The Calculus of Variations and the Action
~6 min read · Tier toggle changes depth (T1 plain · T2 standard · T3 rigour)
You should be comfortable with basic multivariable calculus, specifically partial derivatives and the chain rule, and have a working knowledge of generalized coordinates from the previous lecture.
Why this matters
The action principle replaces the vectorial complexity of force with a single scalar quantity that encodes the entire history of a system.
In Newtonian mechanics, we track forces acting on a particle at every instant to determine its trajectory. This approach is powerful but local; it requires knowing the instantaneous cause of motion. The calculus of variations offers a global perspective. Instead of asking "what force acts now?", we ask "which path makes a specific quantity stationary over the entire time interval?" This shift from local differential equations to global variational principles is the foundation of modern theoretical physics, providing the natural language for quantum mechanics, field theory, and general relativity.
The Idea: Finding the Best Path
Imagine you are a hiker trying to get from point A to point B on a hilly landscape. You want to take the path that requires the least effort. In a flat field, the straight line is the shortest and easiest path. But if there is a river or a steep cliff, the "best" path might be longer in distance but smoother in gradient. The calculus of variations is the mathematical tool that finds this optimal path. It does not look at one step at a time; it looks at the whole journey and compares it to all other possible journeys. In physics, the "effort" is replaced by a quantity called the action, and the "best path" is the one that makes this action stationary (usually a minimum).
This is fundamentally different from Newton's second law. Newton's law is a local rule: at every point, acceleration equals force divided by mass. The variational principle is a global rule: the actual path taken by nature is the one that extremizes the action integral. This global view allows us to handle constraints and complex geometries more elegantly, as we will see in the next lecture when we derive the Euler-Lagrange equations.
Defining the Action
To apply the calculus of variations to mechanics, we must define the quantity we are extremizing. We start with the Lagrangian, \( L \), which is a function of the generalized coordinates \( q_i \), their time derivatives \( \dot{q}_i \), and time \( t \). For a standard mechanical system, the Lagrangian is defined as the difference between the kinetic energy \( T \) and the potential energy \( V \):
The action \( S \) is defined as the time integral of the Lagrangian over the interval between two fixed times \( t_1 \) and \( t_2 \):
Notice that \( S \) is a functional. It takes a function \( q(t) \) as input and returns a single number. The principle of least action states that the actual physical trajectory \( q(t) \) is the one for which the first variation of the action, \( \delta S \), is zero. This is the variational equivalent of setting the derivative of a function to zero to find its minimum.
The Variation Operator
To formalize the idea of "comparing paths," we introduce the variation operator \( \delta \). Suppose we have the true path \( q(t) \). We consider a nearby path \( q'(t) \) defined by adding a small perturbation \( \eta(t) \):
Here, \( \epsilon \) is a small parameter, and \( \eta(t) \) is an arbitrary function that describes the shape of the deviation. Crucially, the endpoints of the path are fixed. This means that at \( t_1 \) and \( t_2 \), the perturbation must vanish:
The variation of the action is the change in \( S \) due to this perturbation. We require that the first-order change in the action be zero for any arbitrary \( \eta(t) \) that satisfies the boundary conditions. This condition, \( \delta S = 0 \), is the starting point for deriving the equations of motion. It ensures that the true path is a stationary point of the action functional.
Why Stationary, Not Minimum?
A common misconception is that the action is always minimized. In many simple systems, such as a free particle in a straight line, the action is indeed a minimum. Over longer intervals, however, the true path can be a saddle point of the action: for a harmonic oscillator this happens once the interval exceeds half a period (a so-called conjugate point). For an ordinary Lagrangian \( L = T - V \) it is never a true maximum, because adding fast wiggles to any path raises the kinetic term without limit. The correct statement is that the action is stationary. This means that the first derivative (variation) is zero, but the second derivative (second variation) can be positive, negative, or zero. This distinction is important because it allows the principle to apply to a wider range of physical phenomena, including systems with oscillatory behavior or unstable equilibria.
The stationarity condition is robust. It does not depend on the specific form of the potential energy or the complexity of the constraints. This universality is one of the great strengths of the Lagrangian formulation. It provides a unified framework for describing diverse physical systems, from classical particles to fields in spacetime.
If the Lagrangian depends explicitly on time (a driven system, or a moving constraint), the energy function \( h = \sum_i \dot q_i\,\partial L/\partial \dot q_i - L \) is not conserved. If it does not, \( h \) is conserved, but \( h \) equals \( T + V \) only when the constraints do not depend on time: for a bead on a hoop spun at constant rate, \( h \) is conserved while \( T + V \) is not. C
Worked example
Problem. Consider a free particle of mass \( m = 2.0\ \mathrm{kg} \) moving in one dimension. The particle moves from \( x_1 = 0\ \mathrm{m} \) at \( t_1 = 0\ \mathrm{s} \) to \( x_2 = 10\ \mathrm{m} \) at \( t_2 = 5\ \mathrm{s} \). Calculate the action \( S \) for the straight-line path connecting these points.
- Identify the Lagrangian. For a free particle, \( V = 0 \), so \( L = T = \frac{1}{2} m \dot{x}^2 \).
- Determine the velocity. The straight-line path implies constant velocity \( v = \frac{x_2 - x_1}{t_2 - t_1} = \frac{10\ \mathrm{m}}{5\ \mathrm{s}} = 2.0\ \mathrm{m/s} \).
- Substitute into the Lagrangian. \( L = \frac{1}{2} (2.0\ \mathrm{kg}) (2.0\ \mathrm{m/s})^2 = 4.0\ \mathrm{J} \).
- Integrate over time. Since \( L \) is constant, \( S = \int_{0}^{5} 4.0\ \mathrm{J} \, dt = 4.0\ \mathrm{J} \times 5.0\ \mathrm{s} \).
Units check. The Lagrangian has units of energy (Joules). Integrating energy over time gives Joule-seconds, which is equivalent to Action. This is consistent with the definition of action as the integral of the Lagrangian.
Three quick tests
- What is the definition of the Lagrangian for a standard mechanical system?
Answer
The difference between kinetic and potential energy: \( L = T - V \). - Why must the variation \( \eta(t) \) vanish at the endpoints \( t_1 \) and \( t_2 \)?
Answer
Because the endpoints of the physical trajectory are fixed by the problem statement; we are comparing paths that start and end at the same points. - Is the action always a minimum?
Answer
No, it is stationary. It is a minimum for short enough time intervals, but over longer ones (for an oscillator, more than half a period) the true path can be a saddle point.
Problem set for PU-201 →
Each derivation in this unit ends with five graded problems with full solutions.
Lecture notes drafted with a local model and checked line by line by the editor; every number in the worked example was recomputed.