With In-Depth Discussions, Logical Reasoning and Practical Applications
Lagrangian mechanics is a reformulation of classical mechanics introduced by Joseph-Louis Lagrange in 1788. It provides a powerful and elegant framework for analyzing the dynamics of systems, especially those with constraints. Unlike Newtonian mechanics, which focuses on forces and accelerations, Lagrangian mechanics is based on energy principles, making it particularly useful for complex systems.
The core idea is to describe the state of a system using generalized coordinates and derive the equations of motion from a single scalar function: the Lagrangian. This approach simplifies the analysis of systems with multiple degrees of freedom and constraints.
This whitepaper explores the foundational principles of Lagrangian mechanics, its advantages over Newtonian mechanics, and its applications to simple and complex systems. We examine constraints, the principle of virtual work, D'Alembert's principle, Hamilton's variational principle, and the deep connection between symmetries and conservation laws.
Constraints are conditions that restrict the motion of a system. They reduce the number of independent degrees of freedom and can be classified according to their mathematical form and time dependence.
Holonomic constraints can be expressed as equations involving only the coordinates and time (integrable relations). They reduce the configuration-space dimension.
Example: A bead sliding on a circular hoop of radius R in a vertical plane.
x² + y² − R² = 0
Non-holonomic constraints cannot be written solely in terms of coordinates and time; they typically involve velocities and are non-integrable.
Example: A rolling wheel without slipping.
v − Rω = 0
where v is the linear speed, R the radius, and ω the angular speed.
Time-independent constraints. The constraint surface in configuration space is fixed.
Example: A particle constrained to the surface of a fixed sphere.
Explicitly time-dependent constraints. The allowed configuration changes with time.
Example: A bead on a wire that is itself rotating or translating.
The relative prevalence and practical importance of constraint types in typical mechanical systems can be illustrated schematically. Holonomic constraints dominate textbook problems; non-holonomic constraints appear in rolling contact, non-holonomic control systems, and certain rigid-body kinematics.
Relative emphasis of constraint classes in classical mechanics literature and applications.
Newtonian mechanics is intuitive and remains indispensable, yet it becomes cumbersome for constrained, multi-body, or non-Cartesian systems:
Each body requires its own set of Newton–Euler equations. Constraint forces must be retained (or eliminated by additional algebraic steps), rapidly increasing the size of the system.
Constraint forces are a priori unknown. Introducing them enlarges the equation set; eliminating them often requires clever projections or Lagrange multipliers after the fact.
Accelerating frames demand fictitious forces (Coriolis, centrifugal, Euler). These terms must be derived carefully and can obscure the underlying physics.
Cartesian components are natural for Newton’s laws. In polar, spherical, or generalized coordinates the force and acceleration expressions become lengthy and error-prone.
Lagrangian mechanics addresses these difficulties by shifting attention from vector forces to a scalar energy function and by working directly with independent generalized coordinates that already incorporate the holonomic constraints.
For a system in static equilibrium the virtual work of all applied forces vanishes for every virtual displacement consistent with the constraints:
δW = Σ Fi · δri = 0
Virtual displacements δr are infinitesimal, instantaneous, and compatible with the instantaneous constraints; they need not correspond to any actual motion that the system ever executes.
Because constraint forces do no virtual work (they are orthogonal to the allowed displacements), they drop out of the principle automatically. Equilibrium conditions can therefore be written solely in terms of the applied forces and the geometry of the constraints.
Red particle under two applied forces (green, blue). Drag to rotate; the oscillating motion suggests a virtual displacement consistent with constraints.
D'Alembert’s principle extends the principle of virtual work to dynamics by introducing inertial forces −ma. The virtual work of the effective forces vanishes:
Σ (Fi − miai) · δri = 0
The system is thereby treated as if it were in equilibrium under the combined action of applied and inertial forces. This is the conceptual bridge from statics to the Lagrangian equations of motion.
Gravity and tension act on the bob. Tension is a constraint force and does no virtual work for a displacement along the arc. D’Alembert’s principle immediately yields the torque equation that becomes the familiar pendulum equation after projection onto the tangential direction.
Generalized coordinates {qi} are any set of independent parameters that uniquely specify the configuration of the system. They need not be Cartesian lengths; angles, arc lengths, or more abstract parameters are admissible.
The momentum conjugate to qi is defined by
pi = ∂L / ∂q̇i
These momenta become the fundamental variables of Hamiltonian mechanics and appear in the quantum commutation relations after canonical quantization.
For a free particle in the plane:
pr = mṙ , pθ = m r² θ̇
The angular momentum pθ is conserved when the potential is central—an immediate consequence of the cyclic nature of θ.
The Lagrangian is the scalar
L(q, q̇, t) = T − V
where T is the kinetic energy and V the potential energy. For standard mechanical systems T is quadratic in the velocities; V depends only on the coordinates (and possibly time).
T = Σi ½ mi vi²
Potential energy is defined only for conservative forces; non-conservative generalized forces can be added later on the right-hand side of Lagrange’s equations.
T = ½ m ẋ² , V = ½ k x²
L = ½ m ẋ² − ½ k x²
Mass-spring oscillator. Kinetic energy peaks at the equilibrium position; potential energy peaks at the extremes.
Time histories of T, V and total energy E = T + V for an undamped harmonic oscillator (E is conserved).
For each independent generalized coordinate the Euler–Lagrange equation holds:
d/dt (∂L / ∂q̇i) − ∂L / ∂qi = 0
(or = Qi if non-conservative generalized forces are present).
The same equations follow from Hamilton’s principle by requiring that the first variation of the action vanish for fixed endpoints.
Generalized coordinate θ (angle from the downward vertical). Length ℓ, mass m:
T = ½ m ℓ² θ̇² , V = −m g ℓ cos θ
L = ½ m ℓ² θ̇² + m g ℓ cos θ
Application of the Euler–Lagrange equation immediately produces
θ̈ + (g/ℓ) sin θ = 0
For small angles sin θ ≈ θ and the motion becomes simple harmonic with angular frequency √(g/ℓ).
The form of the equations is the same in every coordinate system. One simply rewrites T and V in the new variables.
When generalized coordinates already satisfy the holonomic constraints, the constraint forces never appear.
Multi-body mechanisms, robots, and continuum systems are routinely treated by writing a single Lagrangian and applying the same formal steps.
Noether’s theorem associates continuous symmetries of L with conserved quantities (energy, momentum, angular momentum, \ldots).
The Lagrangian density is the starting point of the path-integral formulation and of classical field theories.
Because L is built from energies, qualitative reasoning about stability, boundedness, and resonances becomes more transparent.
A coordinate qi that does not appear explicitly in L (only its velocity does) is called cyclic or ignorable. Lagrange’s equation then reduces to
d/dt (∂L / ∂q̇i) = 0 ⇒ pi = const
The conjugate momentum is a constant of the motion.
In polar coordinates the Lagrangian of a particle in a central potential V(r) is independent of θ:
L = ½ m (ṙ² + r² θ̇²) − V(r)
pθ = m r² θ̇ = ℓ (constant angular momentum)
Angular momentum conservation follows at once from the cyclic character of the azimuthal angle—an instance of Noether’s theorem for rotational invariance.
The actual trajectory of a system between two fixed configurations at times t1 and t2 renders the action stationary:
S[q] = ∫t₁t₂ L(q, q̇, t) dt , δS = 0
The stationary-path condition is precisely the set of Euler–Lagrange equations. Hamilton’s principle therefore elevates the entire content of classical mechanics to a single variational statement.
L = ½ m v². The action is proportional to the integral of kinetic energy. The path that minimizes (or stationaryizes) S is uniform rectilinear motion—Newton’s first law recovered from a variational principle.
Green: true (stationary) path. Colored curves: varied paths with the same endpoints. Only the true path extremizes the action.
The following classic systems illustrate the economy of the Lagrangian method.
θ̈ + (g/ℓ) sin θ = 0
Nonlinear; reduces to simple harmonic motion for small amplitudes.
Two angles θ1, θ2 yield a four-dimensional phase space. The Lagrangian is
L = ½ (m₁ + m₂) ℓ₁² θ̇₁² + ½ m₂ ℓ₂² θ̇₂² + m₂ ℓ₁ ℓ₂ θ̇₁ θ̇₂ cos(θ₁ − θ₂) − (m₁ + m₂) g ℓ₁ cos θ₁ − m₂ g ℓ₂ cos θ₂
The resulting equations are strongly nonlinear and exhibit chaotic behavior for a wide range of initial conditions—an iconic example of deterministic chaos in a low-dimensional mechanical system.
Interactive double-pendulum simulation. Chaos is visible in the sensitive dependence on initial angles and velocities.
L = ½ m ẋ² − ½ k x² ⇒ mẍ + kx = 0
L = ½ m (ṙ² + r² θ̇²) − V(r)
Conservation of angular momentum reduces the radial motion to an effective one-dimensional problem with potential Veff(r) = V(r) + ℓ²/(2m r²).
Lagrangian mechanics replaces the vectorial, force-centric language of Newton with a scalar, energy-based variational framework. The advantages are decisive for constrained systems, multi-body dynamics, and any setting in which generalized coordinates are natural.
The development proceeds systematically:
Beyond classical mechanics the same ideas reappear in Hamiltonian mechanics, quantum mechanics (path integrals), and classical field theory. Mastery of the Lagrangian formalism is therefore both a practical tool and a conceptual gateway to modern theoretical physics.