Multipole expansion of the electrostatic potential

Intuition Beginner

Look at an electric charge from far away and the details of how the charge is shaped become invisible. A pair of equal positive charges close together looks, from a distance, like a single charge twice as big. A pair of equal-and-opposite charges close together looks, from a distance, like nothing at all — until you look more carefully, and then a weak field appears that falls off faster than a point charge's field. The whole programme of the multipole expansion is to make this systematic: any localised collection of charges, viewed from far enough away, looks like a tower of simpler patterns stacked on top of one another, with each successive pattern weaker than the last.

The first pattern in the tower is the monopole — the total charge $Q$ of the system. Its field falls off as $1/r^2$ and its potential as $1/r$, exactly like a point charge sitting at the centre. The second is the dipole — what is left if the total charge is zero but the positive and negative charges are slightly displaced from each other. Its potential falls off as $1/r^2$ and its field as $1/r^3$. The third is the quadrupole, the next pattern that remains if both the total charge and the dipole moment are zero, and it falls off faster still. Each successive level adds detail that becomes visible only as you move closer to the source.

The pay-off is that any complicated charge distribution can be summarised by a short list of numbers — its total charge $Q$, its dipole moment $p$, its quadrupole moment $Q_{ij}$, and so on. Once you know these few numbers, you know the long-distance behaviour of the field without having to track every individual charge. A water molecule has zero net charge, but its bent shape gives it a dipole moment of about $6.2\times 10^{-30}$ coulomb-metres, and this single number explains why water dissolves salt, why ice floats, and why water has such a high boiling point.

The same idea reaches across physics. The Earth has a magnetic dipole moment of about $8.0\times 10^{22}$ ampere-square-metres; that is why a compass needle aligns north. Atomic nuclei have small quadrupole moments measurable in spectroscopy; the fact that the deuteron has a positive quadrupole moment was historical evidence that the nuclear force is not purely central. Even the cosmic microwave background — the leftover radiation from the early universe — is described by its multipole moments on the sky, with the dipole encoding our motion through space and the higher moments encoding fluctuations from the time of recombination.

Visual Beginner

The picture below shows the first three multipoles of a charge distribution placed near the origin. The monopole is a single positive charge whose field lines spread isotropically outward. The dipole is a pair of equal-and-opposite charges; its field lines emerge from the positive charge and curve back into the negative charge, producing the characteristic figure-eight pattern. The quadrupole is two dipoles oriented head-to-head with opposite signs — equivalently, four charges of alternating sign at the corners of a square — and its field is even more localised, falling off faster as you move away.

The crucial visual rule: every step up the tower trades total charge for spatial structure. The monopole has the most charge and the least structure; the higher multipoles have no net charge but more elaborate spatial patterns, and they fall off faster with distance. From very far away you see only the monopole; closer in, the dipole takes over; closer still, the quadrupole and higher terms become important. The multipole expansion is, in essence, a far-to-near zoom lens for the field of a localised source.

Worked example Beginner

A water molecule sits on a tabletop at the origin. Its hydrogen-oxygen-hydrogen geometry gives it a permanent electric dipole moment of magnitude $p=6.2\times 10^{-30}\mathrm{C}\cdot \mathrm{m}$, pointing from the oxygen toward the midpoint of the two hydrogens. Estimate the strength of the electric field this single water molecule produces at a distance of $r=1\mathrm{nm}$ along the axis of its dipole.

Step 1. The molecule has zero net charge, so the monopole contribution is zero. The leading non-zero contribution to the potential at distance $r$ is the dipole contribution, $V=p/(4\pi {ϵ}_0{r}^2)$, valid along the dipole axis.

Step 2. The field along the dipole axis falls off as $1/r^3$. Its magnitude is $E=2p/(4\pi {ϵ}_0{r}^3)=(2\times 6.2\times 10^{-30})/(4\pi \times 8.85\times 10^{-12}\times (10^{-9}{)}^3)$.

Step 3. Computing the denominator: $4\pi \times 8.85\times 10^{-12}=1.11\times 10^{-10}$, times $10^{-27}$ for $r^3$, gives $1.11\times 10^{-37}$. The numerator is $1.24\times 10^{-29}$. So $E\approx 1.1\times 10^{8}V/m$.

Step 4. Compare with the field that a single full electronic charge would produce at the same distance: $E_e=e/(4\pi {ϵ}_0{r}^2)=1.6\times 10^{-19}/(4\pi \times 8.85\times 10^{-12}\times 10^{-18})\approx 1.4\times 10^{9}V/m$. The water-molecule dipole produces a field about ten times smaller at $1\mathrm{nm}$, but the field is still enormous on a laboratory scale.

What this tells us: even though the water molecule is electrically neutral, its dipole moment alone produces extremely strong fields at molecular distances. Those fields are exactly what drive hydrogen bonding between neighbouring water molecules and explain water's remarkable properties as a solvent.

Check your understanding Beginner

Formal definition Intermediate+

Let $\rho ({\mathbf{r}}^{\prime })$ be a static charge distribution supported in a bounded region $V\subset {\mathbb{R}}^3$, and let $\mathbf{r}$ be a point outside $V$ at distance $r$ from the origin. The electrostatic potential at $\mathbf{r}$ is

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}{\int }_V\frac{\rho ({\mathbf{r}}^{\prime })}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}{d}^3{r}^{\prime }.$$

The multipole expansion is the Taylor expansion of the integrand in powers of $r^{\prime }/r$, valid for $r>{\max }_{{\mathbf{r}}^{\prime }\in V}|{\mathbf{r}}^{\prime }|$.

Legendre form. Let $\gamma$ be the angle between $\mathbf{r}$ and ${\mathbf{r}}^{\prime }$. The generating-function identity

$$\frac{1}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}=\frac{1}{r}\sum _{\ell =0}^{\infty }{\left(\frac{r^{\prime }}{r}\right)}^{\ell }{P}_{\ell }(\cos \gamma )$$

(absolutely convergent for $r^{\prime }<r$) gives the Legendre expansion

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\sum _{\ell =0}^{\infty }\frac{1}{r^{\ell +1}}{\int }_V{r}^{\prime \ell }{P}_{\ell }(\cos \gamma )\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime }.$$

Cartesian form. Taylor expanding $1/|\mathbf{r}-{\mathbf{r}}^{\prime }|$ in ${\mathbf{r}}^{\prime }$ around ${\mathbf{r}}^{\prime }=0$ gives, to third order,

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\left[\frac{Q}{r}+\frac{\mathbf{p}\cdot \hat{\mathbf{r}}}{r^2}+\frac{1}{2r^3}\sum _{i,j}{Q}_{ij}{\hat{r}}_i{\hat{r}}_j+\mathcal{O}(r^{-4})\right],$$

where $\hat{\mathbf{r}}=\mathbf{r}/r$ and the Cartesian multipole moments are

$$Q={\int }_V\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime },\mathbf{p}={\int }_V{\mathbf{r}}^{\prime }\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime },$$ $$Q_{ij}={\int }_V\left(3r_i^{\prime }{r}_j^{\prime }-{\delta }_{ij}{r}^{\prime 2}\right)\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime }.$$

The traceless symmetric tensor $Q_{ij}$ has five independent components (six after symmetry minus one trace), matching the five spherical-harmonic components $Y_2^m$ for $m\in \{-2,-1,0,1,2\}$.

Spherical form. In spherical coordinates $(r^{\prime },{\theta }^{\prime },{\phi }^{\prime })$, the addition theorem for spherical harmonics gives

$$P_{\ell }(\cos \gamma )=\frac{4\pi }{2\ell +1}\sum _{m=-\ell }^{\ell }{Y}_{\ell }^m(\theta ,\phi )Y_{\ell }^{m\ast }({\theta }^{\prime },{\phi }^{\prime }).$$

Substituting into the Legendre form yields the spherical multipole expansion

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\sum _{\ell =0}^{\infty }\sum _{m=-\ell }^{\ell }\frac{4\pi }{2\ell +1}\frac{q_{\ell ,m}{Y}_{\ell }^m(\theta ,\phi )}{r^{\ell +1}},$$

with spherical multipole moments

$$q_{\ell ,m}={\int }_V{r}^{\prime \ell }{Y}_{\ell }^{m\ast }({\theta }^{\prime },{\phi }^{\prime })\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime }.$$

Orthonormality of the $Y_{\ell }^m$ on $S^2$ makes this decomposition unique: each $q_{\ell ,m}$ is recovered from $\varphi$ by an angular projection at fixed $r$.

Energy of multipoles in an external field. For a localised distribution $\rho$ in an external potential ${\varphi }_{\text{ext}}(\mathbf{r})$ slowly varying across the support of $\rho$,

$$U={\int }_V\rho ({\mathbf{r}}^{\prime }){\varphi }_{\text{ext}}({\mathbf{r}}^{\prime })d^3{r}^{\prime }=Q{\varphi }_{\text{ext}}(0)-\mathbf{p}\cdot {\mathbf{E}}_{\text{ext}}(0)-\frac{1}{6}\sum _{i,j}{Q}_{ij}({\partial }_i{E}_{\text{ext},j})(0)+\cdots ,$$

after expanding ${\varphi }_{\text{ext}}$ to third order around the origin and using ${\mathbf{E}}_{\text{ext}}=-\nabla {\varphi }_{\text{ext}}$ together with $\nabla \cdot {\mathbf{E}}_{\text{ext}}=0$ for an external field with no sources inside $V$. The torque on the dipole is $\mathbf{\tau }=\mathbf{p}\times {\mathbf{E}}_{\text{ext}}$.

Counterexamples to common slips Intermediate+

• The dipole moment depends on the origin if the total charge is non-zero. Shifting the origin by $\mathbf{a}$ changes $\mathbf{p}$ to $\mathbf{p}-Q\mathbf{a}$. Only when $Q=0$ is $\mathbf{p}$ origin-independent. Similarly, $Q_{ij}$ is origin-independent only if both $Q=0$ and $\mathbf{p}=0$.
• The quadrupole tensor $Q_{ij}$ is traceless by construction. The trace of the integrand is $3r^{\prime 2}-3r^{\prime 2}=0$, so ${\sum }_i{Q}_{ii}=0$ identically. A common slip is to write the moment as $\int r_i^{\prime }{r}_j^{\prime }\rho d^3{r}^{\prime }$ (without the trace subtraction), which is symmetric but not traceless and contains a redundant scalar piece duplicating the monopole information.
• The Cartesian expansion is not the spherical expansion in different coordinates. The two are related but the Cartesian $Q_{ij}$ has different normalisation conventions across textbooks. Jackson 1999 uses $Q_{ij}=\int (3r_i^{\prime }{r}_j^{\prime }-{\delta }_{ij}{r}^{\prime 2})\rho$; Landau-Lifshitz uses $D_{ij}=\int (3r_i^{\prime }{r}_j^{\prime }-{\delta }_{ij}{r}^{\prime 2})\rho$. The spherical moments $q_{\ell ,m}$ are unambiguously defined, which is one reason Jackson moves to the spherical form for systematic work.

Key theorem with proof Intermediate+

Theorem (multipole expansion of the Newton kernel). Let $\rho$ be a charge distribution supported in a ball $B_R=\{{\mathbf{r}}^{\prime }:|{\mathbf{r}}^{\prime }|\le R\}$, and let $\mathbf{r}$ satisfy $r=|\mathbf{r}|>R$. Then the electrostatic potential admits the absolutely convergent expansion

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\sum _{\ell =0}^{\infty }\sum _{m=-\ell }^{\ell }\frac{4\pi }{2\ell +1}\frac{q_{\ell ,m}{Y}_{\ell }^m(\theta ,\phi )}{r^{\ell +1}},q_{\ell ,m}={\int }_{B_R}{r}^{\prime \ell }{Y}_{\ell }^{m\ast }({\theta }^{\prime },{\phi }^{\prime })\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime }.$$

Proof. Start from the Coulomb form

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}{\int }_{B_R}\frac{\rho ({\mathbf{r}}^{\prime })}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}{d}^3{r}^{\prime }.$$

Place the origin at the centre of $B_R$ and let $\gamma$ be the angle between $\mathbf{r}$ and ${\mathbf{r}}^{\prime }$. The law of cosines gives $|\mathbf{r}-{\mathbf{r}}^{\prime }{|}^2=r^2+r^{\prime 2}-2r{r}^{\prime }\cos \gamma$, so

$$\frac{1}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}=\frac{1}{r}{\left(1-2(r^{\prime }/r)\cos \gamma +(r^{\prime }/r{)}^2\right)}^{-1/2}.$$

The generating function for the Legendre polynomials states that

$$(1-2tx+t^2{)}^{-1/2}=\sum _{\ell =0}^{\infty }{t}^{\ell }{P}_{\ell }(x),|t|<1,|x|\le 1,$$

with absolute convergence for $|t|<1$ uniform on compact subsets in $x\in [-1,1]$. Applying this with $t=r^{\prime }/r$ (so $0\le t<1$ since $r^{\prime }\le R<r$) and $x=\cos \gamma$ yields

$$\frac{1}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}=\frac{1}{r}\sum _{\ell =0}^{\infty }{\left(\frac{r^{\prime }}{r}\right)}^{\ell }{P}_{\ell }(\cos \gamma ).$$

Substituting into the Coulomb integral and exchanging sum and integral (justified by the dominated convergence theorem and the uniform bound $|P_{\ell }(\cos \gamma )|\le 1$):

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\sum _{\ell =0}^{\infty }\frac{1}{r^{\ell +1}}{\int }_{B_R}{r}^{\prime \ell }{P}_{\ell }(\cos \gamma )\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime }.$$

This is the Legendre form. To convert to the spherical-harmonic form, apply the addition theorem (proved as a separate result in unit 12.05.02):

$$P_{\ell }(\cos \gamma )=\frac{4\pi }{2\ell +1}\sum _{m=-\ell }^{\ell }{Y}_{\ell }^m(\theta ,\phi )Y_{\ell }^{m\ast }({\theta }^{\prime },{\phi }^{\prime }),$$

where $(\theta ,\phi )$ are the angles of $\mathbf{r}$ and $({\theta }^{\prime },{\phi }^{\prime })$ those of ${\mathbf{r}}^{\prime }$. Substituting:

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\sum _{\ell =0}^{\infty }\sum _{m=-\ell }^{\ell }\frac{4\pi }{2\ell +1}\frac{Y_{\ell }^m(\theta ,\phi )}{r^{\ell +1}}{\int }_{B_R}{r}^{\prime \ell }{Y}_{\ell }^{m\ast }({\theta }^{\prime },{\phi }^{\prime })\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime }.$$

The remaining integral is $q_{\ell ,m}$ as defined in the statement. Convergence of the resulting double sum follows from the convergence of the Legendre series together with the uniform bound $|Y_{\ell }^m|\le \sqrt{(2\ell +1)/(4\pi )}$ on $S^2$. $■$

Bridge. The expansion builds toward 10.07.02 pending multipole radiation, where the static spherical multipole moments $q_{\ell ,m}$ are replaced by their time-Fourier transforms $q_{\ell ,m}(\omega )$ and the static far-field factor $1/r^{\ell +1}$ is replaced by an outgoing-wave Hankel function $h_{\ell }^{(1)}(kr)$, whose large-$r$ asymptotics produce the radiation fields. The foundational reason that the spherical form (rather than the Cartesian form) is the canonical organiser is that the $Y_{\ell }^m$ furnish the irreducible representation of weight $\ell$ for the rotation group $SO(3)$, and this is exactly the structure that appears again in 12.05.02 in the quantum-mechanical decomposition of angular-momentum eigenstates. Putting these together, the multipole expansion identifies the long-range behaviour of any localised source with a tower of irreducible $SO(3)$-tensors, and the bridge is between the analytic Taylor expansion of the Newton kernel and the representation-theoretic decomposition of $L^2(S^2)$ into spherical-harmonic subspaces. The same algebraic structure recurs in gravitational potentials (Legendre's original 1785 setting), in nuclear and atomic spectroscopy, and in cosmological perturbation theory on the CMB sky.

Exercises Intermediate+

Advanced results Master

Theorem 1 (Multipole expansion in spherical harmonics, Jackson §4.1). Let $\rho$ be a charge distribution supported in $B_R=\{|{\mathbf{r}}^{\prime }|\le R\}$. For $r>R$,

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\sum _{\ell =0}^{\infty }\sum _{m=-\ell }^{\ell }\frac{4\pi }{2\ell +1}\frac{q_{\ell ,m}}{r^{\ell +1}}{Y}_{\ell }^m(\theta ,\phi ),$$

with $q_{\ell ,m}=\int r^{\prime \ell }{Y}_{\ell }^{m\ast }({\theta }^{\prime },{\phi }^{\prime })\rho ({\mathbf{r}}^{\prime })d^3{r}^{\prime }$. The series converges absolutely and uniformly on $\{r\ge R+ϵ\}$ for any $ϵ>0$. (Proof: Key Theorem above.) The five real components of the symmetric traceless Cartesian $Q_{ij}$ are linear combinations of the five spherical components $q_{2,m}$, $m=-2,\dots ,2$; the explicit dictionary is given in Jackson 1999 equations 4.5-4.6.

Theorem 2 (Origin dependence of multipole moments). The monopole $Q$ is origin-independent. The dipole $\mathbf{p}$ is origin-independent if and only if $Q=0$. The quadrupole $Q_{ij}$ is origin-independent if and only if $Q=0$ and $\mathbf{p}=0$. More generally, the lowest non-vanishing $2^{\ell }$-pole moment is origin-independent; all higher moments depend on the choice of origin once a lower-order moment is non-zero. (Proof: Exercise 8 above; for the general statement, see Jackson 1999 §4.1.)

Theorem 3 (Irreducibility under $SO(3)$). The $(2\ell +1)$-dimensional space spanned by the spherical multipole moments $q_{\ell ,m}$ for fixed $\ell$ carries an irreducible unitary representation of $SO(3)$ of weight $\ell$. The Cartesian quadrupole tensor $Q_{ij}$ decomposes under $SO(3)$ as ${\mathrm{Sym}}^2{\mathbb{R}}^3=(\text{trace})\oplus (\text{traceless symmetric})=1\oplus 5$, with the trace piece $1$ identically zero by construction and the traceless piece $5$ matching $\ell =2$. The dictionary $\ell =0\leftrightarrow$ scalar, $\ell =1\leftrightarrow$ vector, $\ell =2\leftrightarrow$ traceless symmetric 2-tensor extends to all $\ell$: the rank-$\ell$ traceless symmetric tensor on ${\mathbb{R}}^3$ has exactly $2\ell +1$ independent components, matching the dimension of the spherical-harmonic subspace. This identification is the representation-theoretic content of the multipole expansion and is the structural fact that organises radiative selection rules, nuclear spin-multipole moments, and the angular-momentum content of any localised electromagnetic source.

Theorem 4 (Power radiated by an oscillating dipole, Larmor 1897). A point dipole oscillating sinusoidally as $\mathbf{p}(t)=p_0\hat{\mathbf{z}}\cos (\omega t)$ radiates time-averaged power

$$P=\frac{{\mu }_0{\omega }^4{p}_0^2}{12\pi c}=\frac{{\omega }^4{p}_0^2}{12\pi {ϵ}_0{c}^3}.$$

The radiation is peaked in the equatorial plane (zero along the dipole axis), with angular distribution ${\sin }^2\theta$. (See 10.07.02 pending for the radiative-multipole derivation; for the original derivation see Larmor 1897 Phil. Mag. (5) 44, 503.) The ${\omega }^4$ scaling — fourth-power scaling with frequency — is the same scaling that makes the sky blue: Rayleigh scattering off air molecules is dipole radiation driven by the incident sunlight, and the blue end of the spectrum (higher $\omega$) is scattered $\sim ({\omega }_{\text{blue}}/{\omega }_{\text{red}}{)}^4\approx 10$ times more strongly than the red end. The next multipole — electric quadrupole or magnetic dipole — radiates with power $\propto {\omega }^6$, suppressed by $(\omega R/c{)}^2$ relative to electric dipole, where $R$ is the source size; for atomic transitions $\omega R/c\sim \alpha \approx 1/137$, hence the $\sim 10^{-4}$ suppression of quadrupole / magnetic-dipole transitions relative to electric-dipole transitions and the resulting "forbidden" character of $\ell$-changes greater than $\pm 1$.

Theorem 5 (Magnetic multipole expansion). The magnetic field of a localised current distribution $\mathbf{J}$ admits a multipole expansion parallel to the electric case. The leading term is the magnetic dipole moment $\mathbf{m}=\frac{1}{2}\int {\mathbf{r}}^{\prime }\times \mathbf{J}({\mathbf{r}}^{\prime })d^3{r}^{\prime }$, and the associated vector potential is $\mathbf{A}(\mathbf{r})=({\mu }_0/4\pi )\mathbf{m}\times \hat{\mathbf{r}}/r^2$. The structural parallel to the electric case is exact: free currents source the H-field as free charges source D, magnetic dipoles appear as $\mathbf{m}$ as electric dipoles appear as $\mathbf{p}$, and the multipole expansion organises both into towers of irreducible $SO(3)$ tensors. There is no magnetic monopole — the magnetic series starts at the dipole level, while the electric series starts at the monopole — but the higher multipoles ($\ell \ge 1$) have full parallel structure. The Earth's magnetic field is, to leading approximation, a pure dipole with $|\mathbf{m}|=8.0\times 10^{22}\mathrm{A}\cdot {\mathrm{m}}^2$ (Bocchieri-Loinger 1957 Nuovo Cimento 5 give the classical-dynamics framework); deviations encode the higher multipoles measured by satellite magnetometers.

Theorem 6 (Nuclear electromagnetic multipoles and angular-momentum selection). A nucleus in a state of total angular momentum $J$ has non-vanishing electric multipole moments $Q_{\ell }$ only for $\ell$ even with $0\le \ell \le 2J$, and non-vanishing magnetic multipole moments ${\mu }_{\ell }$ only for $\ell$ odd with $1\le \ell \le 2J$. (Wigner-Eckart theorem; Blatt-Weisskopf 1952 Theoretical Nuclear Physics §III.) The parity restriction reflects the behaviour of the electromagnetic operators under spatial inversion: ${\hat{r}}_i$ is parity-odd, $Y_{\ell }^m$ is parity $(-1{)}^{\ell }$, currents are parity-even. The angular-momentum cap $\ell \le 2J$ comes from the requirement that $J+J\to \ell$ be a legal angular-momentum coupling in the matrix element $⟨J|{\hat{O}}_{\ell }|J⟩$. The deuteron has $J=1$, hence allowed electric multipoles at $\ell =0$ (monopole = total charge) and $\ell =2$ (quadrupole); its measured quadrupole moment $Q=+0.286\mathrm{e}\cdot \mathrm{f}{\mathrm{m}}^2$ was the historical proof that the nuclear force contains a tensor (non-central) component (Rabi 1939; Schwinger-Rarita 1941).

Theorem 7 (Gravitational multipoles and orbit precession). The Newtonian gravitational potential outside a rotating planet of radius $R$ admits the expansion

$$\Phi (r,\theta )=-\frac{GM}{r}\left[1-\sum _{\ell =2}^{\infty }{\left(\frac{R}{r}\right)}^{\ell }{J}_{\ell }{P}_{\ell }(\cos \theta )\right],$$

with the dominant non-spherical coefficient $J_2\approx 1.083\times 10^{-3}$ for Earth, arising from the rotational equatorial bulge of the geoid. Higher $J_{\ell }$ for Earth: $J_3\approx -2.5\times 10^{-6}$ (pear-shaped distortion), $J_4\approx -1.6\times 10^{-6}$. The $J_2$ term causes the line of nodes of a satellite orbit to precess at a rate proportional to $J_2\cos i/(R^2/r^2)$, where $i$ is the inclination of the orbit relative to the equatorial plane; this precession is exploited to design sun-synchronous orbits in which the orbital plane rotates at exactly $360^{\circ }/\text{year}$, keeping the satellite over each ground point at the same local solar time. The same multipole expansion (originated by Legendre 1785 and Laplace 1782 for planetary potentials) is the historical origin of the Legendre polynomials.

Theorem 8 (Cosmological multipoles and the CMB). The cosmic microwave background temperature anisotropy on the sky is expanded as

$$\delta T(\theta ,\phi )=\sum _{\ell =0}^{\infty }\sum _{m=-\ell }^{\ell }{a}_{\ell ,m}{Y}_{\ell }^m(\theta ,\phi ),$$

with the rotationally invariant power spectrum $C_{\ell }=⟨|a_{\ell ,m}{|}^2⟩$ encoding the statistical properties of the primordial fluctuations. The monopole ($\ell =0$) is the mean CMB temperature $T_0=2.725\mathrm{K}$ (Mather-Smoot 1992 Astrophys. J. Lett. 397, COBE FIRAS measurement). The dipole ($\ell =1$) is the kinematic signature of the Solar System's peculiar velocity ${\mathbf{v}}_{\odot }$ relative to the CMB rest frame, $\delta T_1/T_0={\mathbf{v}}_{\odot }\cdot \hat{\mathbf{n}}/c$ with $|{\mathbf{v}}_{\odot }|\approx 368km/s$. The higher multipoles ($\ell \ge 2$) carry cosmological information: the quadrupole through the first acoustic peak ($\ell \approx 200$) encodes the geometry of the universe (spatial flatness from peak position), the baryon-to-photon ratio (from peak-height ratios), and the dark-matter density. WMAP (2003) and Planck (2013-2018) measured the spectrum to $\ell \sim 2500$, fixing the standard cosmological model to percent-level precision. A persistent observational anomaly is the slight suppression of the $\ell =2$ quadrupole below the standard-model prediction — interesting but compatible with cosmic variance.

Theorem 9 (Berry-phase polarization in crystals, King-Smith-Vanderbilt 1993). In a periodic crystal, the macroscopic polarization is

$$\mathbf{P}=\frac{ie}{(2\pi {)}^3}\sum _n{\int }_{\text{BZ}}⟨u_{n\mathbf{k}}|{\nabla }_{\mathbf{k}}|u_{n\mathbf{k}}⟩d^{3}k(\mathrm{mod}e\mathbf{R}/V_{\text{cell}}),$$

where $u_{n\mathbf{k}}$ are the periodic parts of the Bloch wavefunctions, summed over occupied bands and integrated over the Brillouin zone, defined modulo a polarization quantum $e\mathbf{R}/V_{\text{cell}}$ with $\mathbf{R}$ any lattice vector. (King-Smith-Vanderbilt 1993 Phys. Rev. B 47.) The classical multipole-style definition $\mathbf{P}=\int \rho \mathbf{r}{d}^{3}r$ is ill-posed for an infinite periodic crystal because $\mathbf{r}$ is unbounded and the integral diverges; the Berry-phase formula replaces it with a manifestly periodic expression. Differences in polarization across phase transitions (the change $\Delta \mathbf{P}$ from one ferroelectric state to another) are well-defined modulo the polarization quantum, and modern density-functional implementations compute them as line integrals of the Berry connection in $\mathbf{k}$-space. This is the connection to topological insulators (Kane-Mele 2005 Phys. Rev. Lett. 95; Fu-Kane-Mele 2007 Phys. Rev. B 76): the integrated Berry curvature is a ${\mathbb{Z}}_2$ topological invariant when time-reversal symmetry is present, and the multipole moments of insulating crystals become topologically protected observables.

Theorem 10 (Vector spherical harmonics and radiative multipoles, Hansen 1935). The electromagnetic field of a time-varying source $\rho (\mathbf{r},t),\mathbf{J}(\mathbf{r},t)$ in a homogeneous region admits the multipole expansion in vector spherical harmonics:

$$\mathbf{E}=\sum _{\ell ,m}[a_E(\ell ,m){\mathbf{N}}_{\ell ,m}^{(1)}(kr)+a_M(\ell ,m){\mathbf{M}}_{\ell ,m}^{(1)}(kr)],$$

where ${\mathbf{M}}_{\ell ,m}$ and ${\mathbf{N}}_{\ell ,m}$ are the magnetic and electric vector spherical harmonics (Hansen 1935 Phys. Rev. 47), the coefficients $a_E$ and $a_M$ encode the electric and magnetic multipole moments of the source, and the radial Hankel functions $h_{\ell }^{(1)}(kr)$ provide outgoing-wave behaviour. The static expansion of this unit is recovered in the $k\to 0$ limit; the time-varying generalisation introduces the additional structure of magnetic-multipole radiation, which has no static counterpart because Maxwell's static equations admit no magnetic-monopole sources. Gravitational-wave radiation follows a parallel structure with weight $\ell =2$ (quadrupole) as the lowest non-vanishing radiating multipole, both monopole and dipole gravitational radiation being forbidden by mass and momentum conservation respectively; LIGO detects the $\ell =2$ quadrupole signal of binary-merger inspirals, and the $\ell =3,4$ contributions appear in higher-mode analyses of asymmetric mergers.

Synthesis. The multipole expansion is the foundational reason that any localised source admits a far-field description through a short list of numbers: total charge, dipole vector, quadrupole tensor, and so on, with each level falling off one power of $r$ faster than the last. The central insight is that the spherical-harmonic decomposition is not just a convenient choice of basis but the irreducible representation-theoretic decomposition of the rotation group acting on $L^2(S^2)$, and this is exactly the structure that organises angular-momentum eigenstates in quantum mechanics, radiative selection rules in atomic spectroscopy, and the multipole content of nuclear and cosmological observables. Putting these together with the Larmor radiation formula (Theorem 4), the bridge is between static multipole moments and the strength of the corresponding radiating modes: an oscillating $\ell$-pole moment radiates power $\propto {\omega }^{2\ell +2}$, with the dominant contribution coming from the lowest allowed $\ell$ for any given transition.

The pattern recurs across every domain that involves a localised source observed from afar. The gravitational $J_{\ell }$ coefficients of a planet (Theorem 7) are structurally identical to the electric $q_{\ell ,m}$ of a charge distribution; nuclear multipoles (Theorem 6) decompose by the same Wigner-Eckart machinery; the CMB anisotropies (Theorem 8) extract cosmological parameters from the same spherical-harmonic decomposition on the sky. The magnetic expansion (Theorem 5) generalises by replacing $\mathbf{P}$ with the magnetization $\mathbf{M}$, and the structural parallel is exact except that no magnetic monopole appears. The Berry-phase reformulation (Theorem 9) identifies the polarization of a crystal with a topological observable of its electronic band structure, dual to the integrated Berry curvature; this generalises the multipole formalism from finite localised sources to extended periodic systems and reaches forward to topological insulators. Vector spherical harmonics (Theorem 10) build toward 10.07.02 pending multipole radiation by replacing the static $1/r^{\ell +1}$ falloff with outgoing-wave Hankel functions, recovering the static expansion in the $k\to 0$ limit and extending it to gravitational-wave radiation in the parallel $\ell =2$ quadrupole framework that LIGO observes today.

Full proof set Master

Proposition 1 (Cartesian expansion to third order). Let $\rho$ be a charge distribution supported in a ball $B_R$ centred at the origin, and let $\mathbf{r}$ satisfy $r>R$. Then

$$\varphi (\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\left[\frac{Q}{r}+\frac{p_i{\hat{r}}_i}{r^2}+\frac{Q_{ij}{\hat{r}}_i{\hat{r}}_j}{2r^3}+\mathcal{O}(r^{-4})\right],$$

with $Q$, $\mathbf{p}$, $Q_{ij}$ as defined in the Formal definition.

Proof. Expand $1/|\mathbf{r}-{\mathbf{r}}^{\prime }|$ as a Taylor series in ${\mathbf{r}}^{\prime }$ around ${\mathbf{r}}^{\prime }=0$. Define $f({\mathbf{r}}^{\prime })=1/|\mathbf{r}-{\mathbf{r}}^{\prime }|$. Direct differentiation gives

$${\partial }_i^{\prime }f=\frac{r_i-r_i^{\prime }}{|\mathbf{r}-{\mathbf{r}}^{\prime }{|}^3},{\partial }_i^{\prime }{\partial }_j^{\prime }f=-\frac{{\delta }_{ij}}{|\mathbf{r}-{\mathbf{r}}^{\prime }{|}^3}+\frac{3(r_i-r_i^{\prime })(r_j-r_j^{\prime })}{|\mathbf{r}-{\mathbf{r}}^{\prime }{|}^5}.$$

Evaluating at ${\mathbf{r}}^{\prime }=0$:

$$f(0)=\frac{1}{r},({\partial }_i^{\prime }f)(0)=\frac{r_i}{r^3}=\frac{{\hat{r}}_i}{r^2},({\partial }_i^{\prime }{\partial }_j^{\prime }f)(0)=-\frac{{\delta }_{ij}}{r^3}+\frac{3{\hat{r}}_i{\hat{r}}_j}{r^3}.$$

Hence

$$\frac{1}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}=\frac{1}{r}+\frac{r_i^{\prime }{\hat{r}}_i}{r^2}+\frac{r_i^{\prime }{r}_j^{\prime }}{2r^3}(3{\hat{r}}_i{\hat{r}}_j-{\delta }_{ij})+\mathcal{O}(r^{\prime 3}/r^4).$$

Multiplying by $\rho ({\mathbf{r}}^{\prime })$ and integrating:

• Zeroth order: $\int \rho d^3{r}^{\prime }/r=Q/r$.
• First order: $\int r_i^{\prime }{\hat{r}}_i\rho d^3{r}^{\prime }/r^2=p_i{\hat{r}}_i/r^2$.
• Second order: $\int r_i^{\prime }{r}_j^{\prime }(3{\hat{r}}_i{\hat{r}}_j-{\delta }_{ij})\rho d^3{r}^{\prime }/(2r^3)$.

The second-order integrand can be rewritten. Using ${\hat{r}}_i{\hat{r}}_j$ symmetric and the identity ${\hat{r}}_i{\hat{r}}_i=1$:

$$\int r_i^{\prime }{r}_j^{\prime }(3{\hat{r}}_i{\hat{r}}_j-{\delta }_{ij})\rho d^3{r}^{\prime }=\int [3r_i^{\prime }{r}_j^{\prime }{\hat{r}}_i{\hat{r}}_j-r^{\prime 2}]\rho d^3{r}^{\prime }$$ $$=\int [(3r_i^{\prime }{r}_j^{\prime }-{\delta }_{ij}{r}^{\prime 2}){\hat{r}}_i{\hat{r}}_j]\rho d^3{r}^{\prime }=Q_{ij}{\hat{r}}_i{\hat{r}}_j,$$

where in the second step I used ${\delta }_{ij}{\hat{r}}_i{\hat{r}}_j=1$ to write $r^{\prime 2}=r^{\prime 2}{\delta }_{ij}{\hat{r}}_i{\hat{r}}_j$, then absorbed the ${\delta }_{ij}{r}^{\prime 2}$ into the integrand. This is the traceless symmetric tensor $Q_{ij}$ contracted with the unit vectors. Combining the three orders gives the claim. $■$

Proposition 2 (Equivalence of Cartesian quadrupole and $\ell =2$ spherical moments). The five real components of the traceless symmetric Cartesian tensor $Q_{ij}$ are linear combinations of the five complex components $q_{2,m}$, $m\in \{-2,-1,0,1,2\}$, with the real $Q_{ij}$ recovered from the conjugate-symmetric $q_{2,m}$ pair $q_{2,m}$ and $q_{2, -m} = (-1)^m q_{2, m}^$.*

Proof sketch. Direct calculation with explicit spherical harmonics:

$$Y_2^0=\sqrt{\frac{5}{16\pi }}(3{\cos }^2\theta -1),Y_2^{\pm 1}=\mp \sqrt{\frac{15}{8\pi }}\sin \theta \cos \theta e^{\pm i\phi },Y_2^{\pm 2}=\sqrt{\frac{15}{32\pi }}{\sin }^2\theta e^{\pm 2i\phi }.$$

Writing $\sin \theta \cos \phi ={\hat{r}}_x$, $\sin \theta \sin \phi ={\hat{r}}_y$, $\cos \theta ={\hat{r}}_z$:

$$Y_2^0=\sqrt{\frac{5}{16\pi }}(3{\hat{r}}_z^2-1),r^{\prime 2}{Y}_2^0({\theta }^{\prime },{\phi }^{\prime })=\sqrt{\frac{5}{16\pi }}(3r_z^{\prime }{}^2-r^{\prime 2}).$$

The integral against $\rho$ gives $q_{2,0}=\sqrt{5/(16\pi )}\cdot Q_{zz}$. Similar manipulations give

$$q_{2,\pm 1}=\mp \sqrt{\frac{5}{24\pi }}(Q_{xz}\mp i{Q}_{yz}),q_{2,\pm 2}=\sqrt{\frac{5}{96\pi }}(Q_{xx}-Q_{yy}\mp 2i{Q}_{xy}).$$

These five relations express the five spherical components $q_{2,m}$ in terms of the five independent components of the Cartesian quadrupole $Q_{ij}$ (the sixth Cartesian component is fixed by the trace condition $Q_{xx}+Q_{yy}+Q_{zz}=0$). The reality of the Cartesian $Q_{ij}$ is reflected in the conjugate-reality $q_{2,-m}=(-1{)}^m{q}_{2,m}^{\ast }$. The same dictionary at higher $\ell$ relates the $2\ell +1$ components $q_{\ell ,m}$ to the independent components of the rank-$\ell$ totally symmetric traceless Cartesian tensor. $■$

Proposition 3 (Dipole-dipole interaction and the $1/R^3$ falloff). The interaction energy of two ideal dipoles ${\mathbf{p}}_1$ at the origin and ${\mathbf{p}}_2$ at position $\mathbf{R}$ is

$$U_{12}(\mathbf{R})=\frac{1}{4\pi {ϵ}_0{R}^3}[{\mathbf{p}}_1\cdot {\mathbf{p}}_2-3({\mathbf{p}}_1\cdot \hat{\mathbf{R}})({\mathbf{p}}_2\cdot \hat{\mathbf{R}})].$$

Proof. The field at $\mathbf{R}$ due to ${\mathbf{p}}_1$ at the origin is (Exercise 6) ${\mathbf{E}}_1(\mathbf{R})=(1/4\pi {ϵ}_0)[3({\mathbf{p}}_1\cdot \hat{\mathbf{R}})\hat{\mathbf{R}}-{\mathbf{p}}_1]/R^3$. The energy of an ideal dipole ${\mathbf{p}}_2$ in an external field $\mathbf{E}$ is $U=-{\mathbf{p}}_2\cdot \mathbf{E}$. Substituting:

$$U_{12}=-{\mathbf{p}}_2\cdot {\mathbf{E}}_1(\mathbf{R})=-\frac{1}{4\pi {ϵ}_0{R}^3}[3({\mathbf{p}}_1\cdot \hat{\mathbf{R}})({\mathbf{p}}_2\cdot \hat{\mathbf{R}})-{\mathbf{p}}_1\cdot {\mathbf{p}}_2].$$

Reorganising the sign gives the stated form. $■$

The thermal average of this interaction over the two dipole orientations (assuming both dipoles rotate freely with Boltzmann weight $e^{-U_{12}/k_{B}T}$) gives, to leading order in $1/k_{B}T$,

$$⟨U_{12}{⟩}_{\text{thermal}}=-\frac{2}{3}\frac{p_1^2{p}_2^2}{(4\pi {ϵ}_0{)}^2{k}_{B}T{R}^6},$$

the Keesom interaction, one of the three components (along with Debye and London) of the total van der Waals interaction. The $1/R^6$ falloff at thermal-averaged level is a signature of the dipole-dipole correlation; the $1/R^3$ falloff of the un-averaged Hamiltonian remains visible in oriented systems (liquid crystals, polar solvent shells).

Proposition 4 (Larmor formula for dipole radiation). An oscillating electric dipole $\mathbf{p}(t)=p_0\hat{\mathbf{z}}\cos (\omega t)$ radiates time-averaged total power

$$⟨P⟩=\frac{{\mu }_0{\omega }^4{p}_0^2}{12\pi c}.$$

Proof sketch. In the radiation zone $kr\gg 1$ (with $k=\omega /c$), the Liénard-Wiechert form of the retarded fields for a sinusoidally oscillating dipole gives transverse electric and magnetic fields

$${\mathbf{E}}_{\text{rad}}=-\frac{{\mu }_0{\omega }^2{p}_0\sin \theta }{4\pi r}\cos (\omega t-kr)\hat{\mathbf{\theta }},|{\mathbf{B}}_{\text{rad}}|=|{\mathbf{E}}_{\text{rad}}|/c.$$

The Poynting vector magnitude is $|\mathbf{S}|=E^2/({\mu }_{0}c)$. The time-average is

$$⟨|\mathbf{S}|⟩=\frac{1}{{\mu }_{0}c}\cdot \frac{1}{2}{\left(\frac{{\mu }_0{\omega }^2{p}_0\sin \theta }{4\pi r}\right)}^2=\frac{{\mu }_0{\omega }^4{p}_0^2{\sin }^2\theta }{32{\pi }^{2}c{r}^2}.$$

Integrating over a sphere of radius $r$ at infinity:

$$⟨P⟩={\int }_0^{\pi }{\int }_0^{2\pi }⟨|\mathbf{S}|⟩r^2\sin \theta d\theta d\phi =\frac{{\mu }_0{\omega }^4{p}_0^2}{32{\pi }^{2}c}{\int }_0^{\pi }{\sin }^3\theta d\theta \cdot 2\pi .$$

Using ${\int }_0^{\pi }{\sin }^3\theta d\theta =4/3$:

$$⟨P⟩=\frac{{\mu }_0{\omega }^4{p}_0^2}{32{\pi }^{2}c}\cdot \frac{8\pi }{3}=\frac{{\mu }_0{\omega }^4{p}_0^2}{12\pi c}.■$$

This is the Larmor formula for dipole radiation, first derived in Larmor 1897 Phil. Mag. (5) 44, in connection with the magnetic-spectrum shift problem that came to be called the Zeeman effect. Its ${\omega }^4$ dependence drives the colour of the sky (Rayleigh 1871) and the strength of atomic spectral-line emissions; the same formula governs synchrotron radiation in particle accelerators (with $\omega$ replaced by the cyclotron frequency and $p_0$ by the appropriate magnetic-moment / acceleration combination), and the bremsstrahlung emission of decelerating charges in plasma.

Connections Master

• Coulomb's law and Gauss's law 10.01.01. The multipole expansion starts from the Coulomb form $\varphi =(1/4\pi {ϵ}_0)\int \rho /|\mathbf{r}-{\mathbf{r}}^{\prime }|d^3{r}^{\prime }$ derived in 10.01.01, and the convergence of the Taylor expansion in $r^{\prime }/r$ relies precisely on the localised-source assumption that Gauss's law expresses globally. The far-field behaviour identified by the multipole expansion is dual to the Gauss-law accounting that organises the near-field through closed-surface flux integrals.

• Dielectrics, polarization, and the displacement field 10.01.04. The induced dipole moment $\mathbf{p}$ in a polarized medium is the leading-order response of any localized sample to an applied field, and the dielectric-sphere result of 10.01.04 (induced dipole $\mathbf{p}=4\pi {ϵ}_0{R}^3(\kappa -1)/(\kappa +2){\mathbf{E}}_{\infty }$) is a worked example of the multipole formalism in action. The polarization-density definition $\mathbf{P}$ in 10.01.04 is the local-volume version of the global dipole moment $\mathbf{p}$ defined here.

• Spherical harmonics and Legendre polynomials 12.05.02. The expansion machinery of this unit (Legendre polynomials, generating function, spherical-harmonic addition theorem) is supplied by 12.05.02; the irreducibility of the $Y_{\ell }^m$ under $SO(3)$ is the representation-theoretic backbone of the multipole expansion. The quantum-mechanical angular-momentum eigenstates of 12.05.02 are exactly the multipole basis at fixed $\ell ,m$.

• Multipole radiation 10.07.02 pending (pending). The time-varying generalisation of this unit. The static expansion is the $\omega \to 0$ limit of the radiative-multipole expansion; the Larmor formula (Theorem 4 of this unit) is the dipole-level radiation result; the full vector-spherical-harmonic framework (Theorem 10) organises electric and magnetic multipole radiation through the Hansen 1935 vector spherical harmonics.

• Method of images 10.01.06 pending (pending). An alternative solution technique for boundary-value problems with conducting or dielectric boundaries. Image methods construct exact solutions by exploiting the geometry of mirror reflections; the multipole expansion characterises the far-field of any such solution. In simple cases (point charge near a grounded plane, point charge near a sphere), the image charges can be read off as multipole moments in a transformed coordinate system.

• Laplace equation and boundary-value problems 10.01.02. The multipole expansion is the spherical-harmonic / Legendre-polynomial solution of Laplace's equation in the exterior of a localised source, with the multipole moments $q_{\ell ,m}$ playing the role of arbitrary constants in the general harmonic solution. The interior expansion (with $r^{\ell }$ rather than $1/r^{\ell +1}$) is the dual problem of 10.01.02 for cavity electrostatics.

Historical & philosophical context Master

The multipole expansion has its origin in the eighteenth-century gravitational programme, not the nineteenth-century electrical one. Legendre 1785 ^{[Legendre1785]} in the Mémoires de Mathématique et de Physique présentés à l'Académie Royale des Sciences introduced what are now called the Legendre polynomials in the context of expanding the gravitational potential of a homogeneous spheroid; Laplace 1782 ^{[Laplace1782]} in his Théorie des attractions des sphéroïdes et de la figure des planètes developed the spherical-harmonic / multipole framework for planetary potentials, recognising the polynomials' role as the angular basis on $S^2$ for harmonic functions. The framework predated Maxwell's electromagnetic theory by nearly a century; when Maxwell 1873 ^{[Maxwell1873]} in A Treatise on Electricity and Magnetism applied it to electrostatics in Volume I §§129-135, he found a near-perfect transcription: the gravitational and electrostatic problems share the inverse-square law, and the multipole apparatus carries over verbatim with charge replacing mass.

The electrostatic multipole formalism was extensively developed in the nineteenth century by William Thomson (Lord Kelvin) 1845 ^{[Thomson1845]}, whose Reprint of Papers on Electrostatics and Magnetism assembled his and others' results on multipole moments, image methods, and the general boundary-value theory in ${\mathbb{R}}^3$. The radiative generalisation began with Larmor 1897 ^[Larmor1897] in Philosophical Magazine (5) 44, who derived the power radiated by an oscillating dipole and recognised the ${\omega }^4$ dependence that Rayleigh had postulated for atmospheric scattering. Hansen 1935 ^[Hansen1935] in Physical Review 47 introduced the vector spherical harmonics that organise the multipole expansion of electromagnetic radiation in a manifestly $SO(3)$-irreducible basis, replacing earlier scalar-by-scalar treatments. Bocchieri and Loinger 1957 ^{[BocchieriLoinger1957]} in Nuovo Cimento 5 placed the classical-dipole dynamics in a precise geomagnetic context, working out the spin-orbit dynamics of a magnetic dipole moving in the geomagnetic field.

The mid-twentieth-century nuclear-physics era brought the multipole expansion into atomic and nuclear spectroscopy, where it organised electric and magnetic multipole transitions, fixed selection rules through the Wigner-Eckart theorem, and revealed nuclear deformations as multipole moments of the rotating nuclear surface. Bohr and Mottelson 1953 ^{[BohrMottelson1953]} in Mat.-Fys. Medd. Dan. Vid. Selsk. 27 (no. 16) introduced the collective model of nuclear quadrupole deformations, modelling each nucleus as a deformable liquid drop whose surface admits a multipole decomposition with the quadrupole as the leading collective mode; this synthesis of the multipole formalism with collective rotational dynamics earned Bohr, Mottelson, and Rainwater the 1975 Nobel Prize in Physics. The contemporary endpoint is the Berry-phase reformulation of polarization in periodic crystals (King-Smith and Vanderbilt 1993 ^{[KingSmithVanderbilt1993]} in Phys. Rev. B 47), which moves the multipole formalism from finite localised systems to infinite periodic crystals and identifies polarization with a topological observable of the electronic band structure — the bridge to topological insulators and modern condensed-matter physics.

Bibliography Master

@article{Legendre1785,
  author    = {Legendre, Adrien-Marie},
  title     = {Recherches sur l'attraction des sph{\'e}ro{\"i}des homog{\`e}nes},
  journal   = {M{\'e}moires de Math{\'e}matique et de Physique pr{\'e}sent{\'e}s {\`a} l'Acad{\'e}mie Royale des Sciences},
  volume    = {10},
  year      = {1785},
  pages     = {411--434},
}

@article{Laplace1782,
  author    = {Laplace, Pierre-Simon},
  title     = {Th{\'e}orie des attractions des sph{\'e}ro{\"i}des et de la figure des plan{\`e}tes},
  journal   = {M{\'e}moires de l'Acad{\'e}mie des Sciences de Paris},
  year      = {1782},
}

@book{Thomson1845,
  author    = {Thomson, William (Lord Kelvin)},
  title     = {Reprint of Papers on Electrostatics and Magnetism},
  publisher = {Macmillan, London},
  year      = {1872},
  note      = {Reprint of papers from 1845 onward},
}

@book{Maxwell1873,
  author    = {Maxwell, James Clerk},
  title     = {A Treatise on Electricity and Magnetism},
  publisher = {Clarendon Press, Oxford},
  year      = {1873},
  volume    = {1--2},
}

@article{Larmor1897,
  author    = {Larmor, Joseph},
  title     = {On the theory of the magnetic influence on spectra; and on the radiation from moving ions},
  journal   = {Philosophical Magazine, Series 5},
  volume    = {44},
  year      = {1897},
  pages     = {503--512},
}

@article{Hansen1935,
  author    = {Hansen, W. W.},
  title     = {A new type of expansion in radiation problems},
  journal   = {Physical Review},
  volume    = {47},
  year      = {1935},
  pages     = {139--143},
}

@article{BohrMottelson1953,
  author    = {Bohr, Aage and Mottelson, Ben R.},
  title     = {Collective and individual-particle aspects of nuclear structure},
  journal   = {Matematisk-Fysiske Meddelelser, Det Kongelige Danske Videnskabernes Selskab},
  volume    = {27},
  number    = {16},
  year      = {1953},
}

@article{BocchieriLoinger1957,
  author    = {Bocchieri, P. and Loinger, A.},
  title     = {Sul moto di un dipolo magnetico in un campo magnetico},
  journal   = {Il Nuovo Cimento},
  volume    = {5},
  year      = {1957},
  pages     = {1224},
}

@book{Jackson1999,
  author    = {Jackson, John David},
  title     = {Classical Electrodynamics},
  edition   = {3},
  publisher = {Wiley},
  year      = {1999},
}

@book{Griffiths2013,
  author    = {Griffiths, David J.},
  title     = {Introduction to Electrodynamics},
  edition   = {4},
  publisher = {Cambridge University Press},
  year      = {2013},
}

@book{LandauLifshitzVol8,
  author    = {Landau, Lev Davidovich and Lifshitz, Evgeny Mikhailovich},
  title     = {Electrodynamics of Continuous Media},
  series    = {Course of Theoretical Physics},
  volume    = {8},
  edition   = {2},
  publisher = {Pergamon},
  year      = {1984},
}

@article{KingSmithVanderbilt1993,
  author    = {King-Smith, R. D. and Vanderbilt, David},
  title     = {Theory of polarization of crystalline solids},
  journal   = {Physical Review B},
  volume    = {47},
  year      = {1993},
  pages     = {1651--1654},
}