Mechanics I: The “Principles” and Equations of Motion

Mechanics I: The “Principle” and Equations of Motion

Inertial Frames

“What is an inertial system?”

“It is a c.s. in which the laws of mechanics are valid.”

— The Evolution of Physics [1]

Here, c.s. stands for co-ordinate system.

The inertial frame of reference, or inertial frame for short, provides the setting for mechanics and is a cornerstone of classical physics. An inertial frame is a frame of reference in which space can be described as homogeneous and isotropic, and time as homogeneous. Spatial homogeneity means that no position is privileged. Spatial isotropy means that no direction is privileged. Temporal homogeneity means that no instant is privileged. In other words, for an isolated system free from outside influences, changing the location, orientation, or starting time of the experiment as a whole should leave the laws of mechanics governing it unchanged.

Galileo’s inclined-plane thought experiment can help make this abstract idea more concrete. For now, let us set aside the causes of changes in motion and simply observe what a ball does. It speeds up as it rolls downhill and slows down as it rolls uphill. How, then, would it move on a level surface, where it neither climbs nor descends? Experiments show that the ball generally slows to a stop, but it tends to roll farther on a flatter, smoother surface. The distance it rolls therefore depends on the experimental conditions. Now place two inclined planes opposite each other and join them smoothly at the bottom. A ball released from rest on one side rolls down and then climbs the other side. As the experimental conditions improve, the height it reaches approaches its starting height. This suggests that if the impediments present in the actual experiment could be removed, the ball would rise to its original height on the other side. Now keep the release height fixed and gradually reduce the angle of the second incline. To reach the same height, the ball must travel farther and farther. Once the second surface is entirely level, the ball cannot regain its original height however far it travels, so it keeps moving. The thought experiment helps us understand that in an inertial frame, a free particle unaffected by other bodies remains at rest or moves in a straight line at constant velocity.

This is Newton’s first law, which also asserts that inertial frames exist.

Consider a second frame that moves at constant velocity relative to a known inertial frame, with no relative rotation of the coordinate axes. Observers in the two frames may record different velocities for the same object, but they agree on whether its velocity remains constant. The measured velocities differ only by the constant relative velocity of the frames. If the object moves in a straight line at constant speed in one frame, its velocity also remains constant in magnitude and direction in the other. The second frame therefore satisfies the condition for inertial motion and is itself an inertial frame. By contrast, in a non-inertial frame, even an object unaffected by other bodies need not remain at rest or move in a straight line at constant velocity. Viewed from a rotating platform, for example, a free object travelling in a straight line may appear to follow a curved path. Free motion alone gives us no way to single out a privileged inertial frame. Extending this equivalence to all the laws of mechanics gives Galileo’s principle of relativity. The laws of mechanics governing the same physical process have the same form in all inertial frames.

“The same” refers to the form of the laws. Observers in different frames may record different positions and velocities.

How can we determine whether a particular frame is inertial? A single direct measurement cannot fully establish the homogeneity and isotropy described above. We must observe the motion of objects, eliminate the effects of other bodies and the apparatus as far as possible, and check whether the observations agree with the law of inertia. Every experiment is limited in duration, spatial extent, and precision, so it cannot cover every possible case. Repeated experiments can test the applicability of inertial frames, within those limits. For the problems discussed in this part of the series, we can work in inertial frames throughout.

We take the existence of inertial frames as a starting point of the theory. This is also an essential part of Newton’s first law. Although we have yet to introduce the second law, treating the first merely as a special case of the second would overlook its role in the theory. Even if the second law yields uniform straight-line motion under a particular condition, its usual form already presupposes an inertial frame. Deriving that special case therefore cannot establish the existence of inertial frames.

Newton’s first law describes free motion and asserts that there are inertial frames in which bodies move this way, providing the basis for the dynamics that follows.

The “Principle” and Equations of Motion

“The fundamental laws can be put in the form of a principle of least action.”

— The Feynman Lectures on Physics, Vol. II [2]

The principle of least action, referred to below as the principle, is widely used across physics. It offers a unified description by taking the entire trajectory over a time interval and using a quantity called the action to obtain the equations of motion. Laws of motion with different forms can thus be expressed within a single variational framework, illustrating the search for simplicity and unity in physics. In this article, we will approach mechanics through this principle.

“Least” is the traditional term. With the initial and final times and endpoint configurations fixed, any infinitesimal variation of the actual trajectory leaves the action unchanged to first order. The action is then said to be stationary.

The action is defined by

S=:∫t1t2Ldt

In other words, the action is the integral of the Lagrangian L over the given time interval. We write the Lagrangian as L(q,q˙,t) , where q denotes the generalised coordinates, q˙ the generalised velocities, and t time. Why is it enough to stop at velocity? Knowing where an object is now does not tell us what it will do next, since it can pass through the same position with different velocities. For the classical systems considered here, experience shows that once all the generalised coordinates and velocities are specified at one instant, the law of motion determines the subsequent motion. The same law determines the acceleration. In mathematical terms, the motion is governed by second-order differential equations. The variational derivation below shows that when a Lagrangian depends only on the coordinates and their first time derivatives, the resulting equations of motion contain at most second time derivatives of the coordinates.

What happens if we include acceleration in the Lagrangian? If the dependence on acceleration is nondegenerate, variation yields fourth-order equations of motion. To determine a trajectory, we must then specify the initial acceleration and its rate of change independently, in addition to the initial position and velocity [3]. The theory would therefore allow different trajectories with identical initial positions and velocities, depending on the values of those additional initial data. This conflicts with the observed behaviour of the systems considered here, so we use L(q,q˙,t) to describe their dynamics.

Now fix the initial and final times and the generalised coordinates at those times. Infinitely many candidate trajectories satisfy these conditions. Even two trajectories that follow the same path through space correspond to different functions q(t) if they trace it at different speeds. Substituting each candidate trajectory into the integral gives a value of the action. We therefore write the action as S[q] and call it a functional of the trajectory. What condition does the actual trajectory satisfy? According to the principle, if we hold the endpoint times and coordinates fixed and vary the actual trajectory infinitesimally, the first-order change in the action vanishes. The stationarity condition is

δS=0.

To calculate the variation of the action, we impose δt=0 and the endpoint conditions δq|t1=δq|t2=0 .

δ∫t1t2L(q,q˙,t)dt=∫t1t2δL(q,q˙,t)dt=∫t1t2∑α=1n(∂L∂qαδqα+∂L∂qα˙δqα˙+∂L∂tδt)dt=∫t1t2∑α=1n(∂L∂qαδqα+∂L∂qα˙δqα˙)dt

Integration by parts produces a total time derivative. Integrating it gives a boundary term that vanishes under the endpoint conditions.

∫t1t2∑α=1n(∂L∂qαδqα+∂L∂qα˙δqα˙)dt=∫t1t2∑α=1n[∂L∂qαδqα+ddt(∂L∂qα˙δqα)−ddt(∂L∂qα˙)δqα]dt=∫t1t2∑α=1n[∂L∂qαδqα−ddt(∂L∂qα˙)δqα]dt=∫t1t2∑α=1n[(∂L∂qα−ddt∂L∂qα˙)δqα]dt=0

The variations δqα are independent. For the equality to hold for all allowed variations, we must have

∂L∂qα−ddt∂L∂qα˙=0

This is the Euler–Lagrange equation. It provides a general method for obtaining the equations of motion from a Lagrangian. To apply it to a particular system, we still need to determine that system’s Lagrangian. Let us start with a free particle in an inertial frame. What quantities can its Lagrangian L depend on? Write its Cartesian position as q and its velocity as q˙ . Because time is homogeneous, releasing the particle at different times from the same position with the same velocity gives motion governed by the same law. Because space is homogeneous, changing the initial position while keeping the initial velocity fixed simply translates the trajectory. The starting time and position therefore make no difference to the law of free motion. Accordingly, we take the Lagrangian to depend only on velocity.

L=L(q˙)

Before determining its form, let us return to the variation of the action. The total time derivative produced by integration by parts can be integrated directly, leaving a term evaluated only at the endpoints. In deriving the Euler–Lagrange equation, this boundary term vanishes because the variations at the endpoints are zero. This suggests that we obtain the same equations of motion whenever we modify the action in a way that preserves its stationarity condition. One simple example is multiplying the entire Lagrangian by a nonzero constant k . Under L→kL , the variation of the action is multiplied by the same constant. Hence kδS=0 and δS=0 give the same equations of motion. Now let ϕ(q,t) be a sufficiently smooth function and add its total time derivative dϕ/dt to the Lagrangian. On integration, the change in the action depends only on the endpoint values of ϕ .

S⟶S+[ϕ(q,t)]t1t2=S+ϕ(q(t2),t2)−ϕ(q(t1),t1).

The initial and final times are fixed, and δq|t1=δq|t2=0 , so all the trajectories under comparison have the same endpoints. The added boundary term is consequently the same for each trajectory, and its variation vanishes.

L⇔L+ddtϕ(q,t)

The two Lagrangians give the same equations of motion and are equivalent as descriptions of the system’s dynamics.

Given that different Lagrangians can yield the same equations of motion, how should we choose one for a free particle? We have used the homogeneity of time and space to write it as L(q˙) . We now use Galileo’s principle of relativity to determine how it depends on velocity. Consider two inertial frames K and K′ whose axes are aligned. Suppose the transformation from K to K′ changes the particle’s velocity from q˙ to q˙+ϵ , where ϵ is a constant infinitesimal vector of arbitrary direction. We require that the transformation change the Lagrangian by at most a total time derivative. The action then changes only by a boundary term, so the stationarity condition and equations of motion are preserved.

L(q˙+ϵ)−L(q˙)=ddtϕ(q,t)

Expanding the first term on the left in powers of ϵ gives

L(q˙+ϵ)−L(q˙)=L(q˙)+∇L(q˙)Tϵ+12!ϵTH(q˙)ϵ+...−L(q˙)=∇L(q˙)Tϵ+12!ϵTH(q˙)ϵ+...=ddtϕ(q,t)

The right-hand side is a total time derivative and is therefore at most first degree in velocity. Comparing the terms of first order in ϵ shows that ∇LTϵ is also at most first degree in velocity. Since ϵ can point along any coordinate axis, the partial derivative of L with respect to each velocity component is at most first degree in the full velocity vector. Hence L is a polynomial of degree at most two in velocity.

L(q˙)=12q˙TMq˙+Bq˙+C

Here M is a constant symmetric matrix, B a constant row vector, and C a scalar constant. The last two terms together form a total time derivative.

Bq˙+C=ddt(Bq+Ct)

Dropping this term leaves the equations of motion unchanged. The free-particle Lagrangian therefore reduces to

L(q˙)=12q˙TMq˙

In Cartesian coordinates, spatial isotropy requires this quadratic form to have the same value for all velocities of equal magnitude. Evaluate it at velocities of the same magnitude along the eigenvectors of M . The resulting values can agree only if all the eigenvalues are equal. Thus M=mI , giving

L=T=12mr˙2

The coefficient m is the particle’s inertial mass. The expression above is the familiar kinetic energy, denoted by T . With the free-particle Lagrangian in hand, we write the particle’s position vector as r and substitute the Lagrangian into the equation of motion.

∂L∂qα−ddt∂L∂qα˙=0ddt(mr˙)=0

The particle’s mass m is constant, so its velocity vector r˙ is constant as well. This agrees with our earlier description of inertial motion. The equation also identifies a conserved quantity. To extend the idea to generalised coordinates, we define the conjugate momentum pα corresponding to qα by

pα:=∂L∂qα˙

In the Cartesian coordinates x , y , and z of a free particle, the conjugate momenta are mx˙ , my˙ , and mz˙ , respectively. These are the three components of the particle’s momentum p=mr˙ . The equation of motion shows that p stays constant throughout the motion. This is conservation of momentum for a free particle.

p=mr˙=const.

Why is momentum conserved? In terms of conjugate momentum, the Euler–Lagrange equation reads

dpαdt=∂L∂qα

If the Lagrangian does not explicitly depend on a coordinate qα , the right-hand side vanishes and the corresponding conjugate momentum remains constant. A translation of the free particle in any Cartesian direction leaves its Lagrangian unchanged, so all three components of its momentum are conserved. The conservation of momentum for a free particle is thus linked to spatial translation symmetry. Does this connection extend to other conservation laws? In the next article, we will use this question to explore the relationship between symmetries and conservation laws.

Having considered a free particle in an inertial frame, let us now see what happens when a particle is subject to an interaction. Take a fixed source as an example. Its effect on the particle may differ from one position to another. Moving the particle while leaving the source in place may therefore change the particle’s subsequent motion. For interactions that can be described by a function of position, we subtract a function V(q) , called the potential energy, from the free-particle Lagrangian.

L=T−V=12q˙TMq˙−V(q)

In Cartesian coordinates, the negative gradient of the potential energy with respect to position gives the force on the particle.

F=−∇V

The minus sign indicates that the force points in the direction of steepest decrease in potential energy. A force that can be written as the negative gradient of a time-independent potential energy is called conservative. For a single particle of constant mass, substituting L=T−V into the Euler–Lagrange equation gives F=mr¨ . This is the form Newton’s second law takes here. More generally, we can collect the effects of the interactions on the right-hand side of the equation and use the generalised force Fα to express their effect on the coordinate qα .

ddt∂T∂q˙α−∂T∂qα=Fα

In Cartesian coordinates, Fα is the corresponding component of the ordinary force. In other coordinates, its meaning depends on the choice of coordinates. If the interaction is described by a position-dependent potential energy, then Fα=−∂V/∂qα . In other cases, Fα is determined by the particular interaction.

Some velocity-dependent interactions can be included directly in the Lagrangian. Even a term linear in velocity can affect the equations of motion. If the term can be written as the total time derivative of a function, its integral changes only the boundary term in the action, so the term may be omitted.

Does our initial focus on interactions that can be represented by potential energy reveal a limit of the principle, given how often nonconservative forces occur in practice?

We can address this question by broadening our view from the objects we directly observe to the entire interacting system. Modern physics describes the fundamental interactions in nature in terms of gravity, electromagnetism, and the strong and weak interactions. In a particular mechanics problem, we generally consider only a few variables directly relevant to the motion under study. Energy can be transferred to degrees of freedom outside the model through microscopic motion or electromagnetic radiation, for example. The observed motion then exhibits dissipation, which can be described using nonconservative forces. The appearance of these forces is therefore related to how we define the system’s boundaries. The principle can be used to construct a more complete description of the dynamics, while nonconservative forces can account for processes that the simpler model leaves implicit.

Finally, consider an isolated system of two particles in an inertial frame. Suppose their interaction potential V depends only on their relative position r1−r2 . Adding the Euler–Lagrange equations for the two particles gives the rate of change of their total momentum.

ddt(p1+p2)=F1+F2

Since r1 and r2 appear with opposite signs in the relative position, the chain rule gives ∇r1V=−∇r2V . It follows that

F1+F2=−∇r1V−∇r2V=0

The particles’ total momentum is therefore conserved, and the forces they exert on each other are equal in magnitude and opposite in direction. This is Newton’s third law.

References

[1]
A. Einstein and L. Infeld, The evolution of physics: from early concepts to relativity an quanta. New York: Simon and Schuster, 1966.
[2]
R. P. Feynman, R. B. Leighton, and M. L. Sands, The Feynman lectures on physics, New millennium ed. New York: Basic Books, 2011.
[3]
R. P. Woodard, “The Theorem of Ostrogradsky.” Accessed: Nov. 12, 2026. [Online]. Available: http://arxiv.org/abs/1506.02210

Mechanics I: The “Principles” and Equations of Motion
https://l-lllll-06.github.io/en/mechanics-01/
Author
-L_LLLLL-
Posted on
November 13, 2026
Licensed under