The N Norm: Mass, Angle and the Cone
Introduction
This entry reads the ninth section of the mathematical category: the quaternionic norm, and the two cases of its value.
The question is what a product is allowed to change. The Hermitian norm of the previous entry is the measure that survives the largest group of motions of the algebra: left and right multiplication by the compact group leave it alone, and nothing larger does. The quaternionic norm answers the other question. It is the measure that survives multiplication itself, because it is the only one of the four with a product law,
$$ N(\tilde Q_2\tilde Q_1)=N(\tilde Q_2)\,N(\tilde Q_1). $$
A measure that a product multiplies is not a size of an element. It is the eigenvalue of a composition. The Hermitian norm of a product is bounded by the product of the Hermitian norms up to a constant and the bound is attained, so it grows with composition and is not determined by it. The quaternionic norm is determined: the norm of a product is the product of the norms, with no constant and no inequality.
The rule is one line, and it has a condition.
$$ \boxed{\;\tilde Q=\varrho\,e^{i\theta_{\mathrm{pha}}}\,\tilde U_{\mathrm b}\,\tilde u_{\mathrm r},\qquad \varrho^{2}e^{2i\theta_{\mathrm{pha}}}=N(\tilde Q),\qquad \varrho=\lvert N(\tilde Q)\rvert^{1/2},\qquad \theta_{\mathrm{pha}}\in(-\tfrac{\pi}{2},\tfrac{\pi}{2}]\;} $$
The condition is $\varrho\,e^{i\theta_{\mathrm{pha}}}\neq0$. The word divides by the modulus, and the modulus vanishes exactly on the null cone. Off the cone the element is invertible, the four factors exist, and the decomposition is a coordinate system on the group of invertibles. On the cone the modulus is zero, the word does not exist, and what survives is the two-factor pair of the previous entry. The line between the two is the zero set of one complex number.
The single idea is that the norm decides invertibility, and invertibility is the only sharp division the algebra has. Every other partition of the corpus is finer and sits inside one of these two cases. The regular case is the group of invertibles: open, dense, connected. The degenerate case is the null cone: closed, of real codimension two, the wall of the group. Mass, the massless boundary, the reversible process and the zero divisor are four readings of this one cut.
The second idea is that the norm is complex, so the modulus is a branch, and the branch is the only convention in the decomposition. A complex number has two square roots. One of them is chosen by a rule fixed once and for all, and with the rule fixed, all four factors are functions of the element. On the material sector the choice is not free, because there the norm is real: the modulus is either real or purely imaginary, and which one it is is exactly the timelike–spacelike distinction. The convention that looks arbitrary on the algebra is a sign pattern on the sector.
The entry is in three parts, and the title names them. The mass is read off the scale $\varrho$, the modulus of the norm: on the material sector $\varrho=mc$, so $m=\varrho/c$. The angle is the phase angle $\theta_{\mathrm{pha}}$ of the modulus, one number in a half-circle. The cone is the set where both disappear. The boost angle and the rotor angle belong to the word but are not named in the title, because they are motion and orientation and the next entry of the category reads them; here they are carried along so that the four factors are all present.
The strongest objection is that a complex norm cannot be a physical mass. The objection is right about the algebra and wrong about the reading. On the whole algebra the norm is a complex quadratic form, its zero set is a complex cone, and no mass is defined there. On the material sector the norm is real and indefinite, a quadratic form of inertia $(3,1)$, and its level sets are the mass shells. The mass is a level set on a sector, not a value on the algebra, and this entry says so before it uses the word.
A second objection is that a decomposition that exists only off the cone cannot describe the massless case, which is the case physics cares most about. The answer is that the entry has two decompositions and gives each its domain. The four-factor word is the description of the massive element. The two-factor pair is the description of everything, and on the cone it is the description of the massless element: a rank-one amplitude and a rank-one frame. Neither is a limit of the other, and the passage between them is the content of the two walls.
The names of the norm are several and they are one object. It is the quaternionic norm, because it is the norm of the quaternion algebra that the biquaternions complexify. It is the algebraic norm, because it is the determinant in the matrix model. It is the interval, because its level sets on the material sector are the mass shells. This entry writes $N$ and reads it as the interval when the sector is material.
Nothing here is new mathematics. The norm, its multiplicativity, the two cases, the four-factor word and the three pairings are the content of the mathematical articles of the source table. What the entry adds is the ordering: which statement is a definition, which is a theorem, and which is a reading that needs a sector before it means anything.
The Source Table
| Mathematical Content | Source Mathematical Article |
|---|---|
| the norm, its identity, its multiplicativity, the criterion of invertibility in five equivalent forms, the group of invertibles, the two strata of the cone | N-Regularity and N-Degeneracy in the Biquaternion Algebra |
| the four-factor word on the invertibles, the modulus and its branch, the closed forms, the exponential writing with the ranges of the angles, the component form and the parameter count, the seven data in the coordinates, the spectral reading, the three pairings, the characterisation of the maximal compact subgroup, and the word at the wall with the unipotent unit factor, the two strata and the boundary | The Complete Polar Decomposition of an N-Regular Biquaternion |
| the two-factor pair on the cone, the rank-one amplitude, the partial isometry, the two sectors of degeneracy | The Polar Decomposition of an N-Degenerate Biquaternion |
| the two invariants read together, the characteristic polynomial, the four sectors and the independence of the two conditions | Relations between the N Norm and Spectral Regularity and Degeneracy |
| the four forms and the four algebraic norms, the Gram matrices, and the multiplicativity of the quaternionic form alone | The Four Forms and the Four Algebraic Norms over the Biquaternion Algebra |
| the isomorphism onto the complex matrices, the determinant as the norm and the adjugate as the conjugate | $M_2(\mathbb{C})$ Pauli Representation |
| the cone as a topological object, its apex, its link, its real dimension and its contractibility | The Topology of the Zero-Divisor Cone |
| the group of invertibles as a topological group, its retraction onto the compact part and its fundamental group | The Biquaternion Group of Invertibles as a Topological Group |
The Results in Brief
- The quaternionic norm is $N(\tilde Q)=Q_t^{2}+B^{2}=\sum_\mu Q_\mu^{2}$, with $B^{2}=Q_x^{2}+Q_y^{2}+Q_z^{2}$; it is the only one of the four algebraic norms that is multiplicative.
- $N(\tilde Q_2\tilde Q_1)=N(\tilde Q_2)N(\tilde Q_1)$, and the inverse of an invertible element is $\tilde Q^{\natural}/N$; the norm is the determinant in the matrix model.
- The algebra is the disjoint union of the group of invertibles $\mathbb B^{\times}=\{N\neq0\}$ and the null cone $\mathcal N=\{N=0\}$; the group is open and dense, the cone is closed of real codimension two and is its boundary.
- The cone is the origin, the pure stratum of nilpotents of index two, and the non-pure stratum of complex multiples of idempotents; the split is by the scalar part.
- Off the cone the element has the complete polar decomposition $\tilde Q=\varrho\,e^{i\theta_{\mathrm{pha}}}\tilde U_{\mathrm b}\tilde u_{\mathrm r}$ with $\varrho=\lvert N\rvert^{1/2}>0$, $\tilde U_{\mathrm b}$ Hermitian positive of norm one and $\tilde u_{\mathrm r}$ a unit real quaternion.
- The four factors are the scale, the central phase, the boost and the rotor, of real dimensions $1,1,3,3$, summing to eight.
- The modulus is the principal square root of the norm, $\varrho^{2}e^{2i\theta_{\mathrm{pha}}}=N$ with $\varrho\,e^{i\theta_{\mathrm{pha}}}$ in the half-plane $\{\operatorname{Re}(\varrho\,e^{i\theta_{\mathrm{pha}}})>0\}\cup\{it:t>0\}$; the branch is the only convention of the decomposition.
- The scale is the geometric mean of the two eigenvalues and the angle is the mean of their phases: $\varrho=\lvert\lambda_+\lambda_-\rvert^{1/2}$ and $2\theta_{\mathrm{pha}}=\arg(\lambda_+\lambda_-)$.
- On the material sector the norm is real, so the phase is $1$ on spacelike and $i$ on timelike four-vectors, and the scale is $mc$ in the second case and the proper length in the first.
- The amplitude of the previous entry is recovered as $\tilde Q_H=\varrho\tilde U_{\mathrm b}$ and the frame as $\tilde Q_0=e^{i\theta_{\mathrm{pha}}}\tilde u_{\mathrm r}$, so the two articles are two bracketings of one word.
- On the cone the two-factor pair survives with a rank-one amplitude $\tilde Q_H=a e_t+i\vec W/(2a)$ and $\lvert\vec W\rvert=2a^{2}$; the rank is two off the cone, one on it, and zero at the origin.
- The rapidity of the boost is bounded by nothing and the rotor angle is periodic; as an element approaches the cone the scale collapses as $e^{-\theta/2}$ and the rapidity diverges as $e^{\theta/2}$, while the Hermitian norm stays finite.
- The word exists on the complement of two cones, the null cone $N=0$ and the spectral cone $B^{2}=0$; the first destroys the word, the second destroys the eigenbasis.
The Norm
The norm is one complex number and it is the whole of the first invariant.
$$ N(\tilde Q)=\langle\tilde Q\rangle_{\natural}=Q_t^{2}+Q_x^{2}+Q_y^{2}+Q_z^{2}=Q_t^{2}+B^{2}. $$
It is a quadratic form of the eight real coordinates with complex coefficients, and it is not a length. The Hermitian norm of the previous entry is a sum of squared moduli and is positive definite. The quaternionic norm is a sum of squares and is not: on the real quaternion subspace it is the Euclidean form of inertia $(4,0)$, on the pure imaginary line it is negative definite, and on the material sector it is the Minkowski form of inertia $(3,1)$. One formula, four sign patterns.
The identity of the norm is the reason it decides invertibility.
$$ \tilde Q\tilde Q^{\natural}=\tilde Q^{\natural}\tilde Q=N(\tilde Q)\,e_t, \qquad \tilde Q^{-1}=\frac{\tilde Q^{\natural}}{N(\tilde Q)}\quad(N\neq0). $$
The inverse is written down, not solved for. The quaternion conjugate is a linear map, the norm is a scalar, and the product of the two is the identity. In the matrix model the same formula is the adjugate, and the norm is the determinant,
$$ \det\mathsf M_2(\tilde Q)=N(\tilde Q),\qquad \mathsf M_2(\tilde Q^{\natural})=\varepsilon\,\mathsf M_2(\tilde Q)^{\mathsf T}\varepsilon^{-1}. $$
Multiplicativity is one line and it is the property the other three forms lack.
$$ N(\tilde Q_2\tilde Q_1)=N(\tilde Q_2)N(\tilde Q_1). $$
The proof is the identity read twice. $(\tilde Q_2\tilde Q_1)(\tilde Q_2\tilde Q_1)^{\natural}=\tilde Q_2\tilde Q_1\tilde Q_1^{\natural}\tilde Q_2^{\natural}=N(\tilde Q_1)\tilde Q_2\tilde Q_2^{\natural}=N(\tilde Q_1)N(\tilde Q_2)e_t$, and the left side is $N(\tilde Q_2\tilde Q_1)e_t$ by the identity. The same argument fails for the Hermitian norm because the $\star$-conjugate of a product reverses the order but the Hermitian norm is a sum of squared moduli and not a product of conjugates; it fails for the plain and the Krein forms for the same reason.
The norm is not additive, and the failure is the physics. For two elements $X,Y$ the cross term is $2\langle X,Y\rangle$ with $\langle X,Y\rangle=\sum_\mu X_\mu Y_\mu$ the bilinear pairing, so
$$ N(X+Y)=N(X)+N(Y)+2\langle X,Y\rangle . $$
The cross term is the interference of the two summands, and it can cancel the sum. Two elements of norm one can add to an element of norm zero, and two of norm zero can add to an element of norm two. That is what an indefinite quadratic form does, and it is why the interval and not the weight is the measure of a composition.
The criterion of invertibility has five equivalent forms, and the equivalence is the content of the first case. For $\tilde Q\neq0$:
- $N(\tilde Q)\neq0$;
- $\tilde Q$ is invertible in the plain product;
- $\tilde Q$ is not a zero divisor;
- left multiplication by $\tilde Q$ is injective;
- right multiplication by $\tilde Q$ is injective.
The five are one condition because the two-sided inverse is available. The implication from the norm to the unit is the formula for the inverse. The implication from the unit to the absence of a zero divisor is that an invertible element cannot annihilate. The implications to injectivity are the definitions read one-sidedly. And the return from injectivity uses the identity: if $N(\tilde Q)=0$ with $\tilde Q\neq0$ then $\tilde Q\tilde Q^{\natural}=0$ with $\tilde Q^{\natural}\neq0$, so the left translation kills a nonzero element. The only point where the chain needs care is the origin, which satisfies the norm condition and is neither a zero divisor nor an invertible element; the criterion is stated for $\tilde Q\neq0$ for that reason.
In the matrix model the same five forms are the invertibility of a two-by-two matrix. A biquaternion is invertible exactly when its matrix is, the determinant is the norm, and the zero divisors are the singular matrices. The whole of this entry's algebra is therefore the algebra of $2\times2$ complex matrices read in one basis, and the physics is what the basis makes visible.
The Angle
The norm is complex, so it has a modulus, and the modulus has a sign ambiguity. A modulus is a square root,
$$ \varrho^{2}e^{2i\theta_{\mathrm{pha}}}=N(\tilde Q), $$
and the two roots differ by a sign. The root is not a function of the norm until a branch is fixed. The branch that the decomposition uses places $\varrho\,e^{i\theta_{\mathrm{pha}}}$ in the half-plane
$$ \mathcal H=\{\varrho\,e^{i\theta_{\mathrm{pha}}}\in\mathbb C:\operatorname{Re}(\varrho\,e^{i\theta_{\mathrm{pha}}})>0\}\cup\{it:t>0\}, $$
with the closed form
$$ \varrho\,e^{i\theta_{\mathrm{pha}}}=\sqrt{\frac{\lvert N\rvert+\operatorname{Re}N}{2}} +i\operatorname{sgn}\bigl(\operatorname{Im}N\bigr)\sqrt{\frac{\lvert N\rvert-\operatorname{Re}N}{2}}, \qquad\operatorname{sgn}(0)=+1 . $$
The branch is the only convention in the whole decomposition. Every other object — the scale, the phase, the boost, the rotor, the two polar factors — is a function of the element once this root is fixed. Change the branch and the phase and the rotor both change sign while their product does not.
The modulus splits into two real numbers and one of them is the mass.
$$ \varrho\,e^{i\theta_{\mathrm{pha}}},\qquad \varrho=\lvert N\rvert^{1/2}>0 . $$
The scale is the modulus of the norm and the angle is its unimodular direction. In coordinates, writing $N=\lvert N\rvert e^{i\psi}$ with $\psi\in(-\pi,\pi]$, the angle is $\psi/2$ and the modulus is $\lvert N\rvert^{1/2}$. The phase therefore lives in a half-circle, $\theta_{\mathrm{pha}}=\psi/2\in(-\pi/2,\pi/2]$, and the range is not the full circle: the branch has halved it.
The halving is the square root, and it is the reason the angle is called the angle. A rotation by $2\pi$ of the norm is a rotation by $\pi$ of the modulus, and a rotation by $2\pi$ of the modulus is a rotation by $\pi$ of the phase. The phase is an angle in the double cover of the circle, exactly as the rotor angle is, and the two are independent.
On the material sector the angle is not a free parameter. There the element is $\tilde R=ir_t e_t+\vec r$ with $r_t$ and $\vec r$ real, so
$$ N(\tilde R)=-r_t^{2}+\lvert\vec r\rvert^{2}\in\mathbb R . $$
A real norm has only two directions, and they are the two sign patterns. If $N>0$ the modulus is real and positive and the phase is $1$; if $N<0$ the modulus is purely imaginary and the phase is $i$; if $N=0$ the element is on the cone and there is no phase. So on the material sector
$$ e^{i\theta_{\mathrm{pha}}}=\begin{cases}1,&N>0\quad\text{(spacelike)},\\[2pt] i,&N<0\quad\text{(timelike)},\\[2pt] \text{undefined},&N=0\quad\text{(lightlike)} .\end{cases} $$
The scale is the same expression in both cases and its reading is not. On a spacelike four-vector $\varrho$ is the proper length of the displacement. On a timelike four-vector $\varrho$ is $mc$, because $N=-m^{2}c^{2}$ and $\varrho=mc$; the mass is $m=\varrho/c$. One modulus, two names, and the name is decided by the sign of the norm.
The consequence is that matter sits at the endpoint of the phase interval. The phase of a timelike four-vector is exactly $i$, which is the boundary value $\psi/2=\pi/2$ of the half-circle. It is not an interior point of the range that matter happens to occupy. It is the endpoint, and the endpoint is reached because the norm is real and negative, which is the definition of timelike.
The angle is therefore a sign label on the sector and a free parameter off it. This is the sharpest difference between the two readings of one complex number. On the algebra the phase is one continuous parameter of the group of invertibles. On the material sector it takes two values, and the two values are the two halves of the light cone, and the third case is the cone itself.
The Two Cases
The value of the norm separates the algebra into two sets, and the separation is a partition.
| condition on $N$ | set | name |
|---|---|---|
| $N\neq0$ | $\mathbb B^{\times}=\{N\neq0\}$ | the regular case: the group of invertibles |
| $N=0$ | $\mathcal N=\{N=0\}$ | the degenerate case: the null cone |
The two sets have opposite topology and that is the point. The group of invertibles is open and dense. The cone is closed, of real codimension two, and it is the boundary of the group. An element is either invertible or it is not, and there is no third state: the algebra is the disjoint union of the two.
The cone has an apex and two strata, and the split is by the scalar part.
$$ \mathcal N=\{0\}\uplus\mathcal N_{\mathrm p}\uplus\mathcal N_{\mathrm{np}},\qquad \mathcal N_{\mathrm p}=\{Q_t=0,\ \tilde Q\neq0\},\qquad \mathcal N_{\mathrm{np}}=\{Q_t\neq0\}. $$
On the cone the element obeys a rank-one law.
$$ \tilde Q^{2}=2Q_t\tilde Q\qquad(\tilde Q\in\mathcal N), $$
and the two strata are the two ways of reading it. At $Q_t\neq0$ the law is an idempotent equation after rescaling: $\tilde Q/(2Q_t)$ is an idempotent, and the element is a complex multiple of a nontrivial idempotent. At $Q_t=0$ the law is $\tilde Q^{2}=0$: the element is a nilpotent of index two, and it is its own annihilator. The equivalence is exact,
$$ Q_t=0\iff\tilde Q^{2}=0\iff B^{2}=0\qquad(\tilde Q\in\mathcal N), $$
so the two strata are also the two values of the second invariant on the cone.
The cone is stable under the product and is not a subspace. Stability is multiplicativity: a product of two zeros of the norm has norm zero. Not a subspace is the cross term: two null elements add to a null element only when they are proportional, and a check on a hundred random pairs of null elements found a non-null sum in every case.
The norm vanishes nowhere else, and that is a small but useful fact. On the centre $N(\lambda e_t)=\lambda^{2}$, so the only zero is $\lambda=0$. On the real quaternion subspace the norm is a sum of four squares of real numbers and vanishes only at $0$. On the vector subspace $N(\vec Q)=B^{2}$ with $B^{2}$ a sum of three squares, so again only at $0$ over the reals. The cone is carried by the mixed subspaces, where one coordinate is imaginary and another is real.
The two cases are therefore not two kinds of element but two values of one function. This matters for the reading: the massless case is not a limiting species of particle that the algebra distinguishes by a label. It is the zero level of the interval, and the particle content on that level is imported.
The Word
On the invertibles the two-factor pair refines into four factors, and the refinement is the content of the regular case.
$$ \boxed{\;\tilde Q=\varrho\,e^{i\theta_{\mathrm{pha}}}\,\tilde U_{\mathrm b}\,\tilde u_{\mathrm r} =\varrho\,e^{i\theta_{\mathrm{pha}}}\, \exp\Bigl(i\frac{\theta_{\mathrm{rpd}}}{2}\vec u_{\mathrm b}\Bigr) \exp\Bigl(\frac{\theta_{\mathrm{axr}}}{2}\vec u_{\mathrm r}\Bigr)\;} $$
with the ranges
$$ \varrho>0,\qquad \theta_{\mathrm{pha}}\in\Bigl(-\frac{\pi}{2},\frac{\pi}{2}\Bigr],\qquad \theta_{\mathrm{rpd}}\ge0,\qquad \theta_{\mathrm{axr}}\in[0,2\pi], $$
and $\vec u_{\mathrm b},\vec u_{\mathrm r}$ real unit three-vectors, the boost axis and the rotor axis.
| factor | closed form | constraint | real dimension | type |
|---|---|---|---|---|
| scale $\varrho$ | $\varrho$ | $\varrho>0$ | $1$ | positive real, central |
| phase | $e^{i\theta_{\mathrm{pha}}}$ | $\theta_{\mathrm{pha}}\in(-\tfrac{\pi}{2},\tfrac{\pi}{2}]$ | $1$ | unit complex scalar, central |
| boost $\tilde U_{\mathrm b}$ | $\cosh\frac{\theta_{\mathrm{rpd}}}{2}e_t+i\sinh\frac{\theta_{\mathrm{rpd}}}{2}\vec u_{\mathrm b}$ | $N(\tilde U_{\mathrm b})=1$ | $3$ | Hermitian positive, norm one |
| rotor $\tilde u_{\mathrm r}$ | $\cos\frac{\theta_{\mathrm{axr}}}{2}e_t+\sin\frac{\theta_{\mathrm{axr}}}{2}\vec u_{\mathrm r}$ | $N(\tilde u_{\mathrm r})=1$ | $3$ | unit real quaternion |
The four dimensions add to eight, and the addition is a theorem. $1+1+3+3=8$, the real dimension of the algebra. The scale and the phase are central and commute with everything. The boost and the rotor do not commute, and the order of the word is part of the definition: the boost stands on the left.
The word is a coordinate system on the group of invertibles, and the two axes are points of the sphere. The parameter count is the dimension of $\mathbb B^{\times}$, and the ranges of the angles cut the chart. The chart is not global in the angle of the phase, because the branch cuts the circle; it is global in everything else.
The four factors group into the two factors of the previous entry in one way, and into a second pair in another.
$$ \tilde Q_H=\varrho\,\tilde U_{\mathrm b},\qquad \tilde Q_0=e^{i\theta_{\mathrm{pha}}}\,\tilde u_{\mathrm r},\qquad \varrho\,e^{i\theta_{\mathrm{pha}}},\qquad \tilde U=\tilde U_{\mathrm b}\tilde u_{\mathrm r}. $$
The first bracketing is the Cartan pair and the second is the modulus–unit pair. The Cartan pair is Hermitian positive times $\star$-unitary; the modulus–unit pair is the complex modulus times the norm-one element. The first is the pair the previous entry studied on the whole algebra. The second exists only here, because it divides by the modulus.
The two bracketings answer different questions and neither is finer. The Cartan pair separates the size from the orientation. The modulus–unit pair separates the complex scaling from the unimodular part. The first is the decomposition of the algebra and the second is the decomposition of the group, and on the units they carry the same information.
Two worked elements show the four factors separated.
| element | $N$ | $B^{2}$ | $\varrho$ | $e^{i\theta_{\mathrm{pha}}}$ | $\theta_{\mathrm{rpd}}$ | sector |
|---|---|---|---|---|---|---|
| $e_t+\vec e_x$ | $2$ | $1$ | $\sqrt2$ | $1$ | $0$ | regular, regular |
| $e_t+\vec e_x+i\vec e_y$ | $1$ | $0$ | $1$ | $1$ | $\operatorname{arccosh}3$ | regular, degenerate |
| $2ie_t+\vec e_x$ | $-3$ | $1$ | $\sqrt3$ | $i$ | $\operatorname{arccosh}\tfrac53$ | regular, regular |
| $(1+i)e_t$ | $2i$ | $0$ | $\sqrt2$ | $(1+i)/\sqrt2$ | $0$ | regular, degenerate |
| $e_t+i\vec e_x$ | $0$ | $-1$ | — | — | — | degenerate, regular |
| $\vec e_x+i\vec e_y$ | $0$ | $0$ | — | — | — | degenerate, degenerate |
The first line is a pure rotor. The norm is real and positive, the phase is one, the rapidity is zero, and the whole of the element is a scale and a rotation by $\pi/2$ about $\vec e_x$. It is the case in which the word is the quaternion word.
The second line is the case the eigenbasis cannot see. The norm is one, so the element is invertible and the word exists; but $B^{2}=0$, so the two eigenvalues coincide and the element is defective. The rapidity is $\operatorname{arccosh}3$, finite and positive: the boost is real and the word is perfectly well defined, while the spectral decomposition of the same element has collapsed. This is the sharpest single illustration in the entry of the difference between the two decompositions, and it is the subject of the next entry but two.
The third line is matter at rest in a boosted frame. The norm is real and negative, the phase is $i$, and the scale is $mc$: $\varrho=\sqrt3$ with $N=-3$, so $mc=\sqrt3$ and $m=\sqrt3/c$ in the units of the row. The rapidity is $\operatorname{arccosh}(5/3)$, which is not zero, so the mass is not at rest in this frame. The scale is $mc$ no matter what the boost does, and that is the content of the next section.
The last two lines are on the cone. The word does not exist, and the two lines differ in the second invariant: one is spectrally regular and one is not. They are the two sectors of the degenerate case, and they are read in the section after the walls.
The Two Walls
The word exists on the complement of two cones, and the two cones fail differently.
| wall | equation | what happens | what survives |
|---|---|---|---|
| null cone | $N=0$ | $\varrho\to0$, $\theta_{\mathrm{rpd}}\to\infty$ | the two-factor pair, rank one |
| spectral cone | $B^{2}=0$ | the eigenvalues coalesce, $\vec Z$ diverges | the word, and the four factors |
The null cone is where the modulus vanishes and the word has no domain. An element of the cone has no scale and no phase, because the number they are made from is zero. As an element approaches the cone from the regular side the scale and the rapidity move in opposite directions and their product stays finite:
$$ \tilde Q(\theta)=K\,e^{-\theta/2}\,\omega\, \exp\Bigl(i\frac{\theta}{2}\vec u_{\mathrm b}\Bigr)\tilde u_{\mathrm r} \longrightarrow \frac{K\omega}{2}\bigl(e_t+i\vec u_{\mathrm b}\bigr)\tilde u_{\mathrm r}, \qquad\theta\to\infty, $$
with $K>0$ a constant and $\lvert\omega\rvert=1$. Every member of the family is invertible with scale $Ke^{-\theta/2}$ and rapidity $\theta$. The limit is a nonzero null element of rank one. The scale vanishes as $e^{-\theta/2}$ and the boost diverges as $e^{\theta/2}$, so the element converges while both of its factors diverge in opposite directions.
The Hermitian norm does not diverge and does not vanish. Along the same family, in the notation of the previous entry,
$$ H(\tilde Q)=\varrho^{2}\cosh\theta_{\mathrm{rpd}}=K^{2}e^{-\theta}\cosh\theta\longrightarrow\frac{K^{2}}{2}, $$
a finite limit that is the trace of the rank-one projection. So one measure collapses, one diverges, and one stays finite and tends to $K^{2}/2$. The three behaviours are the three measures, and the finiteness of the middle one is why the two-factor pair survives the wall.
The spectral cone is a different failure. At $B^{2}=0$ the two eigenvalues coincide, the minimal polynomial loses a factor, and the element becomes defective unless it is a scalar. The root
$$ \vec Z=-i\frac{\vec Q}{B} $$
used to write the spectral idempotents diverges as $B\to0$, so no spectral idempotent has a limit. The word, however, does: the modulus is continuous in the norm, the norm is continuous, and the Cartan factors are continuous functions of the unit factor. The polar data pass through the spectral wall; the spectral data do not.
The two cones meet on the nilpotent stratum, and there the element is degenerate twice. That stratum is the origin together with the pure nilpotents, and it is the only place where both decompositions fail. It is also the only place where the cone and the spectral cone are the same set.
The walls are independent, and the entry uses that independence. An element can be regular in the norm and degenerate in the spectrum, or the reverse, and the table of worked elements contains one of each. The four combinations are the four sectors of the category, and the next entry reads them as a partition.
The words "the massless limit" name the first wall and not the second. The limit is a limit of the norm to zero, and the family above shows what it does: the element survives, the amplitude survives, the frame loses a rank, and the four-factor word does not exist at the limit. Nothing about the spectrum is involved. An element can be spectrally degenerate and perfectly massive, as the second line of the table shows.
The Massless Pair
On the cone the two-factor pair of the previous entry is the description, and its factors are rank one.
$$ \tilde Q=\tilde Q_H\tilde Q_0,\qquad \tilde Q_H=\bigl(\tilde Q\tilde Q^{*}\bigr)^{1/2}=a\,e_t+\frac{i}{2a}\vec W, \qquad \tilde Q_0=\tilde Q_H^{+}\tilde Q, $$
with $a=\sqrt{H(\tilde Q)/2}$ and $\lvert\vec W\rvert=2a^{2}$. The amplitude is Hermitian positive semidefinite of rank one, the frame is a partial isometry of rank one, and both are functions of the element. At the origin both vanish.
The amplitude is twice a scale times a minimal projection.
$$ \tilde Q_H=2a\,\tilde\Pi_{\vec u},\qquad \vec u=\frac{\vec W}{\lvert\vec W\rvert}, \qquad \tilde Q_0=\frac{\tilde Q}{2a}, $$
and the frame is the element divided by the same scale. The scale here is not the modulus; the modulus is zero. It is the square root of the Hermitian norm, $2a=\sqrt{2H}$, and it is the only scale on the cone.
The two invariants of the cone are the two strata.
$$ \tilde Q=2a\,\tilde\Pi_{\vec u}\quad(Q_t\neq0) \qquad\text{and}\qquad \tilde Q=\vec Q\quad(\tilde Q^{2}=0), $$
an idempotent multiple and a nilpotent. The first is a projection times a complex scalar; the second is a complex isotropic vector. They are both null, both rank one, and not the same kind of object: the first squares to a multiple of itself, the second squares to zero.
The rank of the amplitude is the mass label, and it is discrete.
| element | rank of $\tilde Q_H$ | $N$ | reading |
|---|---|---|---|
| regular | $2$ | $\neq0$ | massive |
| on the cone, nonzero | $1$ | $0$ | massless |
| the origin | $0$ | $0$ | the vacuum |
The rank cannot be one and a half. There is no element whose amplitude is almost rank one, because rank is an integer and the cone is the exact set where it drops. The distinction between massive and massless is therefore not a value of a continuous parameter crossing a threshold: it is a change of rank, and the change happens on a set of measure zero.
The frame on the cone has half the norm of a frame. In the notation of the previous entry, $H(\tilde Q_0)=\tfrac12$ for a nonzero element of the cone, because the frame is a minimal partial isometry and twice the scalar part of a minimal projection is one while twice the scalar part of the unit is two. The missing half is the rank that the cone removed.
The pair on the cone is the continuous limit of the pair off it, in one factor and not the other. The amplitude is continuous everywhere, including across the cone and at the origin. The frame is not: its limit along the family of the previous section is the identity, and its value on the cone is the rank-one projection. The discontinuity is forced by the uniqueness conditions and is not a defect of the formula.
The Three Pairings
The same four factors can be bracketed in three ways, and each bracketing is a different answer to the question what the magnitude here is.
| pairing | first factor | second factor | available when |
|---|---|---|---|
| quaternion | $\varrho\tilde u_{\mathrm r}$ | $e^{i\theta_{\mathrm{pha}}}\,\tilde u_{\mathrm r}^{-1}\tilde U_{\mathrm b}\tilde u_{\mathrm r}$ | $N\neq0$ and the real-coefficient part is nonzero |
| Hamilton root | $\varrho\,e^{i\theta_{\mathrm{pha}}}$ | $\tilde U_{\mathrm b}\tilde u_{\mathrm r}=\cos\Theta+\vec U\sin\Theta$ | $N\neq0$ and $B^{2}\neq0$ |
| Cartan Hermitian | $\varrho\tilde U_{\mathrm b}$ | $e^{i\theta_{\mathrm{pha}}}\tilde u_{\mathrm r}$ | $N\neq0$ |
The Cartan pairing is the only one defined on the whole group of invertibles, and the reason is that it divides by nothing. It takes the Hermitian square $\tilde Q\tilde Q^{*}$, takes the positive square root in the cone, and divides by that root. All three steps exist for every invertible element. The other two pairings divide: the quaternion pairing by the real-coefficient part, the Hamilton pairing by $B$.
The Hamilton pairing is the polar form of a complex number, written for a biquaternion.
$$ \tilde Q=\varrho\,e^{i\theta_{\mathrm{pha}}}\bigl(\cos\Theta+\vec U\sin\Theta\bigr), \qquad \cos\Theta=\frac{Q_t}{\varrho\,e^{i\theta_{\mathrm{pha}}}},\qquad \sin\Theta=\frac{B}{\varrho\,e^{i\theta_{\mathrm{pha}}}},\qquad \vec U=\frac{\vec Q}{B}, $$
and $\Theta$ is the complex angle of the element. Its real part is half the rotor angle and its imaginary part is half the rapidity when the two axes agree; when they do not agree the angle is complex and the axes are two. The name root pairing records that the modulus is a square root of the norm: the pairing is written in the variable $\varrho\,e^{i\theta_{\mathrm{pha}}}$, not in $N$.
The two partial domains are not nested, and their union misses one class. The quaternion pairing fails where the real-coefficient part vanishes; the Hamilton pairing fails where $B^{2}=0$. Both fail together exactly on the nonzero pure imaginary scalars,
$$ \tilde Q=i c\,e_t\quad(c\in\mathbb R,\ c\neq0), \qquad N=-c^{2}\neq0,\qquad Q_t=ic,\qquad B^{2}=0, $$
which are invertible with no real part and no spectral direction. They are the exception that a reader who assumes one pairing is general will miss, and the Cartan pairing covers them.
The three pairings are three notions of magnitude and they are not interchangeable. The modulus $\varrho\,e^{i\theta_{\mathrm{pha}}}$ is a square root of the product of the two eigenvalues. The norm $N$ is that product without the root. The Hermitian factor $\varrho\tilde U_{\mathrm b}$ is the amplitude of the previous entry. A text that says the magnitude of the biquaternion without saying which pairing is in use says nothing, and this is the reason the corpus writes the letter: $N$ for the norm, $H$ for the weight, $\varrho\,e^{i\theta_{\mathrm{pha}}}$ for the modulus.
The Group of Invertibles
The regular case is a group and the decomposition is a coordinate on it.
$$ \Phi:\mathbb B^{\times}\longrightarrow\mathcal H\times SL_2(\mathbb C), \qquad \Phi(\tilde Q)=(\varrho\,e^{i\theta_{\mathrm{pha}}},\tilde U), $$
and $\Phi$ is a bijection. Injectivity is uniqueness of the pair; surjectivity is that $\varrho\,e^{i\theta_{\mathrm{pha}}}\tilde U$ has norm $\varrho^{2}e^{2i\theta_{\mathrm{pha}}}$ with principal root $\varrho\,e^{i\theta_{\mathrm{pha}}}$. In the matrix model it is the factorisation of an invertible matrix into a scalar and a matrix of determinant one.
The norm-one part is the special linear group, and the compact part of it is a sphere and a circle.
$$ \mathbb B^{\times}_{1}=\{\tilde U:N(\tilde U)=1\}\cong SL_2(\mathbb C), \qquad U(2)=\{e^{i\theta_{\mathrm{pha}}}\tilde u_{\mathrm r}:\theta_{\mathrm{pha}}\in\mathbb{R},\ \tilde u_{\mathrm r}\in SU(2)\}. $$
The dimension count is the arithmetic of the four factors. The scale and the phase are two real parameters; the norm-one equation is one complex equation, so it removes two; six remain, which is the real dimension of $SL_2(\mathbb C)$. Inside those six, the Hermitian positive elements of norm one form a hyperbolic space of dimension three and the rotor is a sphere of dimension three.
The group retracts onto $U(2)$, and the retraction is the massless limit done slowly. Shrinking the scale to one and pushing the boost to the identity along the Hermitian positive elements is continuous. So every invertible element is homotopic to a unitary one, and the topology of the group is the topology of $U(2)$.
The fundamental group is the integers and the generator is the phase.
$$ \pi_1(\mathbb B^{\times})\cong\mathbb Z, \qquad \pi_3(\mathbb B^{\times})\cong\mathbb Z, \qquad \pi_2(\mathbb B^{\times})=0 . $$
The generator of $\pi_1$ is the loop of the phase around the cone, and the sphere of the rotor is simply connected. The boost adds no loops, because hyperbolic space is contractible. So the winding is carried by the one compact central direction, and not by the spin.
Nothing in this section is dynamics. The group of invertibles is the set of elements that have an inverse, the retraction is a homotopy, and the winding number is a class of loops. The physical readings are in the interpretations, and each carries its boundary with it.
The Interpretations
I.9.a — The Norm Is the Mass Squared, and the Sector Supplies the Sign
Read: on the material sector, with the four-momentum written $\tilde P=ip_0e_t+\vec p$, real $p_0,\vec p$ and $p_0=E/c$,
$$ N(\tilde P)=-p_0^{2}+\lvert\vec p\rvert^{2}=-m^{2}c^{2}, $$
so the mass shell is the level set $N=-m^{2}c^{2}$ and the mass is $\lvert N\rvert^{1/2}/c$.
The sign is not a convention and it is not decoration. It comes from one complex coordinate in the timelike slot, and it is the same sign that makes the material sector anti-Hermitian. A real four-vector has $N=p_0^{2}+\lvert\vec p\rvert^{2}$, a sum of squares, and no level set of it is a mass shell. The shell exists because one coordinate is imaginary, and that is the whole of the mechanism.
The physical interpretation is that mass is a level set of a quadratic form and not a value of a field. The statement is weak and it is the only one the algebra supports: it says which surfaces the four-momenta of a given mass occupy, not which masses occur. Its strength is in its universality: every element of the sector has a value of $N$, the level sets foliate the sector, and the sign of the value separates the timelike from the spacelike.
The boundary. The mass spectrum is not derived, the level sets are not quantised, and the identification of $\vec p$ with a physical momentum is a modelling choice. On the whole algebra $N$ is complex and the word mass has no reading.
I.9.b — Two Cases and No Third: the Wall of Invertibility
Read: $\mathbb B=\mathbb B^{\times}\uplus\mathcal N$, the group open and dense, the cone closed of real codimension two.
A property decided by one number is a property that cannot be graded. Invertibility is not a matter of degree: the determinant is either zero or it is not, and the set where it is zero has measure zero. There is no element that is nearly invertible in the algebraic sense, and no process that is nearly reversible.
The physical interpretation is that reversibility is a wall and not a slope. A process is invertible in the algebra or it is a zero divisor, and an irreversible one is one that annihilates something. An irreversible process has a kernel, and a kernel in this algebra is at least a complex line.
The boundary. This is a statement about elements and their products, not about time. Nothing in the algebra says that a physical process is described by an element, and nothing forbids an element from being both invertible and thermodynamically irreversible. The reading is structural.
I.9.c — The Modulus Is the Geometric Mean of the Two Eigenvalues
Read: with $\lambda_\pm=Q_t\pm iB$,
$$ N=\lambda_+\lambda_-, \qquad \varrho=\lvert N\rvert^{1/2}=\bigl(\lvert\lambda_+\rvert\,\lvert\lambda_-\rvert\bigr)^{1/2}. $$
The scale is the geometric mean of the two eigenvalue moduli and the norm is their product. The two eigenvalues are the two spectral values of the element, and the mass is a symmetric function of them: it is what they share. A change of frame multiplies the two eigenvalues in opposite directions and leaves the product alone.
The physical interpretation is that the invariant mass of an element is the part of its spectrum that no frame can move. The product of two frequencies is invariant under the boost that Doppler-shifts them, because the shift is reciprocal. This is the algebraic form of the statement that the mass of a system is not the sum of its parts' masses but the invariant of the pair.
The boundary. The two eigenvalues belong to one element and are not two particles. Statements about a pair of modes are statements about one biquaternion's spectrum, and the entry that reads them as a two-body system is later in the category.
I.9.d — The Angle Is the Mean Phase
Read: $\theta_{\mathrm{pha}}=\tfrac12\arg(\lambda_+\lambda_-)=\tfrac12\bigl(\arg\lambda_++\arg\lambda_-\bigr)$.
The phase of the modulus is the mean of the two eigenphases, and the halving is the square root. The angle is therefore not an independent datum: it is the average of the two phases of the spectrum, read through the branch that halves it.
The physical interpretation is that the angle is a relative phase and not an absolute one. Only the sum of the two eigenphases enters, so the angle does not measure a common rotation of the two eigenvalues; it measures the phase that the element carries after the scale has been removed. A global phase is not in the word at all.
The boundary. The mean of two phases is defined modulo $\pi$, and the branch is what turns the class into a number. Statements about the sign of $\theta_{\mathrm{pha}}$ therefore depend on the branch, and the branch is a convention.
I.9.e — On Matter the Angle Is Not Free: It Is the Sign
Read: on the material sector $N\in\mathbb R$, so $e^{i\theta_{\mathrm{pha}}}=1$ for $N>0$, $e^{i\theta_{\mathrm{pha}}}=i$ for $N<0$, and there is no phase for $N=0$.
A free parameter of the algebra becomes a label on the sector. The phase of a biquaternion is one continuous parameter of the group of invertibles. On the material sector it takes exactly two values, and the two values are the two halves of the light cone. The complex structure of the algebra is, on that sector, the choice between a length and a mass.
The physical interpretation is that the difference between a displacement and a four-momentum is a phase of $i$. The same quadratic form reads as a proper length on a spacelike vector and as minus a mass squared on a timelike one, and the object that changes between the two readings is the phase of the modulus. This is the precise sense in which the mass is imaginary in a metric of inertia $(3,1)$: the modulus of the interval is $im$ and not $m$.
The boundary. The phase is a property of the element's norm and not of a state. The reading is available only on the material sector, and it needs the sector before it can be stated. Nothing is claimed about the phase of a general biquaternion, which is continuous and unconstrained.
I.9.f — The Scale Is $mc$ and the Boost Is the Motion
Read: on a timelike four-momentum $\tilde Q_H=\varrho\tilde U_{\mathrm b}$ with $\varrho=mc$, and $H=\varrho^{2}\cosh\theta_{\mathrm{rpd}}$.
The polar factor of the previous entry separates into a scale and a boost, and the two are multiplied. The scale is $mc$ and the boost is the motion, and the Hermitian norm is the product of the two, with the factor $\cosh\theta_{\mathrm{rpd}}$ carrying how much the motion has inflated the amplitude.
The physical interpretation is that an amplitude is not a mass. The amplitude of the previous entry is mass times boost, and it equals the mass only in the frame where the element is at rest. The previous entry said this with the Hermitian norm; this entry says it with the norm, and the two statements are one: $\lvert N\rvert=\varrho^{2}$ and $H=\varrho^{2}\cosh\theta_{\mathrm{rpd}}$, so the ratio of the amplitude to the mass is the Lorentz factor.
The boundary. The identity $\varrho=mc$ uses $N=-m^{2}c^{2}$, and it is a reading of the sector. That the scale is $mc$ in every frame is a property of the decomposition and not a dynamical theorem.
I.9.g — The Massless Limit Is a Limit of the Inverse Parameters
Read: the family $\tilde Q(\theta)=Re^{-\theta/2}\omega\exp(i\theta\vec u_{\mathrm b}/2)\tilde u_{\mathrm r}$ converges to a null element while $\varrho\to0$ and $\theta\to\infty$.
The element has a limit and its factors do not. The scale collapses and the rapidity diverges, and the product stays finite because the divergence is exponential in both directions. The Hermitian norm converges to a finite value, so the amplitude of the previous entry converges too; only the word fails, and it fails at the limit and not before it.
The physical interpretation is that the massless limit is a limit of a product and not of a parameter. A massless particle is not a particle with a small mass: the limit is taken along a family in which the mass and the boost move together, and no member of the family is massless. The statement that a massless particle has no rest frame is the statement that the rapidity is unbounded, read at the limit. For the material momentum the data are $K=2E/c$, $\omega=i$, $\vec u_{\mathrm b}=-\vec p/\lvert\vec p\rvert$, $\tilde u_{\mathrm r}=e_t$ and $\theta=2\theta_{\mathrm{rpd}}$, and the limit is $\tilde P_\Phi$ itself to a relative defect $e^{-2\theta_{\mathrm{rpd}}}$.
The boundary. The family is one path to the cone, and the limiting element depends on the path through the direction and the rotor. The statement is about limits of a family and not about a canonical degeneration.
I.9.h — Rank Is the Mass Label
Read: the amplitude $\tilde Q_H$ has rank two off the cone, rank one on the nonzero cone, and rank zero at the origin; in the matrix model the norm is the determinant.
Massless means singular. The distinction between a massive and a massless element is the distinction between an invertible and a singular matrix, and the determinant is the norm. There is no third case and no intermediate: the rank is an integer.
The physical interpretation is that the mass shell is where the amplitude loses a dimension. The amplitude of a massive element is a two-dimensional positive element with two spectral values; on the cone one of the two vanishes and the amplitude becomes a projection on a line. The mass is a rank, and a rank is not a quantity that can be tuned.
The boundary. Rank does not count states and does not distinguish the massless particles from one another. It is a property of one element, and any statement about a spectrum of masses is imported.
I.9.i — A Lightlike Momentum Is Not a Nilpotent
Read: on the cone $\tilde Q^{2}=2Q_t\tilde Q$, and $Q_t=0\iff\tilde Q^{2}=0$.
The two strata are two different objects with the same norm. The pure stratum is the nilpotents, the elements that annihilate themselves in one product. The non-pure stratum is the complex multiples of idempotents, the elements that rescale themselves. Both are null and only the second has a scalar part.
The physical interpretation is that the physical null momentum is an imaginary multiple of an idempotent and not a nilpotent. The four-momentum of a massless particle is $\tilde P=ip_0(e_t+\vec u)$ with $p_0>0$: its scalar part is $ip_0$, it is null, and it is an imaginary multiple of the projection onto the lightlike direction. The nilpotent is a different object on the same cone: a complex isotropic vector that squares to zero, useful as a model of a null direction that erases itself, and not a momentum.
The boundary. The three objects — the nilpotent, the imaginary idempotent multiple and the real lightlike displacement — live on the same cone and are not the same element. A text that calls the cone the nilpotent cone is using the name of one stratum for the whole, and the physics of null momenta lives on the other.
I.9.j — Three Pairings Are Three Notions of Magnitude, and Two Have Holes
Read: the Cartan pairing is total on the invertibles; the quaternion pairing needs a nonzero real part; the Hamilton pairing needs $B^{2}\neq0$.
The pairing defined everywhere is the one that divides by nothing. The other two are partial and their domains are different. The Hamilton pairing fails exactly at the spectral cone, which is where the two eigenvalues coalesce; there is no direction $\vec U$ and no angle $\Theta$ when the spectrum is degenerate.
The physical interpretation is that the polar form of a spinor is not available at the exceptional point. A formula that writes an element as a modulus times a phase and an axis presupposes a direction, and the exceptional point has none. Any argument that uses the Hamilton form and then specialises to the degenerate case uses a formula outside its domain, and the Cartan form is the one that survives.
The boundary. These are domain statements about three decompositions of one word. They do not rank the pairings physically: each is the right tool where it exists, and the Cartan one is right precisely because its domain is the whole group.
I.9.k — The Branch Cut Lies on Timelike Matter
Read: the modulus is the root in the half-plane $\mathcal H$; across the negative real axis of the norm the phase and the rotor both change sign, while the Cartan pair does not.
The cut is the set of timelike four-momenta. On the material sector the norm is real, so the plane of the norm is the real line, and the negative half of it is exactly $N=-m^{2}c^{2}<0$. The four-factor word is therefore discontinuous on the objects physics calls matter, and continuous on everything else.
The physical interpretation is that the flip is the double cover and not a discontinuity of the element. The flip sends $e^{i\theta_{\mathrm{pha}}}\to-e^{i\theta_{\mathrm{pha}}}$ and $\tilde u_{\mathrm r}\to-\tilde u_{\mathrm r}$ with the product fixed. A rotor has this freedom by construction: its group is $SU(2)$ and the flip is the rotation by $2\pi$. So the cut is where a spinor changes sheet, and the two-factor pair, which is blind to it, is the description in which nothing happens.
The boundary. The cut is a convention of the branch and not a singularity of the algebra. Nothing physical changes across it, and the entry states the fact so that no later argument mistakes the convention for a discontinuity.
I.9.l — Mass Is Multiplicative and Amplitude Is Additive
Read: $N(\tilde Q_2\tilde Q_1)=N(\tilde Q_2)N(\tilde Q_1)$, while the Hermitian norm obeys $H(\tilde Q_2\tilde Q_1)\le2H(\tilde Q_2)H(\tilde Q_1)$.
One measure multiplies under composition and the other does not. The interval is a law: the modulus of a composite is the product of the moduli of its factors, up to the phase that the product of the two carries. The weight is not a law: it is bounded by the product and the bound is attained.
The physical interpretation is that the cone is absorbing. A null factor makes the product null whatever the other factor is, because the product of the norms is zero. This is the algebraic content of the statement that a massless constituent makes the composite massless in the sense of the norm, and it is a statement about composition and not about decay.
The boundary. Multiplicativity is not conservation in time, and the product law is a law of the algebra. The phase of a product is not the sum of the phases: it is the sum up to a sign, because the modulus is a square root, and the sign is the branch again.
The Limits
The norm is not a mass on the algebra. It is a complex quadratic form, its zero set is a complex cone, and the reading as a mass requires the material sector and the sign of the value. Outside the sector the same formula gives a number with a phase, and no mass is defined there.
The modulus is not a function of the norm until the branch is fixed. Two conventions differ by a sign on both the phase and the rotor, and the sign is invisible in the product. Every statement of this entry that uses $\varrho\,e^{i\theta_{\mathrm{pha}}}$ inherits the convention, and a statement that does not survive the change of branch is a statement about the convention.
The norm does not determine the spectrum. The element $e_t$ and the element $\vec e_x$ both have norm one; the first has $B^{2}=0$ and the second $B^{2}=1$, and their minimal polynomials are $X-1$ and $X^{2}+1$. So the mass is coarser than the spectrum, and no argument may infer the eigenvalues from the norm.
Multiplicativity is not conservation. The product law says how the norm of a product is computed from the norms of the factors. It says nothing about a process, a rate, or a time. A conservation law requires a dynamics, and the algebra supplies the identity and not the dynamics.
The cone is not a particle content. The algebra marks the massless locus and says nothing about how many particles sit on it, with what spin, or with what multiplicity. Every such statement is imported, and this entry imports none.
The rank is not a state count. Rank two, one and zero are ranks of the amplitude of one element. They do not count excitations and they do not count the degrees of freedom of a field.
The group of invertibles has a topology and the topology has no coupling. That $\pi_1$ is the integers is a fact of the group. It is not a charge, because a charge requires a field to couple to and the algebra contains no field.
The word is defined on the invertibles and its continuity stops at the wall. The factors diverge in opposite directions along the family that reaches the cone, and the element does not. No claim in this entry extends a four-factor formula to the cone by continuity, because at the cone the formula has no value.
The four factors are not four observables. They are four coordinates of one element. A choice of coordinates is not a measurement, and the scalar nature of the algebra forbids the identification of a coordinate with a number read off a device.
Summary
The quaternionic norm is the multiplicative measure of the algebra; its vanishing is the wall between the group of invertibles and the null cone, and its modulus carries the mass and the angle.
| item | result |
|---|---|
| the norm | $N(\tilde Q)=Q_t^{2}+B^{2}=\sum_\mu Q_\mu^{2}=\langle\tilde Q\rangle_{\natural}$ |
| multiplicativity | $N(\tilde Q_2\tilde Q_1)=N(\tilde Q_2)N(\tilde Q_1)$, the only one of the four |
| the identity and the inverse | $\tilde Q\tilde Q^{\natural}=Ne_t$, $\tilde Q^{-1}=\tilde Q^{\natural}/N$, $\det\mathsf M_2(\tilde Q)=N$ |
| the split | $\mathbb B=\mathbb B^{\times}\uplus\mathcal N$, open and dense against closed of real codimension two |
| the cone | $\{0\}\uplus\{Q_t=0\ \text{nilpotents}\}\uplus\{Q_t\neq0\ \text{idempotent multiples}\}$ |
| the cone law | $\tilde Q^{2}=2Q_t\tilde Q$ on $\mathcal N$, and $Q_t=0\iff\tilde Q^{2}=0$ |
| the modulus | $\varrho^{2}e^{2i\theta_{\mathrm{pha}}}=N$ with $\varrho\,e^{i\theta_{\mathrm{pha}}}\in\mathcal H$, $\varrho=\lvert N\rvert^{1/2}$, $e^{i\theta_{\mathrm{pha}}}=\varrho\,e^{i\theta_{\mathrm{pha}}}/\varrho$ |
| the scale and the phase | $\varrho$ the geometric mean of the eigenvalue moduli, $\theta_{\mathrm{pha}}$ the mean of their phases |
| the word | $\tilde Q=\varrho\,e^{i\theta_{\mathrm{pha}}}\tilde U_{\mathrm b}\tilde u_{\mathrm r}$, of real dimensions $1+1+3+3=8$ |
| the two bracketings | $\tilde Q_H=\varrho\tilde U_{\mathrm b}$ and $\tilde Q_0=e^{i\theta_{\mathrm{pha}}}\tilde u_{\mathrm r}$ |
| the amplitude law | $H=\lvert N\rvert\cosh\theta_{\mathrm{rpd}}$, so $H=\lvert N\rvert$ exactly at zero rapidity |
| the cone pair | $\tilde Q_H=ae_t+i\vec W/(2a)$, $\lvert\vec W\rvert=2a^{2}$, $\tilde Q_0=\tilde Q/(2a)$, $H(\tilde Q_0)=\tfrac12$ |
| the rank | $2$ off the cone, $1$ on the nonzero cone, $0$ at the origin |
| the material sector | $N=-p_0^{2}+\lvert\vec p\rvert^{2}$; phase $1$ spacelike, $i$ timelike; scale $m$ timelike |
| the pairings | Cartan total on the units, quaternion needs a real part, Hamilton needs $B^{2}\neq0$ |
| the group | $\mathbb B^{\times}\cong\mathcal H\times SL_2(\mathbb C)$, retracting onto $U(2)$ |
The value of the norm decides everything downstream. It is zero or it is not; if it is not, the element is invertible and the four factors are coordinates on the group; if it is, the element is null and the two-factor pair is the description, with an amplitude of rank one and a scale equal to the square root of half the weight.
What the Reader Should Retain
The three statements to carry away are these. The quaternionic norm is the only one of the four measures that a product multiplies, and that is why the mass, the massless wall and the group of reversible elements are all read from it and from nothing else. The modulus is a square root and therefore a branch, and on the material sector the branch is not a choice: the norm is real, its sign is the timelike–spacelike split, and the phase is $1$ or $i$ accordingly. And the cone is a wall with two strata: the nilpotents, which square to zero, and the imaginary idempotent multiples, which square to a multiple of themselves, the physical null momenta being of the second kind.
Summary of Notation
| symbol | meaning |
|---|---|
| $\tilde Q=Q_te_t+\vec Q$ | the element, with complex coefficients $Q_t,Q_x,Q_y,Q_z$ |
| $N(\tilde Q)=\langle\tilde Q\rangle_{\natural}$ | the quaternionic norm, $Q_t^{2}+Q_x^{2}+Q_y^{2}+Q_z^{2}$ |
| $B^{2}$ | the vector square $Q_x^{2}+Q_y^{2}+Q_z^{2}$, the second invariant |
| $\tilde Q^{\natural}$ | the quaternion conjugate, $Q_t-\vec Q$, used in the norm identity |
| $\tilde Q^{*}$ | the $\star$-conjugate, the complex conjugate of the quaternion conjugate |
| $H(\tilde Q)=\operatorname{Sc}(\tilde Q\tilde Q^{*})$ | the weight of the previous entry, the additive measure |
| $\mathcal N=\{N=0\}$ | the null cone, the zero-divisor cone, the wall |
| $\mathcal N_{\mathrm p}$, $\mathcal N_{\mathrm{np}}$ | its pure stratum of nilpotents and its non-pure stratum |
| $\mathbb B^{\times}=\{N\neq0\}$ | the group of invertibles, the regular case |
| $\mathbb B^{\times}_{1}=\{N=1\}$ | the norm-one group, isomorphic to $SL_2(\mathbb C)$ |
| $\varrho\,e^{i\theta_{\mathrm{pha}}}$ | the modulus, the principal square root of $N$ in $\mathcal H$ |
| $\varrho,\ \theta_{\mathrm{pha}}$ | the scale $\varrho=\lvert N\rvert^{1/2}$ and the phase angle |
| $\theta_{\mathrm{pha}}$ | the phase angle of the modulus, $\theta_{\mathrm{pha}}\in(-\pi/2,\pi/2]$ |
| $\theta_{\mathrm{rpd}}$ | the rapidity, written $\Xi$ in the previous entry |
| $\theta_{\mathrm{axr}}$ | the rotor angle, written $\Theta$ in the previous entry |
| $\vec u_{\mathrm b},\ \vec u_{\mathrm r}$ | the boost axis and the rotor axis, real unit three-vectors |
| $\tilde U_{\mathrm b},\ \tilde u_{\mathrm r}$ | the boost, Hermitian positive of norm one, and the rotor, a unit real quaternion |
| $\lambda_\pm=Q_t\pm iB$ | the two eigenvalues, of product $N$ and of sum $2Q_t$ |
| $\Theta$ | the Hamilton complex angle, $\cos\Theta=Q_t/(\varrho\,e^{i\theta_{\mathrm{pha}}})$ and $\sin\Theta=B/(\varrho\,e^{i\theta_{\mathrm{pha}}})$ |
| $a=\sqrt{H/2}$ | the cone scale, with $2a$ the norm of the rank-one amplitude |
Further Reading
In the mathematics corpus. The ten articles of the source table, and inside the first of them the sections The Quaternionic Norm, which owns the identity and the five equivalent criteria; The $\natural$-regular case: The Group of Invertibles, which owns the connectedness and the homotopy; and The $\natural$-degenerate case: the origin together with the zero divisors, which owns the two strata and the law $\tilde Q^{2}=2Q_t\tilde Q$. The four-factor word, its modulus, its branch and its three pairings are in The Complete Polar Decomposition of an N-Regular Biquaternion, in the sections The Four Factors, The split of the modulus, The Group of Invertibles and The Three Pairings. The word in the logarithmic variable, with the ranges of the three angles, is The Complete Polar Decomposition of an N-Regular Biquaternion. The rank-one amplitude is in The Polar Decomposition of an N-Degenerate Biquaternion. The independence of the two invariants and the four sectors are in Relations between the N Norm and Spectral Regularity and Degeneracy, and the topology of the cone is in The Topology of the Zero-Divisor Cone.
In this category. The Four Forms and the Four Norms supplies the four measures, their Gram matrices and the proof that only the quaternionic one is multiplicative. The Four Products: Composition, Probability, Causality and Gauge supplies the products whose scalar parts the measures are. The Spectrum and the Two Modes supplies the two eigenvalues, the characteristic polynomial and the identity $N=\lambda_+\lambda_-$. The Two Sectors: The Hermitian Split and the Four Bridges supplies the coordinate split on which the material sector and the sign of the norm depend. The Element: the Six Subspaces and the Two Ledgers supplies the six subspaces and the two ledgers. The H Norm: Amplitude and Frame is the previous entry and supplies the Cartan pair that this entry brackets the other way. The next entries are The Partition: Two Invariants and Four Regimes, which reads the four sectors as a partition, and Mass in Motion: The Region of Both Invariants, which reads the boost on the sector where both invariants are nonzero.
The sources of this entry. The Partition: Two Invariants and Four Regimes states the two cases as the reversible and the irreversible. The Group of Invertibles, the Cone and the Orders places the massless content on the cone. The Exponential and Its Three Pairings compares the three bracketings and their domains. The Massless Regime is the family of §The Two Walls, read as a limit of the polar data. The Light Cone as the Biquaternion Zero-Divisor Cone identifies the two cones. Zero Divisors as a Physical Locus in Biquaternionic Form is the earlier reading of the same wall. Norm and Pairing Conventions fixes the letters and the inertias.