Centre-of-mass & lab frames

Every collision or decay in particle physics can be watched from more than one inertial frame, and special relativity guarantees that the physics is the same in all of them — only the numbers on the dials change. Two frames recur so often that they have names. The lab frame is the one your apparatus sits in: for a fixed-target experiment it is the frame where the struck particle is at rest and a beam ploughs into it. The centre-of-mass (CM) frame is the inertial frame in which the total three-momentum of the system vanishes, $\sum_i\mathbf p_i=0$ — the natural frame in which to think about what energy is actually available to make new particles. Thomson's convention marks CM quantities with a star, so $p^*$ is a momentum measured in the CM frame. The bridge between the two is the four-vector formalism from the prerequisite articles. Energy and momentum combine into a four-momentum $p^\mu=(E,\mathbf p)$ whose square $p^\mu p_\mu=E^2-|\mathbf p|^2=m^2$ is the same number in every frame. Add the four-momenta of a whole system and its square is the squared invariant mass. Applied to a two-body collision this gives the Mandelstam invariant $s=(p_1+p_2)^2$ , and evaluating it in the CM frame — where the momenta are equal and opposite — collapses it to $s=(E_1^*+E_2^*)^2$ . So $\sqrt s$ is the total energy in the centre of mass. That single identity is why colliders exist, why a fixed-target machine's reach grows only as the square root of its beam energy, and how the same event can be re-expressed in whichever frame makes the answer fall out.

Key takeaways

The CM frame is defined by $\sum_i\mathbf p_i=0$ ; the lab frame has the target at rest. CM quantities carry a star, e.g. $p^*$ .
Energy and momentum form a four-vector $p^\mu=(E,\mathbf p)$ whose invariant square is $p^\mu p_\mu=E^2-|\mathbf p|^2=m^2$ — the same in every frame.
The Mandelstam invariant $s=(p_1+p_2)^2$ evaluated in the CM frame gives $s=(E_1^*+E_2^*)^2$ , so $\sqrt s$ is the total centre-of-mass energy.
A fixed target gives only $\sqrt s\approx\sqrt{2E_1 m_2}$ (grows as the square root of beam energy); a symmetric collider gives $\sqrt s=2E$ (the full sum).
The frames are joined by a Lorentz boost of velocity $\beta_{\rm cm}=|\mathbf p_1|/(E_1+m_2)$ ; the CM momentum is $p^*=\sqrt{\lambda(s,m_1^2,m_2^2)}/(2\sqrt s)$ .
Key idea: pick the frame that simplifies the physics, but compute Lorentz-invariant combinations like $s=(p_1+p_2)^2$ , which take the same value in every frame — so the final answer is frame-independent.

The core insight

Pick the frame that simplifies the physics, but compute Lorentz-invariant combinations like $s=(p_1+p_2)^2$ that have the same value in every frame, so the answer is frame-independent.

Where this lives today

Frame kinematics is the daily language of every collider experiment. When the LHC quotes " $\sqrt s=13.6$ TeV" it is reporting exactly the invariant of equation (2), and every discovery threshold is a statement about $\sqrt s$ : the Higgs at 125 GeV, the top quark at 173 GeV, any new particle of mass $M$ — all require $\sqrt s$ comfortably above the sum of the final-state masses. The invariant-mass technique is how those particles are found: reconstruct the four-momenta of a set of decay products, add them, square the sum — a peak in $\sqrt{(\sum p_i)^2}$ over smooth background is a resonance, and this is how the Higgs announced itself in the $\gamma\gamma$ and four-lepton channels in 2012. Fixed-target physics is far from dead, precisely because its forward-boosted geometry is a feature for some goals: neutrino beams (NuMI, DUNE's LBNF) rely on a boosted CM frame to collimate secondaries down a long baseline. And asymmetric $e^+e^-$ colliders such as PEP-II (BaBar) and KEKB/SuperKEKB (Belle) deliberately run unequal beam energies so the CM frame is boosted in the lab, giving short-lived $B$ mesons enough flight distance to measure their decay times — a bespoke choice of frame kinematics engineered into the machine. Every one of these is the same four-vector bookkeeping, tuned to the physics goal.

The physical picture

Watch a single collision from two seats. Sit in the lab and the target hangs motionless while the beam rushes in; all the debris is thrown forward, sharing the beam's momentum, because the whole system was already moving. Now boost yourself into the frame where the two incoming particles carry equal and opposite momenta — the centre of mass. The same collision now looks symmetric: the pair rushes together head-on, and the products stream out back-to-back with no net drift. Nothing about the event changed; only your velocity did. The first figure puts the two views side by side so "boosted forward" and "back-to-back" are visibly the same physics from two frames.

Figure

Lab frame versus centre-of-mass frame for the same 2→2 collision. In the lab the target m₂ is at rest and the products are thrown forward because the centre of mass itself moves at β_cm = p₁ ∕ (E₁ + m₂); boosting to the CM frame makes the incoming momenta equal and opposite, so Σp = 0 and the products emerge back-to-back with total energy the invariant √s = E₁ + E₂*

The reason anyone cares which frame they use is energy bookkeeping. What matters for creating heavy particles is not the beam energy but the invariant energy $\sqrt s$ — what is left over once you subtract the bulk motion, i.e. the energy in the frame where nothing is drifting. In a fixed-target setup most of the beam's energy is locked up in the forward motion of the debris; only the CM slice can be converted to mass. In a symmetric collider the CM frame is the lab frame, so nothing is wasted. The second figure is the number that built every collider on Earth: how $\sqrt s$ climbs with beam energy in the two configurations.

Figure

Available centre-of-mass energy √s versus beam energy on log-log axes. A symmetric collider gives √s = 2E, a straight line of slope 1; a fixed-target machine gives only √s ≈ √(2 m_p E), a line of slope 1∕2. At the LHC beam energy of 7 TeV the collider reaches √s = 14 TeV, whereas the same beam striking a stationary proton would reach only √s = 114.6 GeV — a factor of 122 lost to the forward motion of the debris

The core equations

The whole subject rests on one object: the four-momentum. Energy and the three components of momentum are packaged into a single four-vector, and its Minkowski square is a Lorentz invariant fixed entirely by the particle's mass. This is the on-shell condition, and it holds in every frame.

p^\mu=(E,\mathbf p),\qquad p^\mu p_\mu=E^2-|\mathbf p|^2=m^2 \qquad (1)

where: $p^\mu$ — contravariant four-momentum; $E$ — energy; $\mathbf p$ — three-momentum; $m$ — rest mass. Natural units ( $c=1$ ) and the metric $(+,-,-,-)$ are used throughout, so energy, momentum and mass share the unit GeV. Reading it: the length of the energy-momentum vector is the mass — boosting to another frame slides $E$ and $\mathbf p$ around but never changes their invariant combination. The two frames of this article are now defined by conditions on the total momentum. The CM frame is the one where the summed three-momentum is zero; the lab frame is the one where the target sits still. Because the total four-momentum of the system is itself a four-vector, its square — the Mandelstam variable $s$ — is invariant, and in the CM frame it reduces to the square of the total energy.

\sum_i\mathbf p_i^{\,*}=0,\qquad s\equiv\Big(\sum_i p_i\Big)^2 \;\xrightarrow{\ \rm CM\ }\; \Big(\sum_i E_i^{*}\Big)^2 \equiv E_{\rm cm}^2 \qquad (2)

where: the star marks CM-frame quantities; $s$ — the squared invariant mass of the whole system, a single Lorentz-invariant number; $E_{\rm cm}=\sqrt s$ — the total energy in the centre of mass. Reading it: $\sqrt s$ is the energy budget for making new particles, and it can be computed in any frame because $s$ is invariant — you never actually have to sit in the CM frame to know it. Specialising to a $2\to2$ process, $s$ takes a form you can read off from lab-frame quantities alone. This is the equation that separates a collider from a fixed-target machine: the first term is symmetric in the beams, the fixed-target result carries only a single power of the beam energy.

s=m_1^2+m_2^2+2\big(E_1E_2-\mathbf p_1\!\cdot\!\mathbf p_2\big);\quad \underbrace{s=m_1^2+m_2^2+2E_1m_2}_{\text{fixed target}},\quad \underbrace{\sqrt s=2E}_{\text{symmetric collider}} \qquad (3)

where: subscripts 1, 2 label the two incoming particles; the fixed-target line sets $\mathbf p_2=0$ , $E_2=m_2$ ; the collider line sets equal head-on beams $\mathbf p_1=-\mathbf p_2$ , $E_1=E_2=E$ . Reading it: at high energy the fixed-target $\sqrt s\approx\sqrt{2E_1m_2}$ crawls upward as the square root of beam energy, while the collider delivers the full $2E$ . Finally, to translate a measurement between frames you need the boost that connects them and the momentum each particle carries in the CM. For a fixed-target collision the centre of mass moves with the system's total momentum divided by its total energy, and the common CM momentum follows from energy conservation as the Källén (triangle) function.

\beta_{\rm cm}=\frac{|\mathbf p_1|}{E_1+m_2},\quad \gamma_{\rm cm}=\frac{E_1+m_2}{\sqrt s},\qquad p^{*}=\frac{\sqrt{\lambda(s,m_1^2,m_2^2)}}{2\sqrt s} \qquad (4)

where: $\beta_{\rm cm},\gamma_{\rm cm}$ — velocity and Lorentz factor of the CM frame relative to the lab; $p^{*}$ — magnitude of each incoming particle's three-momentum in the CM; $\lambda(a,b,c)=a^2+b^2+c^2-2ab-2bc-2ca=[a-(\sqrt b+\sqrt c)^2][a-(\sqrt b-\sqrt c)^2]$ . Reading it: boost by $\beta_{\rm cm}$ and the forward-thrown lab debris becomes the symmetric back-to-back CM picture of the first figure.

Derivation AWhy the square root of s is the total centre-of-mass energy

The goal is to prove the identity that everything else hangs on: that the Mandelstam invariant $s=(p_1+p_2)^2$ , defined as a Lorentz scalar out of the incoming four-momenta, equals the square of the total energy measured in the centre-of-mass frame. We will build $s$ from its invariant definition, evaluate it in the CM frame where the three-momenta cancel, and confirm that the leftover is purely the summed energy — so that $\sqrt s$ deserves its name as the available energy, computable from any frame at all.

Step 1. Start from the invariant definition $s=(p_1+p_2)^\mu(p_1+p_2)_\mu$ . Expanding the square gives $s=p_1^2+p_2^2+2\,p_1\!\cdot\!p_2=m_1^2+m_2^2+2(E_1E_2-\mathbf p_1\!\cdot\!\mathbf p_2)$ , using the on-shell condition (1) for each particle. Why: because $s$ is a scalar product of four-vectors it is Lorentz invariant, so this one expression holds in every frame — we are free to evaluate it wherever it is simplest.

Step 2. Move to the CM frame, defined by $\mathbf p_1^{*}+\mathbf p_2^{*}=0$ , i.e. $\mathbf p_2^{*}=-\mathbf p_1^{*}$ . The total four-momentum there is $P^{*}=(E_1^{*}+E_2^{*},\,\mathbf 0)$ — its spatial part is zero by construction. Why: the CM frame is precisely the frame that annihilates the total three-momentum, leaving only energy in the time component.

Step 3. Square that total four-momentum: $s=P^{*}\!\cdot\!P^{*}=(E_1^{*}+E_2^{*})^2-|\mathbf 0|^2=(E_1^{*}+E_2^{*})^2$ . Taking the root, $\sqrt s=E_1^{*}+E_2^{*}\equiv E_{\rm cm}$ . Why: with no three-momentum to subtract, the invariant mass of the system is nothing but its total energy in this frame — this is the sense in which $\sqrt s$ is "the energy available".

Step 4. Because $s$ was invariant from the start, the value obtained in the CM frame equals the value obtained from lab quantities in Step 1. So one can measure $E_1,E_2,\mathbf p_1,\mathbf p_2$ in the lab, form $s$ , and know the CM energy without ever transforming a single vector. Why: this frame-independence is the entire practical payoff — invariants let you compute in the convenient frame and report a frame-independent answer.

Go deeper · Derivation B · 1 minFixed target versus collider — where the energy goes

Derivation BFixed target versus collider — where the energy goes

The aim is to derive, from the single invariant $s$ , the stark difference between a fixed-target machine and a collider, and to show quantitatively that a stationary target wastes almost all of the beam's energy on bulk forward motion. We evaluate the $2\to2$ expression for $s$ in each configuration, take the high-energy limit, and read off how the available energy $\sqrt s$ scales with the beam energy $E$ — the result that dictated the architecture of every modern accelerator.

Step 1. Fixed target: particle 1 (energy $E_1$ , momentum $\mathbf p_1$ ) strikes particle 2 at rest, so $E_2=m_2$ and $\mathbf p_2=0$ . Then $s=m_1^2+m_2^2+2E_1m_2$ . Why: with the target motionless the scalar product $p_1\!\cdot\!p_2$ collapses to $E_1m_2$ , leaving $s$ linear in the beam energy.

Step 2. At high energy $E_1\gg m_1,m_2$ , the constant mass terms are negligible and $s\approx 2E_1m_2$ , so $\sqrt s\approx\sqrt{2E_1m_2}$ . Doubling the beam energy multiplies $\sqrt s$ by only $\sqrt2\approx1.41$ . Why: the available energy grows as the square root of the beam energy, because the CM frame is dragged forward and most of $E_1$ is spent moving the debris, not creating mass.

Step 3. Symmetric collider: equal beams head-on, $E_1=E_2=E$ and $\mathbf p_1=-\mathbf p_2$ , so the total three-momentum is zero — the lab is the CM frame. Then $s=(2E)^2$ and $\sqrt s=2E$ , growing linearly. Why: with nothing drifting, no energy is locked up in bulk motion; the entire $2E$ is available to make particles.

Step 4. Compare at a real beam energy. At $E=7\ \mathrm{TeV}$ (an LHC beam), the collider gives $\sqrt s=14\ \mathrm{TeV}$ , while the same beam on a proton target gives $\sqrt s=\sqrt{2(0.938)(7000)}\ \mathrm{GeV}=114.6\ \mathrm{GeV}$ — smaller by a factor of 122. To match the collider's 14 TeV in fixed-target mode you would need a beam of $s/2m_2\approx1.0\times10^{8}\ \mathrm{GeV}$ . Why: this three-order-of-magnitude gap, visible as the diverging lines of the second figure, is the entire reason particle physics moved from fixed targets to colliding beams.

Go deeper · Derivation C · 1 minThe CM momentum and the boost that connects the frames

Derivation CThe CM momentum and the boost that connects the frames

The purpose here is to pin down the two quantities you need to move a real measurement between the lab and the CM frame: the common CM momentum $p^{*}$ of the colliding pair, and the boost velocity $\beta_{\rm cm}$ that carries one frame into the other. We impose momentum balance and energy conservation in the CM frame to solve for the individual CM energies, extract $p^{*}$ as the Källén function, and then identify the boost from the total lab four-momentum.

Step 1. In the CM frame the momenta are equal and opposite, $|\mathbf p_1^{*}|=|\mathbf p_2^{*}|\equiv p^{*}$ , and energy conservation reads $\sqrt s=E_1^{*}+E_2^{*}=\sqrt{p^{*2}+m_1^2}+\sqrt{p^{*2}+m_2^2}$ . Why: the defining condition $\mathbf p_1^{*}=-\mathbf p_2^{*}$ forces a single momentum magnitude, and the total energy is fixed to the invariant $\sqrt s$ .

Step 2. Isolate one root and square to remove it: from $\sqrt s-E_1^{*}=E_2^{*}$ one finds $s-2\sqrt s\,E_1^{*}=m_2^2-m_1^2$ , giving the individual CM energy $E_1^{*}=(s+m_1^2-m_2^2)/(2\sqrt s)$ and symmetrically $E_2^{*}=(s+m_2^2-m_1^2)/(2\sqrt s)$ . Why: each particle's CM energy is fixed entirely by $s$ and the two masses — no angles enter, because the incoming state is collinear.

Step 3. Recover the momentum from the on-shell relation $p^{*2}=E_1^{*2}-m_1^2$ . Algebra collapses the numerator to the Källén function, $p^{*}=\sqrt{\lambda(s,m_1^2,m_2^2)}\big/2\sqrt s$ , which vanishes exactly at threshold $\sqrt s=m_1+m_2$ . Why: at threshold there is just enough energy to make the two particles at rest in the CM, so their momentum must go to zero — the Källén form encodes this automatically.

Step 4. For the boost, the total lab four-momentum of a fixed-target collision is $P=(E_1+m_2,\ \mathbf p_1)$ . The CM frame is where its spatial part vanishes, reached by a boost of velocity $\boldsymbol\beta_{\rm cm}=\mathbf p_1/(E_1+m_2)$ and factor $\gamma_{\rm cm}=(E_1+m_2)/\sqrt s$ . Why: the CM velocity is just total momentum over total energy; applying this boost turns the forward-thrown lab debris into the symmetric back-to-back CM picture. *Why these are consistent: for equal masses ( $m_1=m_2=m$ ) the general formula $p^{*}=\sqrt{\lambda}/2\sqrt s$ reduces to $p^{*}=\tfrac12\sqrt{s-4m^2}$ , as expected. The boost factor satisfies the standard identity $\gamma_{\rm cm}^2(1-\beta_{\rm cm}^2)=1$ , confirming that $\beta_{\rm cm}$ and $\gamma_{\rm cm}=(E_1+m_2)/\sqrt s$ are consistent.*

Limiting cases

Case	Result	Meaning
Symmetric collider	$\sqrt s=2E$	equal head-on beams; the lab is the CM frame, so the full beam energy is available
Fixed target, high energy	$\sqrt s\approx\sqrt{2E_1m_2}$	available energy grows only as the square root of beam energy — the collider motivation
Fixed target, non-relativistic	$\sqrt s\approx m_1+m_2+\tfrac12\mu v_{\rm rel}^2$	reduces to the Newtonian centre of mass; kinetic energy set by the reduced mass $\mu$
Ultra-relativistic / massless	$s\approx 2E_1E_2(1-\cos\theta_{12})$	$s\to4E_1E_2$ head-on, $s\to0$ collinear; masses drop out
Threshold	$\sqrt s=\sum_f m_f$ , $p^{*}=0$	all final particles at rest in the CM — the minimum energy to create them

Common trapDoubling a fixed-target machine's beam energy doubles its reach

It does not. What sets the reach — the heaviest thing you can make — is the invariant $\sqrt s$ , and for a fixed target $\sqrt s\approx\sqrt{2E_1m_2}$ grows only as the square root of the beam energy. Double $E_1$ and $\sqrt s$ rises by a mere factor of $\sqrt2\approx1.41$ ; to double $\sqrt s$ you must quadruple the beam energy. This is the whole reason the field abandoned fixed-target machines for colliders once it wanted the energy frontier: in a symmetric collider $\sqrt s=2E$ grows linearly, so every extra GeV of beam energy is an extra 2 GeV of reach, none of it wasted on dragging the debris forward. The confusion usually comes from equating "beam energy" with "energy available to make particles" — but in the lab frame the collision products inherit the beam's momentum and carry off most of that energy as bulk forward motion, which can never be turned into mass. Only the CM slice, $\sqrt s$ , is convertible. (A related trap: the invariant mass of a system is not the sum of the particles' masses — it also contains their relative kinetic energy, which is why $\sqrt s$ can exceed $m_1+m_2$ and make heavier states.)

Worked examples

1 · The antiproton threshold at a fixed target (fully worked)

To create an antiproton you must respect baryon-number conservation, so the lightest allowed reaction on a hydrogen target is $p+p\to p+p+p+\bar p$ (net baryon number 2 on both sides, charge +2 on both sides). What beam energy does a fixed-target machine need? Take $m_p=0.938$ GeV.

Step 1. Find the threshold invariant mass. At threshold all four final particles sit at rest in the CM frame, so the minimum invariant mass of the final state is $\sqrt{s_{\min}}=4m_p$ , giving $s_{\min}=(4m_p)^2=16m_p^2$ . This is a CM-frame statement, but $s$ is invariant, so it holds in the lab too.

Step 2. Express $s$ from lab quantities. Beam proton (energy $E_1$ ) on a proton at rest: $s=m_p^2+m_p^2+2E_1m_p=2m_p^2+2E_1m_p$ . Set equal to threshold: $2m_p^2+2E_1m_p=16m_p^2$ , so $2E_1m_p=14m_p^2$ and $E_1=7m_p$ .

Step 3. Convert to numbers. The total beam energy is $E_1=7m_p=\mathbf{6.57\ GeV}$ ; the kinetic energy the accelerator must supply is $T=E_1-m_p=6m_p=\mathbf{5.63\ GeV}$ . The Bevatron was built for 6.2 GeV — a deliberate margin, since target protons bound in nuclei have Fermi motion that lowers the effective threshold.

Step 4. Contrast with a collider. A $pp$ collider needs only $\sqrt s=4m_p=3.75$ GeV, i.e. beams of $2m_p=\mathbf{1.88\ GeV}$ each — less than a third of the fixed-target beam for the identical final state. Fixed target needs a 6.57 GeV beam (T = 5.63 GeV); a collider needs only 1.88 GeV per beam.

2 · Collider versus fixed target at the LHC (fully worked)

The LHC collides two 7 TeV proton beams head-on. What centre-of-mass energy does it reach, and what would it take to match that with a fixed proton target? Take $m_p=0.938$ GeV.

Step 1. Collider CM energy. Equal head-on beams give $\sqrt s=2E=2\times7\ \mathrm{TeV}=\mathbf{14\ TeV}$ — the familiar LHC design energy, and it grows linearly with beam energy.

Step 2. Same beam, fixed target. Put one 7 TeV proton onto a stationary one: $\sqrt s=\sqrt{2E_1m_2}=\sqrt{2(7000)(0.938)}\ \mathrm{GeV}=\sqrt{13132}\ \mathrm{GeV}=\mathbf{114.6\ GeV}$ . The identical beam delivers 122 times less available energy — the vertical red gap in the second figure.

Step 3. Beam needed to match 14 TeV in fixed-target mode. Invert $s=2E_1m_2$ : $E_1=s/(2m_2)=(14000)^2/(2\times0.938)\ \mathrm{GeV}=1.96\times10^{8}/1.876\ \mathrm{GeV}\approx\mathbf{1.0\times10^{8}\ GeV}=10^{5}\ \mathrm{TeV}$ .

Step 4. Interpret. A fixed-target machine would need a beam roughly ten million times more energetic than the LHC's to reach the same physics — an accelerator far larger than the Earth. Collider: 14 TeV; same beam fixed-target: 114.6 GeV; fixed-target beam to match: about 10⁸ GeV. This single comparison is why the energy frontier is built from colliding beams.

3 · Centre-of-mass momentum in Z → μ⁺μ⁻ (fully worked)

A $Z$ boson at rest decays to $\mu^+\mu^-$ . What momentum does each muon carry? Take $m_Z=91.19$ GeV and $m_\mu=0.1057$ GeV.

Step 1. Set up. The $Z$ at rest is already its own CM frame, so $\sqrt s=m_Z$ and the two muons emerge back-to-back with equal momenta $p^{*}$ . Equal masses ( $m_1=m_2=m_\mu$ ) simplify the Källén formula to $p^{*}=\tfrac12\sqrt{s-4m_\mu^2}$ .

Step 2. Plug in. $s=m_Z^2=(91.19)^2=8315.2\ \mathrm{GeV}^2$ and $4m_\mu^2=4(0.1057)^2=0.0447\ \mathrm{GeV}^2$ , so $p^{*}=\tfrac12\sqrt{8315.2-0.0447}=\tfrac12\sqrt{8315.2}=\mathbf{45.59\ GeV}$ .

Step 3. Sanity check. Because the muons are so light compared with the $Z$ , $p^{*}$ is essentially $\sqrt s/2=m_Z/2=45.59$ GeV, and each muon's energy $E^{*}=m_Z/2$ as well — they are ultra-relativistic, $E^{*}\approx p^{*}$ . Each muon carries p\* ≈ 45.6 GeV, back-to-back — very nearly half the Z mass. The tiny difference from exactly $m_Z/2$ is the $m_\mu^2$ term, which the Källén function tracks exactly and which grows important only for heavy decay products such as $Z\to b\bar b$ .

Physics of the portrait

The centre-of-mass energy is an abstract invariant, but it has a very concrete monument: the collider. And the collider has an inventor whose entire motivation was the arithmetic of this article — that a fixed target throws away almost all of a beam's energy, while colliding two beams keeps it. That inventor was Bruno Touschek.

Origins · 1 minBruno Touschek and how this was discovered

Bruno Touschek

1921–1978 · colliding-beam physics and the centre-of-mass energy advantage

Bruno Touschek was born in Vienna in 1921 to a Catholic father and a Jewish mother; Nazi race laws barred him from formal study, so he learned physics half in secret, and during the war worked in Germany with Rolf Widerøe on the betatron even while under threat, at one point imprisoned by the Gestapo. He took his doctorate at Göttingen in 1949 and settled in Rome. There, in a now-legendary Frascati seminar in early 1960, he argued that the way to reach high energy was not to slam a beam into a fixed target — where $\sqrt s$ climbs only as the square root of the beam energy — but to collide two beams head-on, where the lab frame is the centre-of-mass frame and $\sqrt s=2E$ . Better still, a single ring could store electrons and positrons at once: being of opposite charge, they circulate in opposite directions in the same field and meet head-on. Within a year his team built AdA (Anello Di Accumulazione), the first storage-ring collider, proving counter-rotating beams could be stored and collided. Every $e^+e^-$ and hadron collider since — ADONE, SPEAR, LEP, the LHC — descends from that idea. His name also survives in the "Touschek effect," the intra-beam scattering that limits stored-beam lifetime. He died in 1978, having turned the kinematics of the CM frame into the dominant instrument of particle physics.

History

1905

— Einstein's special relativity establishes the equivalence of inertial frames

1908

— Minkowski recasts relativity in four-vector form, making invariants central

1943

— Widerøe files the colliding-beam patent, granted in 1953

1955

— Chamberlain and Segrè discover the antiproton at the 6.2 GeV Bevatron

1961

— Touschek's AdA at Frascati stores counter-rotating e⁺e⁻ beams, the first collider

1971

— CERN's Intersecting Storage Rings become the first proton-proton collider

Going further

Texts. Thomson, Modern Particle Physics, §2.2 (special relativity, four-vectors and Mandelstam variables) · Griffiths, Introduction to Elementary Particles, 2nd ed., Ch. 3 (relativistic kinematics), §3.1–3.4 · Halzen & Martin, Quarks and Leptons, §4.1–4.3 · Peskin & Schroeder, An Introduction to Quantum Field Theory, §5.4 and Appendix (kinematics and phase space) · Schwartz, Quantum Field Theory and the Standard Model, Ch. 5. The Mandelstam variable $s$ introduced here is developed in full, alongside $t$ and $u$ , in the Mandelstam Variables article (forward reference, not re-derived here). Key numerical checks: (1) $s=(p_1+p_2)^2=(E_1^{*}+E_2^{*})^2$ ; (2) fixed-target $\sqrt s=\sqrt{2E_1m_2}$ gives 114.6 GeV for a 7 TeV beam on a proton, versus 14 TeV for the collider; (3) the antiproton threshold $E_1=7m_p=6.57$ GeV, $T=5.63$ GeV; (4) $p^{*}=\sqrt{\lambda}/2\sqrt s$ gives 45.6 GeV for $Z\to\mu^+\mu^-$ . Proton mass $m_p=0.938$ GeV, $m_Z=91.19$ GeV, $m_\mu=0.1057$ GeV are standard PDG values.

Prerequisites. Special Relativity Kinematics · Four-Vector Notation · Natural Units. Leads to. Mandelstam Variables · Luminosity and Event Rates · Collider Experiments.

Ask the tutor: Ask me to boost any real collision — an LHC event, a fixed-target beam, or a decay — between the lab and centre-of-mass frames and watch the same physics fall out of both.

What is the difference between the lab frame and the centre-of-mass frame?

The lab frame is whatever frame your detector sits in — for a fixed-target experiment, the frame where the struck target is at rest and the beam carries all the momentum. The centre-of-mass frame is instead defined by a condition on the total momentum: the unique inertial frame in which the summed three-momenta cancel to zero. In a fixed-target setup these two frames differ, and the CM frame moves forward at $\beta_{\rm cm}=|\mathbf p_1|/(E_1+m_2)$ . In a symmetric collider — equal, opposite beams — the total momentum is already zero, so lab and CM coincide, which is exactly why colliders are so efficient.

Why is √s the "energy available" and not just the beam energy?

Because only the energy in the centre-of-mass frame can be converted into the rest mass of new particles. In the lab frame the products inherit the incoming momentum and stream forward; that bulk kinetic energy is real, but it can never be turned into mass, since momentum must be conserved and the debris cannot simply stop. Subtract that unavoidable forward motion and what remains is $\sqrt s$ , the total energy in the frame where nothing is drifting — the budget you get to spend on making particles. It is why the number physicists quote for a machine's "energy" is always $\sqrt s$ , not the beam energy.

Why do particle physicists work in natural units with c = 1?

Setting $c=1$ (and $\hbar=1$ ) removes bookkeeping factors that carry no physics and lets energy, momentum and mass all be measured in the same unit, the GeV. The on-shell relation then reads simply $E^2-|\mathbf p|^2=m^2$ instead of $E^2-p^2c^2=m^2c^4$ , and $s$ has units of GeV² so that $\sqrt s$ is directly an energy. It is a pure convention — you can restore the $c$ 's at the end by dimensional analysis — but it makes the four-vector algebra far cleaner. The prerequisite article on natural units sets the conventions used here.

Is the invariant mass of a two-particle system just the sum of their masses?

No, and this is a common trap. The invariant mass is $\sqrt s=\sqrt{(p_1+p_2)^2}$ , which includes the particles' relative kinetic energy, not only their rest masses. It equals $m_1+m_2$ only in the special case where the two particles are at rest relative to each other (threshold). In general $\sqrt s>m_1+m_2$ , and that excess is precisely what lets a collision create heavier particles than the ones that went in. When experiments "reconstruct an invariant mass" to hunt for a resonance, they are computing this $\sqrt s$ for a candidate set of decay products and looking for a peak at the parent's mass.

How does the invariant-mass method actually find new particles?

Take the four-momenta of a candidate set of final-state particles — say two photons, or four leptons — measured in the lab. Add them component by component and take the root of the square: $M=\sqrt{(\sum_i p_i)^2}$ . Because this is a Lorentz invariant, it does not matter that the parent was moving; $M$ comes out as the parent's rest mass whenever those products really came from a single decay. Plot $M$ over many events and a genuine particle shows up as a sharp peak on the smooth combinatorial background. This is exactly how the Higgs was seen in 2012 in its $\gamma\gamma$ and $ZZ^*\to4\ell$ channels, at $M\approx125$ GeV.

Why do the B-factories deliberately use unequal beam energies?

Because they want the produced $B$ mesons to move in the lab. At the $\Upsilon(4S)$ resonance the CM energy is only just above twice the $B$ mass, so in the CM frame each $B$ is almost stationary and decays where it was born — impossible to time. By colliding, say, 9 GeV electrons with 3.1 GeV positrons (PEP-II/BaBar), the whole CM frame is boosted along the beam, so the $B$ mesons fly a measurable fraction of a millimetre before decaying. That flight distance gives their proper decay time, which is what makes precision studies of $B$ -meson mixing and CP violation possible — a deliberate, engineered choice of frame kinematics, the same physics of this article used as a design knob.