Real Polar Decomposition

Introduction

This article is about the polar decomposition of a real number: the statement that every nonzero real number is the product of a positive real scale and a sign, and the description of the sign factor as an element of the two-element group.

The subject is elementary, and its place in this blog is that of the base case. The polar decomposition is developed here for the one-dimensional algebra $\mathbb{R}$, in the companion article Modulus–Unit Polar Decomposition of Complex Numbers for the two-dimensional definite algebra $\mathbb{C}$, and in the companion articles on the polar decompositions of the split-complex and the other algebras of the family. The family decomposes every element of every one of these algebras into the same four slots — a scale, a central phase, a boost and a rotor — and which slots are occupied is a property of the algebra. In $\mathbb{R}$ exactly two slots are occupied, the scale and the compact slot, the latter filled by the discrete sign group, and the algebra is the smallest case of the pattern: no branch choice, no boost, no zero divisors, no boundary, and only one continuous parameter.

The plan is as follows. The modulus and the sign factor are defined, and existence and uniqueness are proved. The sign factor is then described through the exponential, and the exponential is organised by the rule $\nu^2 = -1$, $0$, $+1$ that governs every algebra of the family: the sign of the square of the exponent determines whether the exponential is trigonometric, parabolic or hyperbolic, and it is the same rule in $\mathbb{R}$ and in $\mathbb{C}$. The two factors are then identified with two rows of that rule, the absent slots are accounted for, and the group of invertibles, the matrix picture and worked examples close the article.

The conventions and the symbols are those of Real Algebra and of the corpus (Algebra and Element Conventions); the symbols this article uses are collected in §Summary of Notation.

Every numerical value displayed below is exact.

The Modulus and the Sign Factor

The Two Factors

Definition. Let $q \in \mathbb{R}$, $q \neq 0$. The polar decomposition of $q$ is the writing

$$ q = \varrho\,\sigma, \qquad \varrho \in \mathbb{R}, \quad \varrho > 0, \qquad \sigma \in \mathbb{R}, \quad \sigma^2 = 1, $$

in which $\varrho$ is the modulus of $q$ and $\sigma$ is its sign factor.

The second factor is a unit element, an element of norm one for the norm in view (Norm and Unitary Conventions, §The Unitary Convention); the norm in view is $\langle q\rangle=q^2$, so the unit elements are the two roots of $+1$, the group $\{\pm1\} = O(1)$, and the sign factor is the unit factor of the real column. The phrase names the norm-one elements and not every invertible element: the group of invertibles $\mathbb{R}^\times$ of the algebra is the wider set of nonzero reals, treated in its own section, and the two senses of the word are kept apart throughout.

The definition names the two factors before anything is proved about them, so that the propositions below have definite objects to be about. The modulus carries one real parameter. The sign factor is constrained by the equation $\sigma^2 = 1$, which is the two-point set $\{\pm1\}$, so it carries no continuous parameter. The counts add to the real dimension of the algebra, $1 + 0 = 1$, and this additivity is the pattern the whole family follows.

The Modulus from the Norm

The norm of a real number is

$$ \langle q\rangle = q\,q = q^2. $$

It is a single square, so $\langle q\rangle > 0$ for every $q \neq 0$ and $\langle q\rangle = 0$ only at $q = 0$; the modulus is

$$ \boxed{\;\varrho = \sqrt{\langle q\rangle} = |q|\;}. $$

Because the norm is strictly positive off the origin, the square root is a strictly positive real number with no sign choice and no branch choice. The requirement $\varrho > 0$ is therefore met automatically and the modulus is forced. In the companion articles the corresponding object is a square root of an indefinite real form or of a complex number, and there the sign and the branch both demand attention; in $\mathbb{R}$ neither does.

The modulus is multiplicative, because the norm is: $\varrho(q_1 q_2) = \varrho(q_1)\varrho(q_2)$, and also $\varrho(\lambda q_1) = |\lambda|\,\varrho(q_1)$ for real $\lambda$.

The Sign Factor and the Two-Point Set

The sign factor is

$$ \sigma = \frac{q}{\varrho} = \frac{q}{|q|} = \operatorname{sgn} q . $$

It lies on the two-point set

$$ \{\pm1\} = \{\sigma \in \mathbb{R} : \langle \sigma\rangle = 1\}, $$

which is a group under multiplication, since $\langle \sigma\tau\rangle = \langle \sigma\rangle\langle \tau\rangle = 1$ and $\langle \sigma^{-1}\rangle = \langle \sigma\rangle = 1$ when $\langle \sigma\rangle = 1$, and which is compact, being a finite closed subset of $\mathbb{R}$. It is the orthogonal group of the line, written $O(1)$: the real $1\times1$ matrices of norm one are exactly $[\pm1]$, because a $1\times1$ matrix $[g]$ preserves the form $N$ exactly when $g^2 = 1$. The group $O(1) \cong \mathbb{Z}/2$ has two elements, the identity $+1$ and the reflection $-1$.

The sign factor is the element that acts: multiplication by $-1$ is the reflection of the line through the origin, and multiplication by $+1$ is the identity. In the complex case the corresponding unit factor is the circle $\mathrm{U}(1)$, which is connected; here the group $O(1)$ is discrete, and it is the component group of the group of invertibles.

Existence and Uniqueness

Theorem. Every nonzero real number $q$ has exactly one polar decomposition.

Existence. Put $\varrho = \sqrt{\langle q\rangle}$ and $\sigma = q/\varrho$. Since $\langle q\rangle > 0$, the number $\varrho$ is positive, and $\langle \sigma\rangle = \langle q\rangle/\varrho^2 = 1$ because $\langle \lambda q\rangle = \lambda^2\langle q\rangle$ for real $\lambda$. Hence $q = \varrho\sigma$ with $\varrho > 0$ and $\sigma^2 = 1$.

Uniqueness. Suppose $q = \varrho\sigma = r_1\sigma_1$ with $\varrho, r_1 > 0$ and $\sigma^2 = \sigma_1^2 = 1$. Taking norms gives $\varrho^2 = \langle q\rangle = r_1^2$, so $\varrho = r_1$ because both are positive, and then $\sigma = q/\varrho = \sigma_1$.

The pair of factors is therefore unique, with no sign ambiguity and no branch ambiguity. What is not unique is a coordinate for the sign factor: the sign has no continuous coordinate at all, because the group is finite, and the discrete alternative is the whole of what the exponential can offer. This is the base case of the non-uniqueness of the exponential coordinate that appears from $\mathbb{C}$ onward, where the coordinate of the unit factor is a class modulo $2\pi$.

The Exponential Form

The Word as Exponentials

The word $q = \varrho\,s$ is the product of a scale and a sign, and the scale alone is an exponential: with $\varrho = \lvert q\rvert$ and $s = \operatorname{sgn} q$,

$$ q = \exp(\ln\lvert q\rvert)\cdot s , \qquad s \in \{\pm e_t\} . $$

# factor name real dimension exponential case closed form parameter and range unit axis
1 $\varrho$ scale $1$ $\exp(\ln \varrho)$ hyperbolic $\varrho$ $\ln \varrho \in \mathbb{R}$, $\varrho > 0$ —
2 $s$ sign $0$ — discrete $\pm e_t$ $s = \operatorname{sgn} q$ —

The exponent follows the rule of the series: the character of the exponential is trigonometric, hyperbolic or parabolic according to the square of the root of the exponent, the trigonometric case for a root of $-e_t$, the hyperbolic for a root of $+e_t$, and the parabolic for a nilpotent root, where the series truncates to the unit plus the exponent. The sign is the one factor of the series that no exponential produces, since $\exp(\mathbb{R}) = \mathbb{R}_{>0}$ and $-1$ lies outside it; the sign group $\{\pm e_t\}$ is the discrete boundary of the exponential reading and the whole of the compact slot in $\mathbb{R}$. The two factors commute, so the order is free, and the word is nevertheless not a single exponential, because no element of $\mathbb{R}$ has an exponential equal to a negative number.

The Roots of $\pm1$ in $\mathbb{R}$

equation solutions in $\mathbb{R}$ count row of the rule
$\nu^2 = -e_t$ none $0$ trigonometric: empty
$\nu^2 = 0$ $\nu = 0$ only $1$ parabolic: degenerate, $\exp = e_t$
$\nu^2 = +e_t$ $\nu = \pm 1$ $2$ hyperbolic: the real exponential

The table is the complete classification of the exponentials of $\mathbb{R}$. The two roots of $+1$ are the two points of the sign group, and in $\mathbb{R}$ every root of $+1$ has norm one, because $\langle q\rangle = q^2$; these two points are the whole root set. The split-complex algebra, whose roots of $+1$ number four, is the first in the family where this set is larger than two points. The roots of $-1$ are absent here, and their absence is the statement that $\mathbb{R}$ has no continuous rotation, no continuous phase and no continuous rotor; the compact slot is filled by the discrete sign group alone.

The Sign Factor as an Exponential

The sign factor $\sigma$ has norm one, so its modulus is one. It therefore satisfies $\sigma = \pm1$, and the two cases are reached as follows.

  • The identity component. The positive root $\nu = +1$ has sign $+1$, so the hyperbolic row gives $\exp(\theta) = \cosh\theta + \sinh\theta = \exp(\theta) > 0$ for $\theta \in \mathbb{R}$; this is the positive scale. In particular $\exp(0) = +1 = \sigma$ for the identity sign.
  • The nontrivial component. The negative root $\nu = -1$ gives $\exp(-\theta) = \exp(-\theta) > 0$, again a positive scale and not the element $-1$. The element $\sigma = -1$ is therefore not in the image of the exponential of the algebra: $\exp(\mathbb{R}) = \mathbb{R}_{>0}$ throughout.

The exponential of the real algebra is thus the positive ray alone, and the sign factor is not produced by it. The correct statement of the exponential form is therefore

$$ q = \varrho\,\sigma = \exp(t)\,\sigma, \qquad t = \ln \varrho \in \mathbb{R}, \qquad \sigma \in \{\pm1\} = O(1), $$

in which the continuous part is the exponential and the discrete part is the sign. The reconstruction is exact: for $q = -\tfrac{3}{4}$, $\varrho = \tfrac{3}{4}$, $t = \ln\tfrac{3}{4}$ and $\sigma = -1$, and $\exp(t)\sigma = \tfrac{3}{4}\cdot(-1) = q$. The kernel of the exponential is trivial,

$$ \boxed{\;\exp(t) = 1 \iff t = 0, \qquad \ker\exp = \{0\}\;}, $$

in contrast with the complex case, where the kernel is the discrete lattice $2\pi i\,\mathbb{Z}$. The exponential of the real algebra is injective, and this injectivity is why the sign must be carried separately. The pair $(t, \sigma)$ is unique, because the logarithm is unique on the positive reals and the sign is determined; the exponential therefore supplies a canonical continuous coordinate on the identity component, and the sign is the discrete coordinate of the component.

What the Two Factors Are

Position and Inertia

Two independent dichotomies classify the directions of an algebra of the family. The first is position: with respect to the involution, a direction is fixed if $\operatorname{id}(\nu) = \nu$ and anti-fixed if $\operatorname{id}(\nu) = -\nu$. The second is inertia: the sign of $\nu^2$. The two dichotomies are independent. In $\mathbb{R}$ every element is fixed, the anti-fixed direction being only $0$, and the two factors are the following.

factor range position inertia parameters what it is
$\varrho$ $(0,\infty)$ fixed $\nu^2 = +e_t$ $1$ the scale, central, non-compact
$\sigma$ $\{\pm1\}$ fixed $\nu^2 = +e_t$ $0$ the sign, central, compact and discrete

The verification is immediate in the basis: $1$ is fixed and $1^2 = +e_t$, so the scale direction has position fixed and sign $+1$; the sign direction is the same element $1$ up to the two signs, and it is fixed as well. Both rows of the table are therefore in the fixed subspace, and the anti-fixed subspace contributes nothing. Unlike the complex case, where the second factor is the anti-fixed imaginary direction, here both factors have the same position and the same inertia; they differ in the sign and in whether they carry a continuous parameter.

The Two Slots and the Two Absent Ones

The four slots of the family are the scale, the central phase, the boost and the rotor. In $\mathbb{R}$ the scale is the continuous fixed slot, and the second occupied slot is the compact slot, which in $\mathbb{R}$ is discrete and carries the sign. The remaining two slots are absent, for the following reasons.

No boost, and no continuous rotor or phase. A boost would be a unit factor of hyperbolic sign, $\exp(\nu\phi)$ with $\nu^2 = +e_t$ off the fixed subspace, and a rotor or phase would be a unit factor of trigonometric sign, $\exp(\nu\theta)$ with $\nu^2 = -e_t$. In $\mathbb{R}$ the only elements with $\nu^2 = +e_t$ are $\nu = \pm1$, which are fixed: their exponentials are the positive real numbers $\exp(\pm\phi)$, already counted as the scale, so there is no non-fixed hyperbolic direction and no boost. The trigonometric row is empty outright, because no element of $\mathbb{R}$ squares to $-1$; so there is no trigonometric direction at all, hence no continuous rotor and no continuous central phase, and the compact slot is filled by the discrete sign group alone, the two roots of $+1$ being its elements. The group of invertibles contains exactly one continuous one-parameter subgroup, the scale.

The sign is the whole compact part. In the complex algebra the second factor is the connected circle $\mathrm{U}(1)$, and the group of invertibles is connected. Here the second factor is the disconnected group $O(1) = \{\pm1\}$, and it is the component group of $\mathbb{R}^\times$. The two-dimensional definite case is the first in which the second factor is connected, and the passage from $O(1)$ to $\mathrm{U}(1)$ is the first place the family acquires a nontrivial continuous compact factor.

No Boundary

The polar decomposition of $\mathbb{R}$ has no boundary: it holds on $\mathbb{R}\setminus\{0\}$ and the only excluded element is the origin. The reason is the definiteness of the norm. If $q \neq 0$ then $\langle q\rangle > 0$, so $\varrho > 0$ and $\sigma = q/\varrho$ is defined; there is no element with $\langle q\rangle = 0$ other than zero, and so there is no set on which the modulus vanishes while the element does not. The split-complex algebra is the first in the family where this fails, the first where a nonzero element can have a vanishing modulus, and the first where the polar decomposition must be restricted to a cone complement.

The parametrisation as a homeomorphism. The parametrisation of the group of invertibles by the two factors,

$$ \Phi : \mathbb{R}_{>0} \times \{\pm1\} \longrightarrow \mathbb{R}^\times, \qquad \Phi(\varrho, \sigma) = \varrho\sigma, $$

is a bijective local homeomorphism, and a homeomorphism onto $\mathbb{R}^\times$ because the sign coordinate is discrete and the map is continuous with continuous inverse $(|q|, \operatorname{sgn}q)$; its image is all of $\mathbb{R}^\times$. It fails to cover exactly the complement $\mathbb{R}\setminus\mathbb{R}^\times = \{0\}$, a single point rather than a hypersurface, and the sign group $\{\pm1\}$ is finite, so neither the scale variable nor the sign variable has a boundary. There is no set on which the parametrisation fails, because the norm is definite and the zero-divisor class is empty; the absence of a boundary is the mathematical content of the definiteness of $\mathbb{R}$.

The boundary as ends. The scale coordinate $\varrho = |q|$ runs over $(0,\infty)$, whose two ends are $0$ and $+\infty$; the end at $0$ is the removed origin, and the end at $+\infty$ is one of the two ends of the line. The sign coordinate is finite, so it contributes no end. The only boundary of the polar decomposition is thus the single point at the end $\varrho = 0$, where the norm vanishes.

The Group of Invertibles and the Sign Group

The group of invertibles of $\mathbb{R}$ is $\mathbb{R}^\times = \mathbb{R}\setminus\{0\}$, and it is the direct product of the two factors,

$$ \boxed{\;\mathbb{R}^\times \cong \mathbb{R}_{>0} \times \{\pm1\}, \qquad q \longmapsto \bigl(|q|,\ \operatorname{sgn}q\bigr)\;}, $$

which is the polar decomposition read as a group isomorphism. The first factor is one-dimensional and non-compact, the second is finite and compact, and the isomorphism is the reason the parameter counts of the polar decomposition are one and zero.

The group of invertibles is disconnected, with two components:

$$ \pi_0(\mathbb{R}^\times) \cong \{\pm1\} = O(1), \qquad \mathbb{R}^\times = \mathbb{R}_{>0} \sqcup \mathbb{R}_{<0}. $$

The identity component is the positive ray $\mathbb{R}_{>0} = \exp(\mathbb{R})$, and the nontrivial component is its negative $-\mathbb{R}_{>0}$. The sign group is therefore the group of components, $\{\pm1\} \cong \mathbb{R}^\times/\mathbb{R}_{>0}$, and it is identified with the orthogonal group $O(1)$ of the line. In the complex case the second factor $\mathrm{U}(1)$ is connected, so $\mathbb{C}^\times$ has one component; the passage from two components to one is the first effect of the two-dimensional definite case.

Inversion in the polar coordinates reads $(\varrho\sigma)^{-1} = \varrho^{-1}\sigma$, inverting the scale and fixing the sign; the two components are each stable under inversion, and the fixed points of inversion are $\pm1$.

The Matrix Picture

Multiplication by the Sign as a Reflection

In the basis $\{e_t\}$ a real number is the single coordinate $q$, and multiplication by $q$ is the linear map

$$ M_q = [\,q\,] = q\,I, \qquad I = [\,1\,], $$

the $1\times1$ Cayley matrix. The determinant and the trace are

$$ \det M_q = q, \qquad \operatorname{tr}M_q = q. $$

The sign factor $\sigma = -1$ acts by the matrix $[-1]$, which is the reflection of the line through the origin; the sign factor $\sigma = +1$ acts by the identity $[1]$. The two matrices $\pm I$ are exactly the orthogonal group $O(1)$ of the line, and the norm is their determinant squared: the reflection $[-1]$ has $\det[-1] = -1$ and $\langle -1\rangle = 1$. The group $O(1)$ is the set of $1\times1$ matrices preserving $N$, and it is the sign group; its identity component is $\{+1\} = SO(1)$, the special orthogonal group, and the quotient $O(1)/SO(1) \cong \{\pm1\}$ is the same two-element group.

The Interpretation of Each Factor

factor matrix the determinant of the image isometry type
$\varrho$ $[\varrho]$, $\varrho > 0$ $\varrho$ similarity of ratio $\varrho$
$\sigma = -1$ $[-1]$ $-1$ reflection through the origin
$\sigma = +1$ $[1]$ $1$ identity
$q = \varrho\sigma$ $[q]$ $q$ signed similarity: reflection and scaling

Every nonzero real number is a signed similarity of the line, and the polar decomposition is the splitting of that signed similarity into its scaling and its reflection parts. This is the precise sense in which the real polar decomposition is the one-dimensional case of the matrix polar decomposition, where the positive definite factor is the scalar $rI$ and the orthogonal factor is $\pm I$; in the complex case the orthogonal factor is a rotation.

Worked Examples

A Generic Real Number

Take $q = -\tfrac{3}{4}$. Then $\langle q\rangle = \tfrac{9}{16}$, so

$$ \varrho = \sqrt{\tfrac{9}{16}} = \tfrac{3}{4}, \qquad \sigma = \frac{-3/4}{3/4} = -1, \qquad \varrho\sigma = \tfrac{3}{4}\cdot(-1) = -\tfrac{3}{4} = q . $$

The modulus is positive and the sign is the nontrivial element of the sign group; the reconstruction is exact.

The Degenerate Shapes

$q$ $\langle q\rangle$ $\varrho$ $\sigma$ exponential coordinate
$4$ $16$ $4$ $+1$ $\ln 4$
$-\tfrac{3}{4}$ $\tfrac{9}{16}$ $\tfrac{3}{4}$ $-1$ $\ln\tfrac{3}{4}$
$1$ $1$ $1$ $+1$ $0$
$-1$ $1$ $1$ $-1$ $0$
$0$ $0$ — — —

The first two lines are a positive and a negative number: same modulus shape, opposite signs. The third and fourth lines are the two elements of the sign group itself, of modulus one; they are the fixed points of the polar decomposition, $q = \sigma$ with $\varrho = 1$. The last line is the origin, the unique excluded element. The exponential coordinate of a positive number is its logarithm, and the sign is carried beside it; the coordinate of $-1$ is $0$ as for $+1$, because the sign is not in the image of the exponential.

The Kernel of the Exponential

The exponential of the real algebra is injective,

$$ \exp(t_1) = \exp(t_2) \iff t_1 = t_2, $$

so its kernel is $\{0\}$ and the continuous coordinate of a positive element is unique. The non-uniqueness that the complex and higher cases exhibit is absent here: the only ambiguity is the sign, which is a discrete choice and not a kernel element. In this sense the real algebra is the one member of the family whose continuous exponential coordinate is canonical.

Comparison with the Other Members of the Series

algebra norm modulus unit factor occupied slots group of invertibles
$\mathbb{R}$ $q^2$, definite $\sqrt{N} = \lvert q\rvert$, positive real $\operatorname{sgn}q \in \{\pm1\}$ scale, compact sign (rotor and phase together, discrete) two components, $\mathbb{R}_{>0}\times O(1)$
$\mathbb{C}$ $u^2+v^2$, definite $\sqrt{N}$, positive real $\exp(i\theta)$, $\theta$ mod $2\pi$ scale, circle (rotation and central phase together) connected, $\mathbb{R}_{>0}\times \mathrm{U}(1)$
$\mathbb{C}_{\mathrm{spl}}$ $u^2-v^2$, indefinite $\sqrt{\lvert N\rvert}$, positive real $\exp(\phi j)$ or $j\exp(\phi j)$ scale, hyperbola (rotation and boost together) four components, non-compact

The progression is the progression of the rule. In $\mathbb{R}$ only the hyperbolic row is occupied, and it produces the scale; the trigonometric row is empty and the second factor is the discrete sign group, the compact slot in its discrete form. In $\mathbb{C}$ the trigonometric row is occupied by the unit factor for the first time, and the unit factor is a connected circle. In $\mathbb{C}_{\mathrm{spl}}$ the hyperbolic row is occupied instead, and the group of invertibles is no longer compact. The one-dimensional algebra $\mathbb{R}$ is the case in which none of that can happen, which is why it is treated first.

Summary

Every nonzero real number has exactly one polar decomposition $q = \varrho\sigma$, with modulus $\varrho = \sqrt{\langle q\rangle} = |q| > 0$ and sign $\sigma = q/|q| \in \{\pm1\}$. The modulus is fixed and of sign $+1$ and carries one parameter; the sign is fixed, discrete and carries none; the counts add to the real dimension one. The sign group is $O(1) \cong \mathbb{Z}/2$, the orthogonal group of the line, and its two elements are the two components of $\mathbb{R}^\times = \mathbb{R}_{>0}\times\{\pm1\}$, which is why the real group of invertibles is disconnected while the complex one is connected.

The exponential of the real algebra is governed by the rule $\nu^2 = -1, 0, +1$: the trigonometric row is empty, so there is no continuous rotor and no continuous phase; the parabolic row contains only $\nu = 0$, so it is degenerate; and the hyperbolic row contains $\nu = \pm1$, whose exponentials are the positive ray $\exp(\mathbb{R}) = \mathbb{R}_{>0}$. The nontrivial sign $\sigma = -1$ is therefore not an exponential, and the exponential form of the representation reads $q = \exp(t)\sigma$ with $t = \ln|q|$ and $\sigma \in O(1)$; the kernel of the exponential is trivial. Of the four slots of the family — scale, central phase, boost, rotor — exactly two are occupied, the scale and the compact slot, and the second is discrete. The norm is definite, so there are no zero divisors and the polar decomposition has no boundary: the origin is the only excluded element. In the matrix picture the statement is the splitting of the signed similarity $[q]$ into the scalar $[\varrho]$ and the reflection $[\sigma]$.

Summary of Notation

symbol meaning
$\mathbb{R}$ the real algebra, basis $e_t = 1$, sole involution the identity
$q$ a real number
$\langle q\rangle = q^2$ the norm, positive definite
$\varrho = \sqrt{\langle q\rangle} = \lvert q\rvert$ the modulus, a positive real
$\sigma = \operatorname{sgn}q = q/\lvert q\rvert$ the sign factor, an element of $\{\pm1\}$
$\{\pm1\} = O(1) \cong \mathbb{Z}/2$ the sign group, the orthogonal group of the line
$q = \varrho\sigma$ the polar decomposition
$q = \exp(t)\sigma$, $t = \ln\lvert q\rvert$ the exponential form, with the sign carried separately
$\Phi(\varrho,\sigma) = \varrho\sigma$ the parametrisation of the group of invertibles by scale and sign
$\nu$ the exponent direction, classified by $\nu^2 = -e_t$, $0$ or $+e_t$
$\mathbb{R}^\times = \mathbb{R}_{>0}\sqcup\mathbb{R}_{<0}$ the group of invertibles, two components
$M_q = [q]$ the $1\times1$ Cayley matrix of multiplication by $q$
$\exp(\mathbb{R}) = \mathbb{R}_{>0}$, $\ker\exp = \{0\}$ the image and kernel of the exponential

Further Reading

  • Edmund Landau, Grundlagen der Analysis (Akademische Verlagsgesellschaft, 1930), for the order of $\mathbb{R}$, the absolute value and the sign of an element.
  • Walter Rudin, Principles of Mathematical Analysis, 3rd edition (McGraw-Hill, 1976), for the absolute value as a multiplicative norm and the logarithm.
  • Brian C. Hall, Lie Groups, Lie Algebras, and Representations: An Elementary Introduction, 2nd edition (Springer, 2015), for the exponential map, its image and its kernel in the one-dimensional case.
  • John Stillwell, Naive Lie Theory (Springer, 2008), for the circle and the sign group as the base cases of the compact orthogonal groups.
  • Tristan Needham, Visual Complex Analysis (Oxford University Press, 1997), for the passage from the sign decomposition of the line to the polar decomposition of the plane.
  • Richard S. Pierce, Associative Algebras, Graduate Texts in Mathematics 88 (Springer, 1982), for the determinant, the orthogonal group and the polar decomposition in the one-dimensional matrix algebra.