Relativistic Momentum

Every particle accelerator on Earth is a machine for exploiting $\vec{p} = \gamma m\vec{v}$, and every beamline engineer works in its two regimes at once. A synchrotron bends its beam with magnets — a transverse force — and the magnetic rigidity $B\rho = p/q$, the field-times-radius needed to hold a particle on a curve, depends on the momentum and nothing else. This is why beams are specified by momentum in GeV/$c$ rather than by speed. The speed of every proton in the LHC is $0.99999999c$-and-change, a number that distinguishes nothing; the momentum ranges over four orders of magnitude and determines every magnet in the ring. The longitudinal side is where the $\gamma^3$ bites. A linac's cavities push along the motion, and the response is $a_\parallel = F/(\gamma^3 m)$: the same gradient buys $\gamma^3$ times less acceleration as the beam speeds up. Past $\gamma \approx 10$ the particle is no longer meaningfully getting faster, and the entire energy input goes into momentum rather than speed. A 10 GeV electron and a 50 GeV electron cross a laboratory in times that differ by parts per billion — and hit a target with five times the punch.

The trouble is easy to see and impossible to argue away. Take two particles that collide, and write $\sum m\vec{v}$ before and after in the laboratory. Suppose it balances. Now boost to a frame moving at $w$ and recompute — using the relativistic velocity-addition law, because that is how velocities actually transform. The sum no longer balances. Nothing has changed about the collision; only the observer has moved. A Newtonian would find this baffling, and would be right to: under a Galilean boost, $\vec{v} \to \vec{v} - \vec{w}$, so $\sum m\vec{v} \to \sum m\vec{v} - \vec{w}\sum m$, and momentum conservation survives provided mass is conserved. Galilean relativity gets momentum conservation for free. Lorentz boosts do not transform velocities linearly — the composition law is a ratio, not a difference — and the free lunch is over. So: what is conserved? Write the candidate as $\vec{p} = f(v)\,m\vec{v}$ with $f$ unknown, and let a symmetric collision tell you what $f$ has to be. The construction below is due to Lewis and Tolman, and it gives a uniqueness answer rather than a consistency answer: it does not check that $\gamma$ works, it computes $\gamma$.

What breaks the symmetry in the laboratory is not the throwing but the transforming: B's transverse velocity, measured in the lab, is not $u$ but $u/\gamma_w$, because her clock runs slow and the transverse distance she covers per unit of lab time is reduced to match. Conservation in the lab therefore compares $f(u)\cdot u$ against $f(v_B)\cdot u/\gamma_w$, and those are equal only if $f$ knows about $\gamma$. Then comes the move that makes it a proof. Let the transverse speed $u$ go to zero. The collision becomes a grazing one, $v_B \to w$, and $f(u) \to f(0)$, which we have already fixed to be 1 by demanding the Newtonian limit. What is left is $1 = f(w)/\gamma_w$, and $w$ was an arbitrary boost speed. So $f$ is determined at every speed at once. No family of solutions, no undetermined constant: $f = \gamma$, uniquely.

The momentum itself, with the mass that appears in it named precisely:

where: $m$ — the invariant rest mass, a Lorentz scalar, identical in every frame and at every speed; $\vec{v} = d\vec{x}/dt$ — the ordinary coordinate velocity; $\gamma$ — the Lorentz factor of the particle's speed. Physical reading: one equation, one mass. The temptation to read $\gamma m$ as "the mass at speed $v$" is dismantled below; $m$ is a property of the particle, $\gamma$ is a property of its state of motion in your frame, and merging them destroys a scalar to manufacture a component. Momentum is not a free-standing object. It is the spatial part of a four-vector, which is why it needed the $\gamma$:

where: $U^\mu = dx^\mu/d\tau$ — the four-velocity, whose fixed norm $U\cdot U = +c^2$ is derived in Four-Vectors; $\tau$ — proper time. Physical reading: $m$ multiplies a four-vector by a scalar, so $p^\mu$ is a four-vector too, and its norm is fixed. In our signature that norm is $+m^2c^2$, not $-m^2c^2$ — the sign flips in $(-,+,+,+)$, and a formula copied without its metric is a formula copied without its meaning. Unpacking $p\cdot p = +m^2c^2$ into components gives the single-particle energy–momentum relation $E^2 = (pc)^2 + (mc^2)^2$; that unpacking is Four-Momentum's to perform, and we simply use the result. Dividing the spatial slot by the time slot eliminates $\gamma$ and $m$ together:

where: $E = \gamma mc^2$ — the total energy. Physical reading: $\gamma m$ cancels, so this relation carries no mass at all. The momentum of anything is its energy times its speed in units of $c$ — which is why it keeps working in the one place where $\gamma m\vec v$ collapses: at $\beta = 1$, where it reads $p = E/c$. That is the massless case, and equation (3) is how you get there honestly.

Step 5 is the article's owned result, but Step 6 is the one students skip. The failure of $m\vec{v}$ is not a rounding error that relativity tidies up: written out, the Newtonian ledger asserts $\gamma_w = 1$. Relativistic momentum is not a refinement of Newtonian momentum. It is the repair of a law that, taken at face value, denies the existence of other inertial frames.

There is an old and still-circulating way to describe that diverging curve: say the mass grows, write $m_{\text{rel}} = \gamma m$, and keep Newton's $\vec{p} = m_{\text{rel}}\vec{v}$. It is seductive, it reproduces equation (1) by construction, and it is a trap. Not because it gives wrong numbers for the momentum — it does not — but because of what happens the moment you use it for anything else.

That last step is the honest reason the concept was abandoned, and it is stronger than the usual complaint. The usual complaint is that $\gamma m$ is not a Lorentz scalar — true, and damning enough: it is a frame-dependent component of a four-vector wearing the costume of an invariant. But the fatal defect is operational. The whole purpose of a velocity-dependent mass was to rescue $F = ma$, and it does not rescue $F = ma$: there is no mass — not $\gamma m$, not $\gamma^3 m$, not anything — for which $\vec{F} = m_{\!*}\vec{a}$ holds in general, because $\vec{F}$ and $\vec{a}$ are not parallel and no scalar can fix that. So relativity keeps $m$, the invariant, and gives up $F = ma$ in favour of $\vec{F} = d\vec{p}/dt$ — which was always the deeper statement of the second law, and the way Newton wrote it. The force law that follows belongs to Relativistic Force and Newton's Second Law, its covariant form to Four-Force (Minkowski Force); here we stop at the reason a velocity-dependent mass had to go.

1 · An LHC proton, and the moment momentum stops caring about mass

A proton is accelerated to a total energy of $E = 6.8$ TeV in the LHC. Find $\gamma$, $\beta$, and its momentum. Then compare $p$ with $E/c$ and say what the comparison means.

Step 1 — identify. The proton's rest energy is $m_p c^2 = 938.272$ MeV, and the beam energy quoted by a machine is always the total energy $E = \gamma m c^2$. So $\gamma$ is simply a ratio of energies — no velocity needed, and none is measured.

Step 2 — apply. $\gamma = \dfrac{E}{m_pc^2} = \dfrac{6.8\times10^{6}\ \text{MeV}}{938.272\ \text{MeV}} = \mathbf{7247}$. Then from $\gamma^{-2} = 1 - \beta^2$, $\beta = \sqrt{1 - \gamma^{-2}} = \mathbf{0.99999999048}$. The proton is travelling about 2.8 m/s slower than light.

Step 3 — the momentum. Use the invariant rather than $\gamma m v$, which would demand ten significant figures of $\beta$: $pc = \sqrt{E^2 - (m_pc^2)^2} = \sqrt{(6\,800\,000)^2 - (938.272)^2}\ \text{MeV} = \mathbf{6\,799\,999.94\ \text{MeV}}$. So $pc$ falls short of $E$ by just $\mathbf{0.065}$ MeV — 65 keV out of 6.8 TeV.

Step 4 — the boundary number. $\dfrac{pc}{E} = \beta = 0.99999999048$. The proton's momentum equals $E/c$ to eight decimal places. At this energy the mass term in $E^2 = (pc)^2 + (mc^2)^2$ has essentially switched itself off, and the proton's energy–momentum bookkeeping is within one part in $10^8$ of a photon's. It is not massless — that 65 keV deficit is the mass — but for adding up momenta in a collision, $p = E/c$ is already correct to more digits than any detector can measure. Ultrarelativistic particles are cheap to work with, and this is why.

2 · Push a fast electron sideways and watch it refuse to go where you pushed

An electron travels at $\beta = 0.8$ along $x$. A force of magnitude $F$ is applied at $45°$ to its velocity, in the $xy$-plane. Which way does it accelerate?

Step 1 — identify. The naive answer is "at $45°$, obviously" — force and acceleration are parallel, everyone knows that. The relativistic answer comes from Step 6 of the derivation: $\vec{a} = \dfrac{1}{\gamma m}\Big[\vec{F} - \dfrac{(\vec{v}\cdot\vec{F})}{c^2}\vec{v}\Big]$.

Step 2 — apply. At $\beta = 0.8$, $\gamma = 1/\sqrt{1 - 0.64} = 5/3 = 1.6667$. Take $\vec{F} = \tfrac{F}{\sqrt2}(1, 1)$ and $\vec{v} = 0.8c\,(1,0)$. Then $\vec{v}\cdot\vec{F} = 0.8c\,F/\sqrt2$, and the subtracted term is $\dfrac{0.64\,F}{\sqrt{2}}(1,0)$.

Step 3 — the components. $\vec{a} \propto \tfrac{F}{\sqrt2}\big(1 - 0.64,\ 1\big) = \tfrac{F}{\sqrt2}\big(0.36,\ 1\big)$. The longitudinal component of the force is suppressed by exactly $1/\gamma^2 = 0.36$ relative to the transverse one — the $\gamma^3$-versus-$\gamma$ split, seen from the other side.

Step 4 — the boundary number. The acceleration points at $\arctan(1/0.36) = \mathbf{70.2°}$ to the velocity, not $45°$. The electron is deflected $25.2°$ away from the direction of the applied force, swung toward the transverse direction because it is so much harder to push forward than sideways. This is not a subtlety or a higher-order correction: it is a quarter-turn, at a speed that electron accelerators reach in the first metre. And it is the reason a scalar mass — any scalar mass — cannot be made to work. Scalars do not rotate vectors.

The first physicist to take Einstein's 1905 paper seriously in print was not a young radical. He was a forty-eight-year-old Berlin professor who had spent five years insisting he did not believe his own quantum hypothesis — and who now, reading a paper by an unknown patent clerk, immediately set about putting relativistic mechanics into the form it still has.

Texts. Rindler, Introduction to Special Relativity, 2nd ed., §§26–29 · Taylor & Wheeler, Spacetime Physics, 2nd ed., ch. 7 · Morin, Introduction to Classical Mechanics, ch. 12 · Griffiths, Introduction to Electrodynamics, 4th ed., §12.2 · Planck, "Das Prinzip der Relativität und die Grundgleichungen der Mechanik", Verh. Deutsch. Phys. Ges. 8, 136 (1906) · Lewis & Tolman, "The Principle of Relativity, and Non-Newtonian Mechanics", Phil. Mag. 18, 510 (1909) · Okun, "The concept of mass", Physics Today 42(6), 31 (1989).

Provenance. Signature $(+,-,-,-)$, so $U\cdot U = +c^2$ and $p\cdot p = +m^2c^2$. Everything above is recomputed in verify_aq-sr-relativistic-momentum.py: the transverse velocity transformation $v_y = -u/\gamma_w$; the symmetric-collision ledger $f(u) = f(v_B)/\gamma_w$ solved for $f$, with $f = \gamma$ confirmed to satisfy it identically in $u$ and $w$ (not merely in the limit) and $f \equiv 1$ shown to fail by exactly $\gamma_w$; $d\gamma/dt = \gamma^3(\vec v\cdot\vec a)/c^2$; $F_\parallel = \gamma^3 m a_\parallel$, $F_\perp = \gamma m a_\perp$ via $\gamma^2 = 1 + \gamma^2\beta^2$; the force–acceleration misalignment at $\beta = 0.8$ with $\vec F$ at $45°$ giving $\vec a$ at $70.2°$; and the LHC numbers $\gamma = 7247$, $\beta = 0.99999999048$, $pc = 6\,799\,999.94$ MeV, $E - pc = 0.065$ MeV. Anchors: $\gamma(0.9) = 2.2942$, $\gamma(0.99) = 7.0888$, $\gamma(0.999) = 22.366$, $m_pc^2 = 938.272$ MeV.

Scope. The energy–momentum relation $E^2 = (pc)^2 + (mc^2)^2$ is used here and derived at Four-Momentum, which owns the norm $p\cdot p = +m^2c^2$; $U\cdot U = +c^2$ is derived at Four-Vectors. The relativistic force law proper belongs to Relativistic Force and Newton's Second Law and its covariant form to Four-Force (Minkowski Force); this article develops $\vec F = d\vec p/dt$ only as far as the two-masses argument requires and stops there. The energy $E = \gamma mc^2$ is stated and used; Relativistic Energy owns it.

Builds on. Four-Vectors · Four-Velocity · Four-Momentum · Proper Time and Proper Length. Leads to. Relativistic Energy · Energy–Momentum Relation · Conservation of Four-Momentum · Relativistic Force and Newton's Second Law.