Approximation Validity: Physical Arguments

1. The General Principle: Comparing What's Kept with What's Lost

At its core, assessing the validity of an approximation is a game of scales. The fundamental strategy is to compare the magnitude of the terms you have neglected in your model to the magnitude of the terms you have kept. If the neglected physics contributes only a small correction, the approximation holds. When the neglected terms become comparable to or larger than the ones you've kept, the approximation breaks down, often spectacularly. The art lies in identifying the correct physical quantities to compare.

2. Perturbation Theory: It's Not Just About $\lambda$

In our initial look at perturbation theory, the rule of thumb was that the expansion parameter, $\lambda$, must be small. However, this is a mathematically incomplete picture. The physical context, specifically the energy spectrum of the system, is crucial. A small $\lambda$ may not be small enough if the system has closely-spaced energy levels.

A simple, exactly solvable two-level system reveals the essential physics.

The video's analysis of the two-level system leads to a crucial physical condition. For perturbation theory to be reliable, the energy shift induced by the perturbation must be small compared to the energy separation between the states that the perturbation mixes.

Formally, for any two unperturbed states $|n^{(0)}⟩$ and $|k^{(0)}⟩$, the condition is:

Physical Argument: The perturbation $V$ causes transitions, or "mixing," between the unperturbed eigenstates. The denominator $E_n^{(0)}-E_k^{(0)}$ in the perturbation theory formulas represents the "energy cost" of this mixing. If the perturbation's matrix element $V_{kn}$ is large compared to this energy cost (i.e., if the levels are nearly degenerate), the perturbation will cause a drastic restructuring of the eigenstates. The original states are no longer a good starting point, and the series will converge poorly or not at all. This condition is far more physically meaningful than the simple statement $\lambda \ll 1$.

3. The Semiclassical (WKB) Approximation

The WKB (Wentzel-Kramers-Brillouin) method is a powerful tool for finding approximate solutions to the one-dimensional Schrödinger equation in regions where the potential $V(x)$ varies slowly. It's fundamentally a semiclassical approximation, assuming the wavefunction locally resembles a plane wave, $\psi (x)\sim e^{ip(x)x/\hslash }$, but where the momentum $p(x)=\sqrt{2m(E-V(x))}$ itself changes with position.

The question of validity becomes: what, precisely, does "slowly varying" mean?

The derivation shows that the validity of the WKB approximation rests on the condition:

where $\lambda (x)=h/p(x)$ is the position-dependent de Broglie wavelength.

Physical Interpretation: This elegant condition states that the fractional change in the de Broglie wavelength over a distance of one wavelength must be very small. In other words, the particle's quantum nature (its wavelength) must not change abruptly from one point to the next. The potential landscape should be smooth on the scale of $\lambda (x)$.

An equivalent statement, also derived in the video, is that the change in potential energy over one de Broglie wavelength must be much smaller than the particle's kinetic energy:

This reinforces the idea that the background potential is not changing fast enough to significantly alter the particle's kinetic state over the distance of its own wavelength.

Breakdown of WKB

This physical condition immediately tells us where the WKB approximation must fail: at the classical turning points, where $E=V(x)$.

As $x$ approaches a turning point, the classical momentum $p(x)\to 0$, causing the de Broglie wavelength $\lambda (x)\to \infty$. Consequently, its derivative $d\lambda /dx$ diverges, and our validity condition is catastrophically violated. This is a classic example of an approximation signaling its own breakdown. To get a global solution, one must use a different approximation near the turning points and then "match" it to the WKB solutions on either side.

4. Mean-Field Theory and the Ginzburg Criterion

Let's turn to condensed matter, a field where you have extensive experience. Mean-Field Theory (MFT) is a cornerstone for understanding interacting many-body systems like magnets and superconductors. Its central move is to approximate the interaction of one particle with its neighbors by an interaction with an average, or "mean," field generated by them. For instance, in the Ising model, the interaction term is approximated as:

where $m=⟨{\sigma }_j⟩$ is the average magnetization (the order parameter) and $z$ is the number of nearest neighbors.

This approximation explicitly neglects fluctuations of the spins around their average value. The validity of MFT therefore depends on whether these fluctuations are small enough to be ignored. The Ginzburg criterion provides the physical condition for this.

The Ginzburg criterion establishes that MFT is valid when the reduced temperature $t=|T-T_c|/T_c$ is much larger than a characteristic value, the Ginzburg reduced temperature $t_G$.

Physical Interpretation:

• $d$ is the spatial dimension of the system.
• $a$ is a microscopic length scale (e.g., the lattice constant).
• $R_{\ast }$ is a length scale related to the range of the interaction.

The criterion tells us that MFT is valid when fluctuations are small. This happens under two main conditions:

1. High Dimensionality ($d>4$): The exponent $\frac{2d}{4-d}$ becomes negative for $d>4$. In this case, $t_G$ is effectively zero, and the criterion is always satisfied. Fluctuations become less important in higher dimensions because a particle has more neighbors, making the average field a better representation. $d=4$ is called the upper critical dimension for this type of model.
2. Long Interaction Range ($R_{\ast }\gg a$): If the interaction range is very large, each particle interacts with many others, and fluctuations tend to average out. This makes $t_G$ extremely small, widening the regime where MFT is valid. This is why MFT (in the form of BCS theory) is extraordinarily successful for conventional superconductors, where the "interaction range" is the size of a Cooper pair, which can be hundreds of lattice spacings. Conversely, for a typical magnet where interactions are nearest-neighbor ($R_{\ast }\sim a$), $t_G\sim 1$, and MFT fails badly near the critical point.

5. A Note on the Adiabatic Theorem

Finally, let's briefly touch upon the adiabatic theorem, which you noted you were less comfortable with. The theorem states that if a system starts in an energy eigenstate of a Hamiltonian $H(0)$, and the Hamiltonian is changed slowly over time to $H(T)$, the system will end up in the corresponding eigenstate of $H(T)$. The condition for validity is another beautiful example of our general principle.

The adiabatic approximation is valid if the rate of change of the Hamiltonian is slow compared to the energy gaps of the system. More precisely, for any two states $|n⟩$ and $|m⟩$:

Physical Interpretation: The system has natural internal timescales, given by ${\tau }_{nm}\sim \hslash /|E_n-E_m|$. The approximation holds if the timescale over which the Hamiltonian changes is much longer than these internal timescales. If you try to "drive" the system faster than it can naturally respond, you will induce transitions, kicking it out of its initial eigenstate. This is why the approximation must fail if the system encounters a level crossing, where $E_n=E_m$ for some $m$. At this point, the energy gap vanishes, the internal timescale becomes infinite, and any finite rate of change is "too fast."

Conclusion

Today we've transformed the process of approximation from a mere calculational tool into an object of physical inquiry itself. By consistently asking "What have we neglected, and how does it compare to what we've kept?", we can derive powerful, intuitive conditions that define the boundaries of our theories.

Key Takeaways:

• Validity is Physical: Assessing an approximation's domain of validity is a physical exercise in comparing the scales of different effects.
• Perturbation Theory: Fails when the perturbation strength is comparable to the energy gaps of the unperturbed system, leading to strong state mixing.
• WKB Approximation: Valid when the de Broglie wavelength changes slowly. It breaks down at classical turning points where the wavelength diverges.
• Mean-Field Theory: Valid when fluctuations are small. The Ginzburg criterion quantifies this, showing MFT works well in high dimensions or for long-range interactions, but fails near criticality in low dimensions.
• Adiabatic Theorem: Valid when the Hamiltonian changes on a timescale much longer than the system's internal dynamical timescales (set by inverse energy gaps). It breaks down at level crossings.

Mastering this skill of critical assessment is what separates a technician from a theorist. It allows you to build and deploy simplified models with confidence, always aware of their inherent limitations and the direction in which to look for new physics when they fail.

In our next module, we will confront a common type of failure head-on: the appearance of divergent series in perturbation theory. This is not always a disaster. We will see how to work with such series and extract meaningful physical results, beginning our exploration of asymptotic and semiclassical methods.