Generalized coordinates are variables used in classical mechanics to specify the configuration of a system. They may be distances, angles, or other parameters rather than the Cartesian coordinates of individual particles. For a system with independent configuration variables, a set gives a local description of every allowed configuration. Their principal advantage is that a suitable choice can incorporate geometric constraints directly, reducing the number of variables and simplifying the equations of motion.
Configuration and degrees of freedom
A configuration describes the positions of a system’s components, without specifying their velocities. A system’s configuration space is the set of all its allowed configurations; generalized coordinates label points in this space.
For unconstrained point particles moving in three dimensions, Cartesian coordinates are needed. Suppose the particles satisfy independent constraints of the form
These are called holonomic constraints because they can be expressed as relations involving positions and, possibly, time. Where the constraints are regular and independent, the number of configuration degrees of freedom is
A set of independent generalized coordinates then parameterizes the positions as
At a fixed time, the derivatives , considered together for all particles, must have rank . This ensures that the coordinates describe independent variations of configuration.
Generalized coordinates need not have the same dimensions. One coordinate may be a length and another an angle. They also need not be unique: different coordinate systems may describe the same physical configurations.
Although the term sometimes includes redundant variables used together with constraint equations, a minimal set of generalized coordinates consists of locally independent variables.
Examples
Plane pendulum
A point mass suspended from a fixed pivot by a massless rod of length , moving in a vertical plane, has one configuration degree of freedom. With the pivot at the origin, the vertical coordinate directed upward, and measured from the downward vertical,
The single generalized coordinate automatically enforces the constraint . There is no need to solve separately for and .
Polar coordinates
For a particle moving freely in a plane, polar coordinates can be used instead of :
Here the number of coordinates is not reduced. The benefit is that the coordinates match circular geometry or rotational symmetry.
Rigid bodies and linked mechanisms
An unconstrained rigid body in three-dimensional space has six configuration degrees of freedom: three for translation and three for orientation. Its position may be described using its center of mass, while its orientation can be described locally by three angles.
A planar double pendulum has two configuration degrees of freedom, commonly represented by the angles of its two rods. In a robotic mechanism, joint angles and sliding-joint displacements often provide natural generalized coordinates, although closed mechanical loops can impose additional constraints.
Generalized velocities
The time derivatives
are called generalized velocities. Differentiating the position mapping gives
The first term describes motion due to changing coordinates. The second accounts for explicit time dependence in the mapping, such as motion of a support or a prescribed moving surface.
Generalized velocities are coordinate rates, not necessarily Cartesian velocity components. In polar coordinates, for example, the transverse speed is , not .
For point particles with masses , the kinetic energy is
When the position mapping has no explicit time dependence, this becomes
where
The coefficients form the system’s mass matrix. They describe how mass and geometry determine the kinetic energy associated with coordinate rates. With an explicitly time-dependent mapping, kinetic energy can also contain terms linear in and terms independent of them.
Generalized forces and virtual work
A virtual displacement is an infinitesimal change of configuration consistent with the constraints while time is held fixed. For independent generalized coordinates,
If forces act on the particles, their virtual work is
where
is the generalized force associated with .
A generalized force is not necessarily an ordinary force. Its units depend on its associated coordinate: has units of energy. For a linear displacement, has units of force; for a rotation angle, it is a torque.
If the forces derive from a position-dependent potential energy , then
Constraint forces that do no virtual work on allowed displacements are called ideal. Such forces can disappear from the equations for independent generalized coordinates. This is why a pendulum’s equation can be obtained without first calculating the rod tension. Not all constraint forces are ideal, so their elimination is not automatic.
Lagrange’s equations
Generalized coordinates are central to Lagrangian mechanics. For a system described by a Lagrangian
the equations of motion are
where represents generalized forces not already included through the potential. If all forces are represented by , the right-hand side is zero.
These equations can be derived from Newton’s laws of motion using virtual work and ideal constraints, or from a stationary-action principle. In the latter formulation, the action is
and the physical trajectory makes its first variation vanish for suitable variations with fixed endpoints.
For the plane pendulum,
so Lagrange’s equation yields
The use of an angle rather than a Cartesian position does not alter the physical law; it expresses that law in variables adapted to the constraint.
Generalized momenta and Hamiltonian mechanics
The momentum conjugate to is
A generalized momentum need not equal mass times a coordinate rate. For a particle of mass in plane polar coordinates with a velocity-independent potential,
Here is the component of angular momentum perpendicular to the plane.
If does not depend explicitly on a coordinate , that coordinate is called cyclic or ignorable. In the absence of an additional generalized force in that coordinate, its conjugate momentum is conserved:
This is a coordinate expression of the connection between symmetry and conservation formalized by Noether’s theorem.
When the relation between and is invertible, a Legendre transformation defines
This leads to Hamiltonian mechanics, in which are coordinates on phase space. Configuration space and phase space are distinct: the former specifies configurations, while the latter also includes momenta.
Nonholonomic constraints
Not every restriction on motion can be incorporated by reducing the number of configuration coordinates. A nonholonomic constraint restricts possible motions without being reducible to a relation among positions and time alone. Rolling without slipping is a common source of such constraints.
A frequent form is
A velocity constraint is genuinely nonholonomic when it is not integrable into a configuration constraint. A system with such constraints may still require coordinates to specify its configuration while having fewer independently admissible instantaneous velocities. Thus configuration-space dimension and the number of independent velocity parameters need not coincide.
Nonholonomic systems generally require constrained dynamical equations, often with Lagrange multipliers or suitably chosen independent velocity variables. Substituting velocity constraints into a Lagrangian and applying unconstrained Lagrange equations is not generally valid.
Coordinate choice and limitations
A coordinate description is usually local rather than global. Polar coordinates are singular at , and common three-angle descriptions of rigid-body orientation have singular configurations. These are failures of the coordinate representation, not necessarily of the physical motion.
Redundant representations can avoid some difficulties. For example, unit quaternions describe orientation using four components subject to one normalization constraint. They do not give four independent rotational degrees of freedom.
Other practical considerations include:
- Symmetry: coordinates aligned with a symmetry can reveal conserved momenta.
- Constraint structure: coordinates that satisfy geometric constraints can remove unnecessary variables and reaction forces.
- Numerical behavior: a mathematically valid coordinate choice can become poorly conditioned near a singular configuration.
- Force recovery: eliminating constraint forces simplifies motion equations, but those forces may require a separate calculation.
- Time dependence: a moving coordinate mapping can complicate the relation between the Hamiltonian and physical energy.
Generalized coordinates specify configurations, not complete dynamical states. For regular second-order equations of motion, initial coordinate values and coordinate rates are normally both required.
Historical development and applications
The use of variables adapted to constraints grew out of the analytical mechanics of the eighteenth century. Joseph-Louis Lagrange’s Mécanique analytique, published in 1788, presented a systematic formulation in which suitable independent variables and virtual work replaced many separate Cartesian force equations.
Generalized coordinates are used in mechanism design, robotics, spacecraft attitude dynamics, vibration analysis, and molecular dynamics. In deformable-body models, a finite set of amplitudes can serve as generalized coordinates for approximate displacement patterns. In continuum theories, this finite-dimensional description extends to fields, with a coordinate-like variable associated with each spatial point.