Diagnosing Perturbative Expansion Breakdown

1. The Catastrophe of Division by Zero: Level Crossings

The most immediate and catastrophic failure of the non-degenerate perturbation formulas we derived last time occurs when the denominator, $E_n^{(0)}-E_k^{(0)}$, goes to zero. This happens when the perturbation connects two or more states that have the same unperturbed energy—a situation known as degeneracy.

This is a direct violation of the convergence condition we found for the two-level system. The resulting infinities in the correction terms are a clear signal that the straightforward expansion has failed. The physical reason is that in the presence of degeneracy, the unperturbed eigenstates $|n^{(0)}⟩$ are not, in general, the correct "zeroth-order" states. Even an infinitesimal perturbation can drastically mix the degenerate states, selecting a new basis that diagonalizes the perturbation within the degenerate subspace. The naive expansion fails because it assumes the states are only slightly perturbed.

The key diagnostic here is straightforward: if your perturbation theory calculation yields infinite corrections due to vanishing energy denominators, you are dealing with a degenerate or near-degenerate system. The regular expansion breaks down. The proper procedure, which you can see outlined in the next part of the same video (<ts start="00:07:35" end="00:25:21">systematic analysis</ts>), involves diagonalizing the perturbation matrix within the degenerate subspace to find the correct leading-order energies and states. For our purposes, recognizing the symptom and its cause is the crucial step.

2. The Unphysical Growth: Secular Terms

Another common failure mode appears in problems involving oscillations or any time-dependent evolution. Often, a perturbation will cause a small shift in the frequency of an oscillator. A regular perturbation expansion, however, is not well-suited to capture this. Instead of a slightly shifted frequency, the expansion often produces terms that grow in time, such as $t\sin (t)$ or $t\cos (t)$. These are called secular terms.

Their presence is unphysical. A small, constant perturbation to a stable oscillator should result in a bounded, periodic motion with a slightly different frequency, not a solution that grows to infinity. This secular growth signals that the expansion is not uniformly valid; while it might be a good approximation for very short times, it inevitably breaks down as $t$ becomes large (specifically, when $ϵt\sim 1$).

The reading shows exactly how a straightforward expansion for the Duffing oscillator, $y^{\prime \prime }+y+ϵy^3=0$, leads to a secular term. The problem is that the expansion is trying to approximate a function like $\cos (\omega (ϵ)t)$ with a series based on functions with a fixed period, like $\cos (t)$ and $\sin (t)$.

To see this diagnosed in a lecture format, let's watch a video that sets up a similar problem.

The diagnostic is the appearance of these secular terms. When you see them in a calculation, it's a red flag that your expansion is not capturing the long-term behavior correctly. The physical cause is almost always a perturbation-induced shift in a fundamental frequency or period of the system. This breakdown motivates more sophisticated techniques like the Poincaré-Lindstedt method or the method of multiple scales, which are designed to handle such frequency shifts and provide uniformly valid solutions.

3. The Series Itself: Intrinsic Divergence

In the last lesson, we saw that the perturbation series for the anharmonic oscillator diverges. This factorial growth of coefficients is not just a pathology of one specific model; it is a generic feature of perturbation theory in quantum mechanics and, crucially for your interests, in Quantum Field Theory. This divergence is a diagnostic sign of deep physical phenomena that a simple power-series expansion is blind to.

Dyson's Argument and Vacuum Instability

One of the most elegant arguments for the inevitable divergence of QED perturbation theory was put forward by Freeman Dyson. It is a physical argument, not a mathematical one.

The perturbation series for any physical quantity $F$ in QED is a power series in the fine-structure constant $\alpha \approx 1/137$.
$F(\alpha )={\sum }_{n=0}^{\infty }{c}_n{\alpha }^n$
If this series were convergent, it would have a non-zero radius of convergence. This would imply that the function $F(\alpha )$ is analytic at $\alpha =0$, and the series should also converge for some small negative values of $\alpha$.

But what does $\alpha <0$ mean physically? Since $\alpha =e^2/(4\pi {ϵ}_0\hslash c)$, a negative $\alpha$ implies a negative $e^2$, or an imaginary charge $e$. This would flip the sign of the Coulomb force. Like charges would attract, and opposite charges would repel. In such a world, the vacuum would be unstable. A virtual electron-positron pair, which normally would annihilate, would instead fly apart explosively, releasing energy and creating more pairs. The vacuum would "disintegrate".

Since the theory is pathologically unstable for $\alpha <0$, physical quantities like the ground state energy cannot be well-behaved (analytic) functions at $\alpha =0$. Therefore, their Taylor series expansion around $\alpha =0$ cannot be convergent.

Factorial Growth and Non-Perturbative Effects

Dyson's argument tells us the series must diverge, but it doesn't tell us how. The characteristic factorial growth of perturbative coefficients, $c_n\sim n!$, is itself a diagnostic signature. In QFT, this is related to the number of Feynman diagrams at order $n$, which also grows factorially.

More deeply, this behavior signals the existence of non-perturbative effects—physical phenomena that cannot be captured by any finite power series in the coupling constant. These effects are typically exponentially small, scaling like $e^{-C/g^2}$ where $g$ is the coupling. A Taylor series expansion of such a function around $g=0$ has all coefficients equal to zero, so perturbation theory is completely blind to it. The presence of these non-perturbative contributions renders the total physical quantity non-analytic at $g=0$, forcing the perturbative part of the answer to be a divergent, asymptotic series.

Canonical examples of non-perturbative effects include:

• Quantum Tunneling (Instantons): In a double-well potential, the energy splitting between the symmetric and anti-symmetric ground states is proportional to $e^{-S/\hslash }$, where $S$ is the classical action for the tunneling trajectory (the instanton). This splitting is invisible to standard perturbation theory expanded around one of the wells.
• Strong Coupling Phenomena: Effects like confinement in Quantum Chromodynamics (QCD) do not have a description in terms of a perturbative expansion in the strong coupling constant.

The following resource provides a dense but powerful overview of these ideas, which are at the frontier of theoretical physics.

The diagnostic here is the behavior of the coefficients themselves. When you compute perturbative corrections and find that they grow factorially, it is a sign that the problem likely contains essential non-perturbative physics that your expansion has missed.

Conclusion

You are now equipped with a diagnostic toolkit to probe the limits of perturbative methods. A breakdown is not a failure of physics, but a message from it. By learning to read these messages, you elevate yourself from a calculator to a strategist.

Key Takeaways:

• Level Crossings: If a perturbation connects degenerate or nearly degenerate states, the non-degenerate formulas will produce infinite coefficients. This signals the need for degenerate perturbation theory.
• Secular Terms: In problems with oscillations, the appearance of terms that grow unboundedly with time (e.g., $t\sin (t)$) indicates that the expansion is not uniformly valid. This typically points to a perturbation-induced frequency shift, requiring methods like multiple scales.
• Factorial Divergence: When the coefficients of the perturbation series grow like $n!$ or $\Gamma (n)$, the series is asymptotic and divergent. This is a profound signal of underlying non-perturbative physics (like tunneling) that is invisible to a simple power-series expansion. In QFT, this is linked to the proliferation of Feynman diagrams and physical arguments about vacuum stability.

In our next lesson, we will explore one of the most powerful non-perturbative techniques in the physicist's arsenal: the variational method. It provides a way to approximate solutions, particularly ground states, without relying on a small parameter, making it a perfect complement to the perturbative methods we have studied.