Identifying Small Parameters for Perturbation Theory

Welcome to the first lesson of our module on approximation methods. In our last discussion, we explored how physicists compare competing theories using the principles of parsimony and predictive power. This often involves effective theories that are valid at specific energy scales and can be expressed as an expansion in some "small" parameter. This naturally raises a crucial question: how do we rigorously identify what makes a parameter "small" enough to serve as the basis for a reliable approximation?

This lesson will equip you with the foundational skill of perturbation theory: identifying and classifying the small parameters that make this powerful technique possible. You will learn that the "smallness" of a parameter is not an absolute concept but is always relative and, crucially, dimensionless. We will delve into the systematic process of non-dimensionalization, which transforms a physical problem into a form where these critical dimensionless ratios emerge naturally.

1. The Necessity of Being Dimensionless

At the outset, we must confront a subtle but critical point. What does it mean for a physical quantity to be "small"? For instance, in your QFT course, you frequently encounter Planck's constant, $ℏ \approx 1.05 \times 1 0^{- 34} J \cdot s$ . This number is numerically tiny, but does that automatically make it a valid expansion parameter? If we changed our units to, say, MeV and femtoseconds, the numerical value would change. A quantity with physical dimensions cannot be small in and of itself; its value is contingent on the system of units we choose to adopt.

The entire framework of perturbation theory rests on finding a truly small number to expand in, one that is independent of our arbitrary choice of units. This means the expansion parameter must be dimensionless.

When is a parameter considered small for perturbation and how ...

This discussion on Physics Stack Exchange directly addresses this foundational question. It clarifies why perturbation theory must be formulated in terms of a dimensionless parameter.

Please read the question and the first part of the top answer. Focus on these key points: The initial question, which highlights the ambiguity of a "small" dimensionful parameter. The core of the answer, which explains that we formally introduce a dimensionless parameter \lambda for the expansion. The original physical parameter \epsilon is then related to \lambda.

The key insight is that our task is to take a physical problem, with all its dimensionful parameters (masses, charges, velocities), and reformulate it. We need to find the inherent ratios of physical scales within the problem. The small parameter $ϵ$ will always be a ratio of two quantities with the same units—a ratio of lengths, a ratio of energies, a ratio of times, etc.—making it a pure number. The process for revealing these ratios is called non-dimensionalization.

2. Unveiling the Small Parameter: Non-dimensionalization

Non-dimensionalization is a systematic procedure for recasting a problem's equations in terms of dimensionless variables and parameters. It achieves two goals simultaneously: it simplifies the algebra by reducing the number of parameters, and it reveals the fundamental dimensionless groupings that govern the system's behavior. One of these groupings will often be our small parameter $ϵ$ .

The general strategy, as described in the Hunter notes, is to identify the "characteristic scales" of the problem—a typical length $L$ , a typical time $T$ , or a typical energy $E_{0}$ —and rescale all variables by these quantities. For example, in a fluid dynamics problem, if a sphere of radius $L$ moves at speed $U$ through a fluid with kinematic viscosity $ν$ , we can define a dimensionless position $x^{*} = x / L$ , time $t^{*} = t U / L$ , and velocity $u^{*} = u / U$ . When you substitute these into the Navier-Stokes equations, you find that the entire dynamics depends on a single dimensionless number: the Reynolds number, $R e = U L / ν$ . If you are studying a situation where viscous forces dominate inertial forces (e.g., a tiny particle in honey), then $R e ≪ 1$ , and the Reynolds number itself becomes the small parameter for a perturbative expansion.

Let's see this procedure applied to a concrete mechanics problem. We'll analyze the motion of a rocket launched from Earth, but unlike the introductory physics version, we will account for the fact that gravity weakens with altitude. This weakening is a "small" correction to the constant-gravity model.

Lecture 11: Regular perturbation methods for ODEs

This lecture by Steven Strogatz provides a masterclass in setting up a perturbation problem, starting with a physical model and systematically deriving the small parameter.

Please watch the following two segments: The Projectile Problem: Prof. Strogatz sets up the equation of motion for a rocket, including the inverse-square law for gravity. Pay attention to how he frames the problem as a small correction to the standard constant-g case. Scaling and Non-dimensionalization: This is the most critical part for our current lesson. Observe the step-by-step process of choosing characteristic scales for distance (x_c) and time (t_c), defining dimensionless variables, and rewriting the ODE. See how the small, dimensionless parameter \epsilon emerges naturally from this process.

Let's recap the essential steps from the video, as they represent a general algorithm:

Write the Full Equation of Motion: Start with Newton's second law, keeping all the physical parameters. For the rocket, this was:
$m x ¨ = - \frac{GM m}{( R + x ) ^{2}}$
where $R$ is the Earth's radius. This was rewritten as:
$x ¨ = - \frac{g R ^{2}}{( R + x ) ^{2}}$
Identify Characteristic Scales: Look at the simplified ( $ϵ = 0$ ) version of the problem to find natural scales. For a simple projectile launched with velocity $v_{0}$ under constant gravity $g$ , the time to reach the peak is $v_{0} / g$ and the maximum height is $v_{0}^{2} / (2 g)$ . Based on this, Strogatz chooses the characteristic scales:
- Characteristic time: $t_{c} = v_{0} / g$
- Characteristic distance: $x_{c} = v_{0}^{2} / g$
Define Dimensionless Variables: Rescale the original variables by these characteristic scales.
- Dimensionless time: $τ = t / t_{c}$
- Dimensionless position: $y = x / x_{c}$
Rewrite the ODE: Substitute these new variables into the full equation of motion and simplify. Through a chain of algebra, the equation becomes:
$\frac{d ^{2} y}{d τ ^{2}} = - \frac{1}{( 1 + ϵy ) ^{2}}$
where the dimensionless parameter $ϵ$ has emerged:
$ϵ = \frac{x _{c}}{R} = \frac{v _{0}^{2}}{g R}$

This final expression for $ϵ$ is the payoff. It is manifestly dimensionless—a ratio of two lengths (characteristic altitude to Earth's radius) or, equivalently, a ratio of two velocities squared ( $v_{0}^{2}$ to $g R$ , where $g R$ is related to orbital velocity). The physical assumption that we are launching a "toy rocket" that doesn't go very high translates directly into the mathematical statement $ϵ ≪ 1$ . We have successfully identified our small parameter.

While this intuitive method of choosing characteristic scales is often sufficient, a more formal algorithm called the Buckingham Pi theorem can also be used. As we covered in Module 2, this theorem states that a system with $k$ variables and parameters described by $r$ fundamental dimensions can be described by $k - r$ independent dimensionless groups. This provides a systematic way to find all the relevant dimensionless numbers (like the Reynolds number and our $ϵ$ ) in a complex problem.

3. Classifying Perturbation Problems: Regular vs. Singular

Now that we know how to find $ϵ$ , we can use it to classify the problem. This classification depends on what happens to the problem in the limit $ϵ \to 0$ . This distinction is fundamental because it dictates the entire mathematical strategy for finding a solution.

[PDF] Asymptotic Analysis and Singular Perturbation Theory

This textbook chapter by John Hunter clearly defines and illustrates the crucial difference between regular and singular perturbation problems.

Please read the following short sections: Section 1.1.2 introduces the definitions of regular and singular perturbation problems. The text of Examples 1.1 and 1.2 provides simple, concrete algebraic illustrations of this distinction. Focus on the qualitative change in Example 1.2.

Let's formalize the distinction:

Regular Perturbation Problem: The problem for $ϵ \to 0$ is qualitatively the same as the problem for small, non-zero $ϵ$ . The unperturbed problem is a good "first guess" for the full problem. The rocket problem is regular: setting $ϵ = 0$ gives $y ¨ = - 1$ , the equation for constant gravity. The behavior (parabolic motion) is qualitatively similar to the $ϵ \neq = 0$ case. Solutions to regular problems can typically be found as a straightforward power series in $ϵ$ :
$y (τ) = y_{0} (τ) + ϵ y_{1} (τ) + ϵ^{2} y_{2} (τ) + \dots$
Singular Perturbation Problem: The limit $ϵ \to 0$ fundamentally changes the character of the problem. As in Hunter's Example 1.2, $ϵ x^{3} - x + 1 = 0$ , the order of the equation changes. The $ϵ = 0$ problem is a linear equation with one root, while the full problem is a cubic equation with three roots. Two solutions are "lost" in the limit. This often happens in differential equations when $ϵ$ multiplies the highest derivative term. The solution to a singular problem is often not a simple power series and requires more sophisticated techniques.

For now, it is enough to recognize this classification. The rocket problem we analyzed is a regular perturbation problem because the unperturbed ( $ϵ = 0$ ) problem (constant gravity) and the perturbed problem (variable gravity) are qualitatively similar, and the parameter $ϵ$ does not multiply the highest derivative.

Conclusion

In this lesson, you have taken the first, most critical step in applying perturbation theory. You've learned to move beyond a vague notion of "smallness" to a rigorous, systematic procedure for identifying the parameters that make perturbative analysis possible.

Here are the key takeaways:

A parameter for a perturbative expansion must be dimensionless. Its smallness is defined by being a ratio of two competing physical scales.
Non-dimensionalization is the core technique used to reveal these dimensionless parameters. It involves rescaling the problem's variables using characteristic scales.
A physical problem can be classified as regular or singular based on its behavior in the $ϵ \to 0$ limit. A regular problem is qualitatively unchanged, while a singular problem undergoes a fundamental change, such as a change in the number of solutions or the order of a differential equation.

Having learned to identify and classify the small parameter, you are now prepared to tackle different types of perturbation problems. In our next lesson, we will focus on the distinction between regular and singular problems in more detail, exploring the rich and often surprising physics that emerges from singular limits.

Recommended resources

1. The Necessity of Being Dimensionless

2. Unveiling the Small Parameter: Non-dimensionalization

3. Classifying Perturbation Problems: Regular vs. Singular

Conclusion