The Mass Shell and the Scalar Part: What a Congruence Preserves

Introduction

The slogan of the two sectors is that mass is what the frame cannot move and information is what the frame leaves free. The slogan is true, and it is true for a reason that is easy to state one step too strongly. In the material sector two different things pin the same coordinate, and they agree. In the informational sector two different things pin two different coordinates, and they do not. One sector has a redundancy that hides an asymmetry; the other exposes it.

The article is about that asymmetry and about nothing else. It is short because the content is a table plus the reason the table has the shape it has.

The findings are stated in advance.

  1. One quadratic form, two planes. The form $\mathcal{N}=\mathrm{Re}\,(q_0^{2}+q_1^{2}+q_2^{2}+q_3^{2})$ is $-m^{2}c^{2}$ on the anti-Hermitian plane and $\tfrac14(1-\lvert\vec s\rvert^{2})$ on the Hermitian one. There is no second definition to reconcile.
  2. A congruence preserves $\mathcal{N}$ and moves $2\,\mathrm{Sc}$, in both sectors, without exception. Worst deviation in $\mathcal{N}$ over a hundred boosts: $3.3\times10^{-16}$; the change in $2\,\mathrm{Sc}$ has median $0.165$ and maximum $2.314$ over the same hundred.
  3. In the material sector the two mechanisms agree. The definition fixes $\mathcal{N}=-m^{2}c^{2}$ and the group preserves $\mathcal{N}$, so an on-shell element stays on shell at the same mass while $E/c$ runs from $1.000$ to $3.762$ over nine rapidities.
  4. In the informational sector they do not coincide. The definition fixes $2\,\mathrm{Sc}=1$ and the group does not preserve it, so the image of a state is off the slice. At $\beta=0.5$ with $\lvert\vec h_\Psi\rvert=1.6$, a boost of $+0.4$ gives $2\,\mathrm{Sc}=0.808317724032$.
  5. The projection is therefore forced, and the running impurity is its consequence. The read radius follows $z'=(z+t)/(1+zt)$ to $2.2\times10^{-16}$, and the impurity moves with it.
  6. The quantity that survives is the scale-free ratio $N/(\mathrm{Sc})^{2}$, unchanged at $0.300000000000000$ under a scale of $0.5$, $2$ and $7.3$.
  7. Reading $N=0$ as massless is an error in the informational sector. There $N=0$ means pure, and the Hermitian form on an idempotent is $\tfrac12$, not zero.

The notation is that of the companions. An element is $\tilde Q=q_0e_t+q_1e_1+q_2e_2+q_3e_3$; the material plane is the anti-Hermitian one, with $q_0$ imaginary and $\vec q$ real; the informational plane is the Hermitian one, with $q_0$ real and $\vec q$ imaginary. The state is $\tilde{Ϛ}_{\Psi}=\tfrac12(e_t+i\vec s_\Psi)$; the momentum is $\tilde p_\Phi=i(E/c)e_t+\vec p$. The congruence is $\mathrm{Cg}_{\tilde U}:\tilde Q\mapsto\tilde U\tilde Q\tilde U^{*}$; $2\,\mathrm{Sc}$ is twice the $e_t$ coefficient; $\langle\cdot\rangle_{\natural}$ is the quaternionic norm.

One Form on Two Planes

The two readings of $N$ are one expression. Write $\mathcal{N}(\tilde Q)=\mathrm{Re}\sum_{k=0}^{3}q_k^{2}$. On the informational plane $q_0$ is real and $q_j=i s_j/2$, so $$ \mathcal{N}=\tfrac14-\tfrac14\lvert\vec s_\Psi\rvert^{2}, $$ which is the impurity, at most $\tfrac14$ and zero on the pure states. On the material plane $q_0=iE/c$ and $q_j=p_j$, so $$ \mathcal{N}=-\frac{E^{2}}{c^{2}}+\lvert\vec p\rvert^{2}=-m^{2}c^{2}, $$ which is minus the mass shell.

So there are not two definitions of one symbol. There is one polynomial, evaluated on two linear subspaces of the complexification: the Hermitian subspace where the scalar coefficient is real, and the anti-Hermitian subspace where it is imaginary. The two ranges look unrelated — one is bounded above by $\tfrac14$ and the other is unbounded below — and they are the same function of different arguments.

The signs carry the physics and they are worth stating once. The impurity is positive and bounded because a state has unit scalar part and a vector no longer than one; the mass shell is negative because the Minkowski signature puts the time component in with the opposite sign. Neither sign is a convention of the framework; both are inherited from the algebra, and the passage between them is the change of plane.

What the Congruence Does

In both sectors the congruence does exactly two things. It preserves $\mathcal{N}$ and it moves $2\,\mathrm{Sc}$.

material, $\mathbb{B}_{\Phi}$ informational, $\mathbb{B}_{\Psi}$
plane anti-Hermitian Hermitian
$\mathcal{N}$ $-m^{2}c^{2}$ $\tfrac14(1-\lvert\vec s\rvert^{2})$
what the definition fixes $\mathcal{N}=-m^{2}c^{2}$ $2\,\mathrm{Sc}=1$
$\mathcal{N}$ under the congruence fixed fixed
$2\,\mathrm{Sc}$ under the congruence runs, $2iE/c$ runs
does the group preserve the fixed coordinate yes no
what the sector's own reading does stays put must project, then runs

The numbers behind the table. Over a hundred random (state, boost) pairs the worst deviation in $\mathcal{N}$ is $3.3\times10^{-16}$, and the change in $2\,\mathrm{Sc}$ from its value of one has median $0.165$, minimum $6.7\times10^{-5}$ and maximum $2.314$. The group is an isometry of the form and a shear of the slice.

Material. On shell at nine rapidities from $-2$ to $2$, with $\lvert\vec p\rvert$ and $E/c$ computed from $m$ and $\eta$, the invariant $E^{2}/c^{2}-\lvert\vec p\rvert^{2}$ stays at $m^{2}c^{2}$ to $1.8\times10^{-15}$ while $E/c$ runs from $1.000$ to $3.762$ and $\lvert\vec p\rvert$ to $3.627$. A boost carries an on-shell element to another on-shell element of the same mass, and the mass does not move.

Informational. At $\beta=0.5$ and $\lvert\vec h_\Psi\rvert=1.6$, so $\Theta=0.8$, a boost of $+0.4$ along the axis sends $2\,\mathrm{Sc}$ from $1$ to $0.808317724032$; a boost of $-0.3$ sends it to $1.247551186271$. Both match $\cosh\theta-\tanh\Theta\sinh\theta$ to twelve digits. The image is a positive element off the slice.

The Asymmetry

In the material sector the two mechanisms are one mechanism. Mass is fixed by the definition of the object — a particle is on its mass shell — and mass is also fixed by the group, because the group preserves $\mathcal{N}$. The two statements coincide, and the coincidence is why the slogan reads so cleanly: the frame cannot move the mass is true twice over, and a reader never has to ask which statement is meant.

In the informational sector they are different mechanisms, and only one of them holds. The definition fixes $2\,\mathrm{Sc}=1$, because a state is a density operator and the trace of its $2\times2$ image is one. The group does not preserve $2\,\mathrm{Sc}$; it moves it by $\cosh\theta+s_{\parallel}\sinh\theta$. So the sector has a coordinate that the definition requires and the group spends.

That is the whole distinction, and it is not a defect. A state is a normalised object, so its normalisation is a convention that has to be restored after every group element; a four-momentum is not normalised, so there is nothing to restore. The asymmetry is the asymmetry between a ray and a vector.

It also explains why the material reading never needed a projection step. The corpus's material articles quote $m$ and $E$ and never divide the element by anything, because nothing was disturbed. The informational articles all divide, and the division is the reason the impurity runs.

What Follows

The running impurity is a consequence, not a second fact

Divide the image by its own $2\,\mathrm{Sc}$ and the radius obeys velocity addition. With $z=-\tanh\Theta$ and $t=\tanh\theta$, $$ z'=\frac{z+t}{1+zt}, $$ verified to $2.2\times10^{-16}$ over twelve pairs, and the read modulus is $\tanh\lvert\Theta-\theta\rvert$ to the same accuracy. The impurity follows because it is a function of the radius alone.

So the running is not an extra postulate about thermodynamics. It is what a normalised object does when a group that ignores its normalisation acts on it. The material sector has no analogue because it has no normalisation to spend.

The quantity that survives a rescaling

Off the slice $\mathcal{N}$ has a scale, and the scale-free combination is the state's own. Doubling an element multiplies $\mathcal{N}$ by four and $\mathrm{Sc}$ by two, so $$ \frac{\mathcal{N}}{\bigl(\mathrm{Sc}\,\tilde Q\bigr)^{2}} $$ is unchanged. Measured on one state under scales $0.5$, $2$ and $7.3$: $\mathcal{N}$ runs $0.018750$, $0.300000$ and $3.996750$, $2\,\mathrm{Sc}$ runs $0.5$, $2$ and $7.3$, and the ratio is $0.300000000000000$ in all four cases including the original.

This is the property that makes the impurity a property of the state and not of the representative. A state written as $2\tilde{Ϛ}_\Psi$ is the same state with a doubled scalar part, and the only reading that is blind to the doubling is the ratio. The companions use the slice because the slice is conventional; the ratio is what the conventional choice is protecting.

The Correct Statement and the Incorrect One

The inverted table

A table that puts the running quantity on $\mathcal{N}$ in the informational column has the group backwards. The congruence preserves $\mathcal{N}$ and moves $2\,\mathrm{Sc}$, in both sectors and without exception. Any account that makes the impurity the invariant and the normalisation the running quantity is describing a different group, and no such group acts here.

The inversion is easy to make because the words invite it. "Impurity" sounds like a thing that a change of frame should alter, and "norm" sounds like a thing that should be fixed; in the material sector both intuitions point the right way and in the informational sector the first one is wrong. The table is worth keeping precisely because the intuition fails on one row.

The two statements

The correct one. The frame cannot move $\mathcal{N}$. It moves the slice, and the impurity read on the re-fixed slice therefore runs.

The wrong one. A boost purifies a state, or a boost lowers its entropy, without qualification. No boost purifies anything: $\mathcal{N}$ of the element is fixed, and the only route to a lower read impurity is a different frame and not a different state.

The wrong statement is not merely imprecise; it is a category error. Purity is a property of the cone, and a boost is a symmetry of the cone. Symmetries do not change invariants, so a boost cannot change purity. What a boost changes is which state an observer holds, and that is a statement about observers.

Two Misreadings

$N=0$ is not masslessness

On the informational plane $\mathcal{N}=0$ means pure. At $\lvert\vec s\rvert=1$ the form is $0$; at $0.999$ it is $4.9975\times10^{-4}$; at $0.5$ it is $0.1875$. Purity is the cone, and the cone is reached exactly by the rank-one states.

On the material plane $\mathcal{N}=0$ means null. It is the light cone, and it is reached by $E^{2}/c^{2}=\lvert\vec p\rvert^{2}$.

The two zero sets are different subsets of different planes, written with the same equation. Confusing them produces the false statement that a pure state is trivial, or that a photon is a pure state. The corpus's own warning is that nullness is nullness for this form and is not a statement that a pure state has zero energy, zero probability or zero value of the Hermitian form — whose value on an idempotent is $\tfrac12$, verified on a rank-one projector: $\tfrac12\mathrm{tr}\,\rho=0.5$, and $\rho^{2}=\rho$.

The centre is the other fixed point and it is worth naming. The maximally mixed state has $\tfrac12\mathrm{tr}=0.5$ and $\mathcal{N}=\tfrac14$; the idempotents have $\tfrac12\mathrm{tr}=0.5$ and $\mathcal{N}=0$. The Hermitian form does not distinguish them and the quaternionic form does, which is why the second is the one the geometry is drawn with.

$N$ needs the slice or the ratio

Off the slice, $N$ alone is a form with a scale and not a property of a state. Doubling the element changes it by four and leaves the purity alone. A reading of "the impurity of this element" without a normalisation is a reading of a representative, and representatives are not unique.

The remedy is one of two, and both are in use. Either place the element on the slice $2\,\mathrm{Sc}=1$ and read $\mathcal{N}$ there, or take the ratio $\mathcal{N}/(\mathrm{Sc})^{2}$, which is scale-free. The first is the companions' convention and the second is the invariant statement; they agree on the slice and only the second survives a rescaling.

Objections

The objection is that the asymmetry is a bookkeeping artefact, since the material sector could also be normalised and the informational sector could also be left unnormalised. The asymmetry is not about the choice of convention; it is about which coordinate the group respects. Normalising the material element would add a constraint the group does not violate, so the constraint would be idle; leaving the informational element unnormalised would leave the reading undefined, since the entropy needs the normalisation. The sector's physics fixes which coordinate has to be pinned, and the group fixes whether the pinning is stable.

A second objection is that "fixed twice" says no more than "the definition happens to agree with the group". That is exactly what it says, and the phrase is used because the agreement is what a reader assumes and what fails on the other side. A statement that is true twice in one place and once in the other is the sharpest way to say where the mirror stops.

A third objection is that the framework should be able to say why the two sectors differ rather than only that they do. The difference has a one-line cause: a four-momentum is an element of the algebra and a state is a normalised element of the algebra, so one carries a constraint the group can violate and the other does not. Whether that is an explanation or a restatement is a fair question, and the article does not settle it.

What Is Not Claimed

It does not claim the two sectors are unrelated. They share the form, the group and the sign structure; what differs is which coordinate the definition pins.

It does not claim that mass runs in any frame. Mass is invariant, and $\mathcal{N}$ is preserved by the congruence in both sectors; there is no frame in which a particle's mass changes.

It does not claim that entropy is an invariant. The von Neumann entropy is a function of the spectrum, the spectrum is a function of the element, and the congruence is a similarity up to the normalisation — so the entropy of the element is fixed and the entropy of the read state moves. The distinction is the article's subject and is not collapsed anywhere in it.

It does not extend the argument beyond one element. One particle, one state, no interaction; a multi-element object would need the tensor structure and the article does not have it.

Summary

One form, two planes. $\mathcal{N}=\mathrm{Re}\sum q_k^{2}$ is $-m^{2}c^{2}$ on the anti-Hermitian plane and $\tfrac14(1-\lvert\vec s\rvert^{2})$ on the Hermitian one; the two readings are one polynomial on two subspaces.

The congruence preserves $\mathcal{N}$ and moves $2\,\mathrm{Sc}$, always. $\mathcal{N}$ deviates by at most $3.3\times10^{-16}$ over a hundred boosts, while $2\,\mathrm{Sc}-1$ has median $0.165$ and maximum $2.314$.

Mass is fixed twice. The definition puts a particle on its mass shell and the group preserves the shell, so $E/c$ runs from $1.000$ to $3.762$ at fixed mass, to $1.8\times10^{-15}$.

Normalisation is fixed once. The definition demands $2\,\mathrm{Sc}=1$ and the group moves it to $0.808317724032$ at $\theta=+0.4$, $\Theta=0.8$; the projection is forced and the radius then follows velocity addition to $2.2\times10^{-16}$.

Two statements, one of them false. The frame cannot move $\mathcal{N}$ is true. A boost purifies a state is a category error: a symmetry of the cone does not change the cone's invariant.

Two traps. $\mathcal{N}=0$ is pure on one plane and null on the other, and the Hermitian form gives $\tfrac12$ on both an idempotent and the maximally mixed state, so only the quaternionic form separates them. And $\mathcal{N}$ off the slice is a form with a scale, so the reading needs the slice or the ratio $\mathcal{N}/(\mathrm{Sc})^{2}$, which is $0.300000000000000$ under every rescaling tried.

Summary of Notation

symbol meaning
$\tilde Q=q_0e_t+\vec q$ an element of the algebra
$\mathcal{N}=\mathrm{Re}\sum_k q_k^{2}$ the one form; mass shell or impurity
$2\,\mathrm{Sc}$ twice the $e_t$ coefficient; the normalisation convention
$\tilde p_\Phi=i(E/c)e_t+\vec p$ the material momentum; anti-Hermitian plane
$\tilde{Ϛ}_\Psi=\tfrac12(e_t+i\vec s_\Psi)$ the state; Hermitian plane
$\langle\cdot\rangle_{\natural}$ the quaternionic norm
$\langle\cdot\rangle_*$ the Hermitian form; $\tfrac12$ on an idempotent
$\mathcal{N}/(\mathrm{Sc})^{2}$ the scale-free reading; $1-\lvert\vec s\rvert^{2}$ on the slice

Further Reading

  • W. R. Hamilton, Lectures on Quaternions, Hodges and Smith, 1853 — the quaternionic norm and its multiplication law
  • P. A. M. Dirac, "Application of quaternions to Lorentz transformations", Proceedings of the Royal Irish Academy 50 (1945) 261–270 — the biquaternion representation of the Lorentz group and its invariants
  • J. L. Synge, "Quaternions, Lorentz transformations, and the Conway–Dirac–Eddington matrices", Communications of the Dublin Institute for Advanced Studies 21 (1972) — the two planes and the indefinite form
  • S. L. Adler, Quaternionic Quantum Mechanics and Quantum Fields, Oxford University Press, 1995 — the quaternionic norm, its indefiniteness and its use in state space
  • M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, Cambridge University Press, 2000 — the density operator, its trace and its purity
  • A. Peres, Quantum Theory: Concepts and Methods, Kluwer, 1993 — the state as a positive operator of unit trace and what the trace condition is for
  • The Lorentz Group on the Bloch Ball: Projective Action and Two Metrics — the four-vector, the Lorentz action and $2\,\mathrm{Sc}$ as the running quantity
  • Impurity in a Boosted Frame: The Invariant and the Read Value — the projection, the read impurity and its maximum
  • Inverse Temperature as a Rapidity: Translation, Addition and the Common Bound — the thermal family and the shift law
  • Energy and Information: Four Conservation Laws from One Identity — the sector table and the two readings of the norm