nLab physical unit


under construction



physics, mathematical physics, philosophy of physics

Surveys, textbooks and lecture notes

theory (physics), model (physics)

experiment, measurement, computable physics



When formulating a theory of physics in terms of mathematics one typically models the range of certain physical quantities by torsors over some group of transformations.

For instance a wavelength would be identified as an element in the positive real numbers, >0\mathbb{R}_{\gt0} , being a torsor over the multiplicative group >0 ×\mathbb{R}_{\gt0}^\times of positive reals.

In order for the “coordination” of the mathematical theory with physical experiment to take place, one needs to choose an identification of this abstract torsor with the (idealized) one that it is supposed to model in nature. Such a choice is equivalent to a choice of unit (in the mathematical sense), hence a choice of element of the torsor. In this context this is then a physical unit.

For instance picking an element in >0\mathbb{R}_{\gt0} and declaring this to be the length of the path travelled by light in a vacuum in 1/299 792 458 second means defining a physical unit of length (in this example: of the meter).

Notice that choice of unit is also called choice of gauge. This is indeed the same “gauge” as in “gauge theory”, as it is how (Weyl 23) introduced the concept of gauge theory: as a theory in which the choice of unit of length may change along paths in space.

Relation to physical constants

Physical units are often called physical constants. But by definition physical units are arbitrary choices made in the desciption of a physical system. Of course once made, one wants to keep these choices constant, such as to be useful.

The actual constants of nature, when they are “dimensionful”, are instead elements of these abstract torsors. For example, the Planck length belongs to the torsor of length, a torsor over >0 ×\mathbb{R}_{\gt0}^\times. We can describe it by its ratio to another element of the same torsor, such as the meter — the Planck length is approximately 1.61610 35m1.616 \cdot 10^{-35} \mathrm{m} — but the number appearing in this statement is completely driven by our arbitrary choice of the meter.

An actual constant of nature corresponds to a specific, non-arbitrary real number only when it is a dimensionless quotient of physical quantities. For instance the fine structure constant is the quotient

αe 2(4πϵ 0)c \alpha \coloneqq \frac{e^2}{ (4 \pi \epsilon_0) \hbar c} \in \mathbb{R}

where ee is the electric charge of the electron, \hbar is Planck's constant, etc. These quantities are such that their dimensionalities cancel out, meaning their quotient is a member not of an abstract torsor but of \mathbb{R} itself, and hence a real number characterizing nature independently of any conventions about how to parameterize it.

For computing such a quotient concretely, one expresses each of the constituent quantities as some real multiple of an appropriate physical unit: e.g. ee as a multiple of the coulomb?, cc as a multiple of the meter per second, etc. If the units are chosen such that their quotient is unity, then the quotient of the numbers is the actual physical constant.

Units of length in Lagrangian field theory

under construction

Let Σ p,1\Sigma \simeq \mathbb{R}^{p,1} be Minkowski spacetime and let EfbΣE \overset{fb}{\to} \Sigma be a fiber bundle thought of as a field bundle. Write {ϕ a}\{\phi^a\} for local coordinates on the typical fiber of this bundle.

The total space of the corresponding jet bundle J Σ (E)jb ΣJ^\infty_\Sigma(E) \overset{jb^\infty}{\to} \Sigma carries an action

sc: ××J Σ (E)J Σ (E) sc \;\colon\; \mathbb{R}^\times \times J^\infty_\Sigma(E) \longrightarrow J^\infty_\Sigma(E)

of the multiplicative group of units ×\mathbb{R}^\times of the real numbers, given on the induced jet coordinates by

x μ rx μ ϕ a ϕ a ϕ ,μ 1,μ k a r kϕ ,μ 1μ k a. \begin{aligned} x^\mu & \mapsto r x^\mu \\ \phi^a & \mapsto \phi^a \\ \phi^a_{,\mu_1, \cdots \mu_k} & \mapsto r^{-k} \phi^a_{,\mu_1 \cdots \mu_k} \end{aligned} \,.

Let then

L (): nΩ p+1(J Σ (E)) \mathbf{L}_{(-)} \;\colon\; \mathbb{R}^n \longrightarrow \mathbf{\Omega}^{p+1}(J^\infty_\Sigma(E))

be a smoothly nn-parameterized collection of Lagrangian densities, equipped with an R ×R^\times-action

scp: ×× n n scp \;\colon\; \mathbb{R}^\times \times \mathbb{R}^n \longrightarrow \mathbb{R}^n

on n\mathbb{R}^n.

Observe that the Euler-Lagrange equations induced by a Lagrangian density L\mathbf{L} equal those induced by the rescaled Lagrangian rLr \mathbf{L}, and that the presymplectic current Ω BFV\Omega_{BFV} induced by L\mathbf{L} scales linearly with rr itself. Upon quantization, this rescaling of Ω BFV\Omega_{BFV} may be absorbed in Planck's constant. In conclusion, as long as Lagrangian densities scale homogeneously the rescaled Lagrangian induces the same physics.

Hence we require that the combined scaling action of ×\mathbb{R}^\times on J Σ (E)J^\infty_\Sigma(E) via scsc and on the parameters in n\mathbb{R}^n via scpscp is homogeneous on L\mathbf{L} in that there exists dimdim \in \mathbb{Z} such that for every r ×r \in \mathbb{R}^\times we have

sc r *L (scp())=r dimL (). sc_r^\ast \mathbf{L}_{( scp(-))} = r^{dim} \mathbf{L}_{(-)} \,.

Then a parameter a: na \colon \mathbb{R}^n \to \mathbb{R} such that there exists ww \in \mathbb{Z} with

scp r *a=r wa scp_r^\ast a = r^{w} a

is said to have dimension [length] w[length]^{w}.

For example the Lagrangian density for the free scalar field

L (): 1Ω p+1(J Σ (Σ×) \mathbf{L}_{(-)} \;\colon\; \mathbb{R}^1 \longrightarrow \mathbf{\Omega}^{p+1}(J^\infty_\Sigma(\Sigma \times \mathbb{R})

given by

L m12(η μνϕ ,μϕ ,νm 2ϕ 2)dvol \mathbf{L}_{m} \;\coloneqq\; \tfrac{1}{2} \left( \eta^{\mu \nu}\phi_{,\mu} \phi_{,\nu} - m^2 \phi^2 \right) dvol

is parameterized by the mass mm. For the Lagrangian to scale homogenously with r p1r^{p-1} the mass parameter has to have dimension [length] 1[length]^{-1}. To indicate this action one writes the mass in the combination mc/m c / \hbar, called the inverse Compton wavelength, so that the homogenously scaling collection of Lagrangians is

L mc/12(η μνϕ ,μϕ ,ν(mc)ϕ 2)dvol \mathbf{L}_{m c / \hbar} \;\coloneqq\; \tfrac{1}{2} \left( \eta^{\mu \nu}\phi_{,\mu} \phi_{,\nu} - \left( \tfrac{m c}{\hbar} \right) \phi^2 \right) dvol



fundamental scales (fundamental/natural physical units)


For a mathematical description of physical units and the associated “physical dimensions”, including a discussion on how densities can be used to define real and complex powers of physical quantities, see

Additional references:

Discussion in the context of philosophy of science includes

Last revised on June 16, 2023 at 06:13:55. See the history of this page for a list of all contributions to it.