Frame Shift · Guide · glidepathscience.org
GuideFrame Shift
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a textbook would take them. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about fifteen minutes to read. The worked examples follow a cart of their own, never one of the sim's problems.
Part 1 · The essentials
Motion is always measured from somewhere
Ask whether a parked car is moving and almost everyone says no. Ask a passenger in a train rolling past and the answer changes: from the window, the car slides backward. Both answers are correct, because a velocity is never a property of an object alone. It is the rate at which the object's position changes, and a position has to be measured from something: a mark on the ground, a seat on the train, a point on a ruler. The thing we measure from, together with its rulers and clocks, is called a frame of reference.
This sim follows two observers. Observer A stands beside a level, frictionless track and measures everything with a ruler fixed to the track. Observer B rides past on a platform at a steady velocity u relative to the track. Both take right as positive. Because B moves at constant velocity, B is an inertial observer, just as A is: neither feels pushed, and a ball set down on B's platform stays put. For two such observers, a velocity measured by B is the velocity A measures minus B's own velocity:
$v_B = v_A - u$
Here $v_A$ is the object's velocity in A's frame, $v_B$ is its velocity in B's frame and u is B's velocity relative to A. A cart sitting still on the track has $v_A = 0$, so B sees it moving at $-u$: to a rider passing at 3.5 m/s, the "resting" cart moves left at 3.5 m/s.
The usual mistake is to treat "at rest" as an absolute fact about the cart. It always means at rest relative to something, here the track; the cart has a definite velocity in every frame, and the frames disagree about its value. Nor does B see the cart moving B's way: B passes the cart, so in B's frame the cart falls behind.
If observers disagree about velocities, what can they agree on? Quite a lot, as it turns out, and the shared quantities are where any comparison has to start.
What every inertial observer shares
A and B read the same mass off the cart and the same force off the gauge between the string and the cart, and their clocks agree on how long the force acts, Δt. Mass and force do not depend on who is watching, and time intervals are the same for observers whose relative speed is far below the speed of light (Part 2 returns to that condition).
Since B's velocities are A's minus a constant, B's velocity changes exactly as A's does. Subtracting the same u at the start and at the end leaves the change untouched, so both observers measure the same change in velocity Δv and the same acceleration. Newton's second law then gives the acceleration from the shared force and mass alone:
$a = F/m$
Here F is the net force on the cart, in newtons, and m its mass, in kilograms. With a constant force the acceleration is constant, and the two constant-acceleration relations give the final velocity and the displacement from an initial velocity $v_i$:
$v_f = v_i + a\Delta t$ $\Delta x = v_i\Delta t + \frac{1}{2}a\Delta t^2$
On a velocity–time graph, the two observers' lines have the same slope: they are parallel, one shifted from the other by u, and they never cross. A common slip is to think that B, being in motion, sees a steeper or shallower line. Another is to multiply by the mass instead of dividing, or to stop at FΔt, which is the impulse in newton-seconds, not a velocity.
A's numbers come from the frame the problem was written in, which is why they feel like the "real" ones. The next step is to translate them into B's frame.
Changing frames: velocities and rulers
The velocity rule from the first section applies at every instant, to the initial and the final velocity alike. Displacements need one more idea. A displacement is read off a ruler: where the object ends up minus where it started, both measured on the same ruler. A's ruler is fixed to the track. B's ruler rides along with B, so during the run it slides a distance uΔt to the right. The cart's displacement on B's ruler is its displacement on A's ruler minus how far B's ruler moved:
$v_B = v_A - u$ $\Delta x_B = \Delta x_A - u\Delta t$
This one subtraction is behind almost every disagreement in the rest of the sim. If B rides faster than the cart's average velocity, B's ruler outruns the cart, and on B's ruler the cart ends up behind where it started, with a negative displacement, even though it is moving forward along the track at the end. That surprises nearly everyone at first. The usual errors are to use A's displacement for B, or to add uΔt instead of subtracting it.
Different velocities and different displacements lead straight to different energies, which is where the real surprise lies.
Kinetic energy depends on the frame
Kinetic energy is the energy an object has because it moves. For a cart of mass m moving at velocity v it is
$K = \frac{1}{2}mv^2$ $\Delta K = K_f - K_i$
Each observer must use the velocities measured in their own frame. Because velocities differ between frames, so do kinetic energies: the cart at rest on the track has no kinetic energy for A and $\frac{1}{2}mu^2$ for B. Squaring removes the sign, so K is never negative, whichever way the cart moves.
What about the change in kinetic energy? Here intuition usually goes wrong. Since B's velocities are A's shifted by a constant, it seems that B's energies should also be shifted by a constant that cancels in ΔK. That argument works for any quantity linear in v: Δv and the change in momentum, Δp = mΔv, are the same for both observers. It fails for K because K is quadratic. Expanding B's kinetic energy shows why:
$\frac{1}{2}m(v - u)^2 = \frac{1}{2}mv^2 - muv + \frac{1}{2}mu^2$
The last term is the same at the start and the end and cancels. The middle term, the cross term $-muv$, does not: it changes by $-mu\Delta v$ as the velocity changes. A related trap is to compute $\frac{1}{2}m(\Delta v)^2$, which equals ΔK only for an object that starts at rest. For anyone else, square each velocity first and then subtract.
Two observers, two values of ΔK. If energy bookkeeping is to survive, the work done on the cart must also change from frame to frame, and by the same amount.
Work follows the ruler
The work a constant force does on an object is the force times the object's displacement along it. For our cart, with the force along the track,
$W = F\Delta x = \Delta K$
The first equality defines the work; the second is the work–energy theorem: the net work done on an object equals the change in its kinetic energy. Both observers agree on F. They disagree on Δx, because each reads it off their own ruler, so they disagree on W. A force that points right the whole time does negative work in any frame where the cart's displacement points left. No friction or reversed push is needed, only a moving ruler.
The striking result is that the theorem holds for each observer with their own numbers. It does not promise that everyone gets the same value of W; it promises that within any one inertial frame, the work and the change in kinetic energy agree. The tempting errors are to use A's displacement in B's work, or to use FΔt, the impulse, as if it were the work. FΔt is the change in momentum, which every observer agrees on; the work is not.
A negative total can hide a positive part. To see how, follow the rider at 5.0 m/s through the run, instant by instant.
The turnaround, power and the zero-work rider
Power is the rate at which a force does work. For a force F on an object moving at v along it, the power is
$P = Fv$
and it carries the sign of v. For a rider faster than the cart, the cart starts out moving backward and the force does negative work; the cart slows, stops for an instant, then moves forward, and the force does positive work. The moment of rest, the turnaround, is when the cart keeps pace with the rider, its velocity on the track equal to u. In the rider's frame,
$v_B = 0$ at $t = (u - v_i)/a$
where $v_i$ is the cart's initial velocity on the track. At that instant the power is zero, and on a graph of power against time the curve crosses the axis there. The work up to any moment is the area under the power curve: negative below the axis, positive above. Two errors are common: inverting the ratio to a/u, and taking the turnaround to happen when the force switches off.
One rider sees zero net work over the whole run. Zero work means zero displacement on that rider's ruler, so the rider must cover the same distance as the cart: they ride at the cart's average velocity. With a constant force, that is halfway between the cart's initial and final velocities:
$u = \Delta x_A/\Delta t = (v_i + v_f)/2$
This rider does not see a stationary cart. The cart starts moving backward, stops halfway through and ends moving forward at the speed it started with. Equal speeds at both ends mean equal kinetic energies, so ΔK = 0 and W = 0, with plenty of motion in between. The rider who keeps pace with the cart at the end is a different person: that one sees the cart end at rest, having lost all its kinetic energy.
Every observer's books balance. Yet the winch that pulls the cart runs on a battery, and the battery cannot spend different amounts of energy for different observers. So where does the energy go?
Who pays? The battery, the track and the Earth
Something has to supply the force: here a battery-driven winch bolted to the far end of the track, winding in a string tied to the cart. The battery's energy goes into winding string onto the drum, so it spends the force times the length of string wound in. That length is the change in the distance between the winch and the cart, and a distance between two objects measured at one instant is the same for every observer. So every observer agrees on the battery's bill:
battery $= F \times$ (string wound in) $= F\Delta x_A$
The string wound in equals the cart's displacement in A's frame, because the winch is fixed to the track. For A the battery's energy becomes the cart's kinetic energy. For a rider at u the cart may gain far less, or even lose energy, while the battery spends the same amount. The rest goes to the track, the winch and the Earth they are bolted to. The string pulls the winch toward the cart, and in the rider's frame the whole Earth is already moving at −u, the same way, so that pull does work on it. The Earth's velocity changes by far too little to measure, but its kinetic energy changes by uFΔt. Energy is conserved in every inertial frame; each observer has to include everything the winch pushes on.
Two mistakes are tempting: to say energy is not conserved for a moving observer, and to blame heat in the winch. The winch here is ideal, with no losses, and the frame argument would be the same with them.
Part 2 derives these results in general: for any force history, for any observer, and for observers who accelerate.
Part 2 · Going further
The Galilean transformation
The rules of Part 1 come from one pair of equations. Let observer B move at constant velocity u relative to A, with their origins together at t = 0. An event that A places at position x and time t, B places at
$x' = x - ut$ $t' = t$
This is the Galilean transformation; primes mark B's coordinates. The second equation says the clocks agree. Differentiating the first with respect to time, once and then twice, gives
$v' = v - u$ $a' = a$
A constant u drops out of the second derivative. Mass and force are the same for both observers, so if F = ma holds in A's frame, it holds in B's with the same F, m and a. Newton's laws are invariant under a Galilean change of frame: no experiment in mechanics can tell a uniformly moving laboratory from one at rest. Velocity, position and everything built from them are not invariant, and the rest of this part follows them.
Time intervals are invariant too, and so is the distance between two points measured at one instant: both points shift by the same ut. A common slip is to think a moving observer needs an extra force to make F = ma work. Only an accelerating observer does, as a later section shows.
With the transformation in hand, the frame dependence of kinetic energy takes one line of algebra.
How kinetic energy transforms
Write B's kinetic energy using $v' = v - u$ and expand. With p = mv, the momentum in A's frame,
$K' = \frac{1}{2}m(v - u)^2 = K - up + \frac{1}{2}mu^2$
Over any interval the last term is constant and cancels, and the middle term changes by $-u\Delta p$. So the change in kinetic energy in B's frame is
$\Delta K' = \Delta K - u\Delta p$
where Δp is the change in momentum, the same in both frames. An observer moving with the force (u and Δp the same sign) sees less kinetic energy gained; one moving against it sees more. The formula is exact, for any u, and it needs nothing about how the momentum changed. The common mistake is the sign of u: an observer riding left has u < 0, and then $-u\Delta p$ adds energy. Another is to keep the $\frac{1}{2}mu^2$ term in a change, where it cancels.
The work must shift by exactly the same amount, or the work–energy theorem would hold in only one frame. It does.
How work and power transform, for any force history
For a constant force, B's displacement is $\Delta x' = \Delta x - u\Delta t$, so B's work is $F\Delta x' = W - u(F\Delta t)$. The product FΔt is the impulse J. In general the impulse is the integral of the force over time, and by Newton's second law it equals the change in momentum, the impulse–momentum theorem:
$J = \int F\,dt = \Delta p$ $W' = W - uJ$
The work shifts by $-uJ$ and the kinetic energy by $-u\Delta p$. Since J = Δp in every inertial frame, the shifts are identical, and W = ΔK in one inertial frame implies it in all of them. Work is not invariant, and neither is kinetic energy; their relation is.
Nothing here needs a constant force. At every instant B's power is the force times the velocity in B's frame,
$P' = F(v - u) = P - Fu$ $P'(t) = F(v_A(t) - u)$
so the two power curves are always Fu apart, and integrating, $W' - W = -u\int F\,dt = -uJ$, whatever the shape of the force history. For a constant force B's power is a straight line in time: negative while the cart moves backward in B's frame, zero at the turnaround, positive after, and the net work is the signed area of the two triangles, when the turnaround falls inside the run. The common errors are to expect the answer to depend on how the force rose and fell, and to use the track-frame velocity in B's power.
So far we have compared two observers at a time. Plotting every observer at once shows the structure behind all of these numbers.
Every observer on one line
Think of u as a variable and plot each observer's account of the same run against it. The kinetic energies at the start and the end, $K_i(u) = \frac{1}{2}m(v_i - u)^2$ and $K_f(u) = \frac{1}{2}m(v_f - u)^2$, are parabolas in u with the same $\frac{1}{2}mu^2$ term. Their difference, the net work each observer finds, is therefore a straight line:
$W(u) = K_f(u) - K_i(u) = W_A - u\Delta p$
Its slope is −Δp, in joules per meter per second (that is, kg·m/s), and it crosses zero at the zero-work observer:
$u^∗ = W_A/\Delta p = (v_i + v_f)/2$
The second form holds for any force history, since $W_A = \frac{1}{2}m(v_f^2 - v_i^2)$ and $\Delta p = m(v_f - v_i)$. For a constant force it is also the cart's average velocity on the track, $u^∗ = \Delta x_A/\Delta t$. Equivalently, in any frame $\Delta K = \Delta p\cdot(v_i' + v_f')/2$. The zero-work observer is not the cart's rest frame: no inertial observer keeps pace with an accelerating cart. The usual errors are a positive slope and taking $u^∗$ to be the final velocity.
The cart alone is an open system: the string reaches in and does work on it. Close the system and the frame dependence disappears.
Isolated systems
Apply the kinetic-energy rule to each part of a system and add. Each part's ΔK shifts by −u times its own Δp, so the total shifts by −u times the total change in momentum:
$\Delta K_{total}' = \Delta K_{total} - u\Delta p_{total}$
With no net external force, total momentum is conserved, $\Delta p_{total} = 0$, and the total change in kinetic energy is the same in every inertial frame. So "elastic" is a frame-free statement: a collision that conserves kinetic energy for one inertial observer conserves it for all, and an inelastic one loses the same amount for everyone. The tempting error is to shift one body and forget its partner, whose Δp is equal and opposite.
The cart-and-winch system becomes isolated once the track and the Earth are inside the boundary, as in Part 1.
The Earth's share and the invariant battery
Put the winch, the track and the Earth, with combined mass M, into the system. The string gives them an impulse −FΔt. In a frame moving at u they start at −u, so their change in kinetic energy is
$\Delta K_E = \frac{1}{2}M\left(u + F\Delta t/M\right)^2 - \frac{1}{2}Mu^2 = uF\Delta t + (F\Delta t)^2/2M$
The first term is u times the impulse; the second is A's value, a tiny recoil energy. Their ratio is $F\Delta t/2Mu$, so the second is negligible whenever u is much larger than $F\Delta t/2M$, which for the Earth, M = 5.97 × 10²⁴ kg, and the impulses here is below 10⁻²⁴ m/s:
$\Delta K_E = uF\Delta t + (F\Delta t)^2/2M \approx uF\Delta t$
The Earth's change in velocity is far too small to see, yet its change in kinetic energy is not, because $\Delta K \approx p\cdot\Delta v$ and in this frame the Earth's momentum is enormous. With u > 0 the string pulls the Earth the way it already moves, so its kinetic energy rises; the sign is the easy thing to lose.
The battery's share comes from the internal force pair between the winch and the cart. The total work of an attracting pair is the force times the decrease in the bodies' separation, the string wound in, measured at single instants, so every observer agrees on it. This assumes an ideal winch and a massless, unstretchable string; when the string pays out, an ideal winch works as a regenerative brake and charges its battery.
Every observer so far has been inertial. An observer who speeds up breaks the pattern, and repairing it introduces a new kind of force.
An accelerating observer
Let B start at velocity u and speed up at a constant $a_B$. Then $x' = x - ut - \frac{1}{2}a_Bt^2$, and in B's frame the cart's acceleration is $a - a_B$. Newton's second law fails unless B adds a fictitious force $-ma_B$ on the cart, a force with no agent that exists only because the frame accelerates. The work–energy theorem needs its work too:
$W_{fict} = -ma_B\Delta x'$ $W_{real} + W_{fict} = \Delta K'$
Here $\Delta x'$ is the cart's displacement in B's frame, including the $\frac{1}{2}a_B\Delta t^2$ that B's own acceleration adds to the ruler's motion. The fictitious force points opposite to B's acceleration. Leaving out the fictitious work makes the theorem appear to fail by exactly that work; leaving out the $\frac{1}{2}a_B\Delta t^2$ term gives the wrong displacement, and the balance fails too. In the sim the fictitious force is drawn dashed, the library's mark for a force with no agent.
One assumption has run underneath everything: that clocks agree, t′ = t. The last section asks how good that is.
The relativistic edge
The Galilean transformation is the low-speed limit of the Lorentz transformation of special relativity. Near the speed of light, c = 299,792,458 m/s, moving clocks run slow and simultaneity depends on the frame. The kinetic energy becomes $(\gamma - 1)mc^2$, with $\gamma = 1/\sqrt{1 - v^2/c^2}$. For v much less than c this expands to $\frac{1}{2}mv^2\left(1 + \frac{3}{4}v^2/c^2 + …\right)$: the Galilean results are the leading term, and the corrections are of relative size $(v/c)^2$, with v the observer's speed or the cart's. The common mistakes are to take the corrections to be of order v/c, or the Galilean rules to be exact below some speed: they are an approximation at every speed, a superb one at everyday speeds.
Reference
Common mistakes
- "At rest" as a fact about the object. At rest means at rest relative to something.
- Adding u instead of subtracting it. $v_B = v_A - u$ and $\Delta x_B = \Delta x_A - u\Delta t$.
- Using one observer's ruler for another. Each observer's work uses their own displacement.
- Letting the offset cancel in ΔK. It cancels in Δv and Δp, not in K, which is quadratic.
- Computing ½m(Δv)². Square each velocity first; the shortcut works only from rest.
- Using the impulse as the work. FΔt is Δp, the same for everyone; FΔx is the work.
- Mixing frames in one balance. Never compare A's kinetic energy with B's work.
- Losing energy in a moving frame. Include the track and the Earth: their ΔK is uFΔt.
- Forgetting the fictitious work. An accelerating observer needs $-ma_B$ and its work.
Equation sheet
| Part 1 | |
|---|---|
| shared | $a = F/m$ (the same for every observer) |
| motion | $v_f = v_i + a\Delta t$, $\Delta x = v_i\Delta t + \frac{1}{2}a\Delta t^2$ |
| change frames | $v_B = v_A - u$, $\Delta x_B = \Delta x_A - u\Delta t$ |
| energy | $K = \frac{1}{2}mv^2$, $\Delta K = K_f - K_i$ |
| work | $W = F\Delta x = \Delta K$ (in each frame); $P = Fv$ |
| turnaround | $v_B = 0$ in B's frame |
| zero-work rider | $u = \Delta x_A/\Delta t$ |
| battery | F × string wound in (the same for all) |
| Part 2 | |
|---|---|
| Galilean | $x' = x - ut$, $v' = v - u$, $a' = a$, $t' = t$ |
| energy | $K' = K - up + \frac{1}{2}mu^2$; $\Delta K' = \Delta K - u\Delta p$ |
| impulse, work | $J = \int F\,dt = \Delta p$; $W' = W - uJ$ |
| power | $P' = F(v - u) = P - Fu$ |
| every observer | $W(u) = W_A - u\Delta p$; $u^∗ = (v_i + v_f)/2$ (constant F: $u^∗ = \Delta x_A/\Delta t$) |
| Earth | $\Delta K_E = uF\Delta t + (F\Delta t)^2/2M \approx uF\Delta t$ |
| accelerating B | $W_{fict} = -ma_B\Delta x'$; $W_{real} + W_{fict} = \Delta K'$ |
Symbols: m mass (kg) · F force (N) · Δt its duration (s) · $v_i$, $v_f$ initial and final velocities · u B's velocity relative to the track (right +) · $a_B$ B's acceleration · K, W energy and work (J) · P power (W) · p momentum · J impulse (N·s) · M the track and Earth · primes, subscript B: B's frame; subscript A: the track's.
Further reading
D. Kleppner and R. Kolenkow, An Introduction to Mechanics, 2nd ed. (Cambridge University Press, 2014), on Galilean transformations, work and energy, and noninertial frames; OpenStax, University Physics Volume 1, §4.5 (relative motion), chapter 7 (work and kinetic energy) and chapter 9 (impulse and collisions). The Earth's mass is from NASA's Earth Fact Sheet. The Field Manual lists the model's assumptions.
Frame Shift v1.4 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org