Non-Relativistic Theory, 3rd ed., Pergamon Press, 1977.
• J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, 3rd ed., Cambridge University Press, 2020.
• R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.