You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In classical electrodynamics, the fundamental entities are the electric field $\mathbf{E}$ and the magnetic field $\mathbf{B}$. These fields are mathematically derived from a scalar potential $\phi$ (the electrostatic equivalent of classical voltage potential) and a vector potential $\mathbf{A}$. The fields are defined as:
Bot E and B can be written as the four-dimensional extension of the curl operator, formally known as the 4-curl or the exterior derivative.
The electric field is not just "kind of like" the curl of the time and spatial components; it is the exact mathematical counterpart to the magnetic field when the curl operation is extended into four-dimensional spacetime.
1. The 4-Curl
In standard 3D vector calculus, the curl of the vector potential $\mathbf{A}$ generates the magnetic field via the antisymmetric combination of spatial derivatives:
$$B_k = \partial_i A_j - \partial_j A_i$$
In covariant formulation, this antisymmetric derivative is generalized across all four dimensions of spacetime to operate on the complete 4-potential $A_\mu$. This operation generates the electromagnetic field tensor $F_{\mu\nu}$:
When you evaluate the purely spatial components (e.g., $\mu=1, \nu=2$), you get the standard 3D curl, yielding the magnetic field components $B_k$. When you evaluate the mixed time-space components (e.g., $\mu=0, \nu=i$), you get the electric field.
2. Resolving the Sign Discrepancy
You correctly observed that in the classical equation $\mathbf{E} = -\nabla\phi - \frac{\partial \mathbf{A}}{\partial t}$, the two terms share the same sign (an overall negative), whereas a traditional curl features a subtraction between terms.
The fundamental 4-curl equation $F_{\mu\nu} = \partial_\mu A_\nu - \partial_\nu A_\mu$does use a minus sign. The reason the final 3D equation looks different is due to the Minkowski metric signature $(+,-,-,-)$, which flips the sign of the spatial components when lowering indices.
Here is the step-by-step derivation showing how the standard subtraction in the 4-curl mathematically produces the two negative signs for the electric field.
Step A: Define the Covariant Vectors
The contravariant 4-potential is defined with positive spatial components:
$$A^\mu = (\phi, \mathbf{A})$$
To use it in the covariant 4-curl equation, we lower the index using the metric, which negates the spatial components:
The fundamental operation generating the electric field is identical in mathematical structure to the operation generating the magnetic field. The apparent discrepancy in the signs in 3D vector calculus is strictly an artifact of flattening 4D covariant tensor math down into 3D vectors.
Two of Maxwell's Laws As Identities
The definitions of E and B above automatically satisfy two of the four governing equations of the field—the homogeneous Maxwell equations—strictly through vector identities:
Gauss's Law for Magnetism:$\nabla \cdot \mathbf{B} = 0$. This is guaranteed because the divergence of a curl is inherently zero ($\nabla \cdot (\nabla \times \mathbf{A}) = 0$). This also dictates the absence of magnetic monopoles.
Faraday's Law:$\nabla \times \mathbf{E} + \frac{\partial \mathbf{B}}{\partial t} = 0$. Substituting the potential definitions yields $-\nabla \times (\nabla\phi) - \frac{\partial}{\partial t}(\nabla \times \mathbf{A}) + \frac{\partial}{\partial t}(\nabla \times \mathbf{A}) = 0$, which holds true because the curl of a gradient is always zero.
These two homogenous equations are combined into a single, elegant tensor equation known as the Bianchi identity.
Expressed in terms of the covariant electromagnetic field tensor $F_{\mu\nu}$, the identity is:
This can be proved by simply expanding the electromagnetic tensor components in terms of the 4-vector potential components: $$F_{\mu\nu} = \partial_\mu A_\nu - \partial_\nu A_\mu$$. All components cancel each other out.
The fact that these two equations are homogeneous (equal to zero) is the direct mathematical expression of the experimental observation that isolated magnetic charges (monopoles) and magnetic currents do not exist.
If magnetic monopoles and a corresponding magnetic current density did exist, the framework of classical electrodynamics would become perfectly symmetric, and the homogeneous equations would become inhomogeneous.
Here is how the equations would change if we introduce a magnetic charge density $\rho_m$ and a magnetic current density $\mathbf{J}_m$.
Gauss's Law for Magnetism
Currently, the equation states that the magnetic field has no divergence, meaning magnetic field lines never originate or terminate at a point:
$$\nabla \cdot \mathbf{B} = 0$$
If magnetic monopoles existed, they would act as point sources or sinks for the magnetic field, exactly as electric charges do for the electric field. The equation would become inhomogeneous:
$$\nabla \cdot \mathbf{B} = \mu_0 \rho_m$$
Faraday's Law
Currently, the equation states that a time-varying magnetic field induces a circulating electric field, with no independent source term:
If magnetic charges existed and were in motion, they would constitute a magnetic current $\mathbf{J}_m$. Just as an electric current generates a circulating magnetic field (Ampere's Law), a magnetic current would generate a circulating electric field. The equation would become inhomogeneous:
(The negative sign is a convention required to maintain energy conservation and the directionality defined by Lenz's Law).
The Collapse of the Potential Framework
The non-existence of magnetic monopoles is precisely what allows us to define the magnetic field as the curl of a vector potential ($\mathbf{B} = \nabla \times \mathbf{A}$).
Because the divergence of any curl is strictly zero ($\nabla \cdot (\nabla \times \mathbf{A}) = 0$), defining $\mathbf{B}$ this way fundamentally bakes the non-existence of monopoles into the mathematics. If magnetic monopoles were discovered, the definition $\mathbf{B} = \nabla \times \mathbf{A}$ would have to be modified or abandoned entirely, because it mathematically prevents $\nabla \cdot \mathbf{B}$ from ever equalling a non-zero $\rho_m$.
Covariant Formulation and the Continuity Equation
The remaining two inhomogeneous Maxwell equations govern how the fields are generated by charge density $\rho$ and current density $\mathbf{J}$:
Though these two equations were originally established by experimental observation, the two inhomogeneous laws can be derived directly from the principle of least action applied to the sum of the free-field and interaction Lagrangian densities.
Relativistically, the potentials and sources are grouped into 4-vectors. The 4-potential is $A^\mu = (\phi/c, \mathbf{A})$, and the 4-current density is $J^\mu = (c\rho, \mathbf{J})$.
$\phi$ is the electric scalar potential. In purely electrostatic situations (where $\frac{\partial \mathbf{A}}{\partial t} = 0$), $\phi$ is exactly equivalent to the classical voltage potential $V$, and $\mathbf{E} = -\nabla\phi$. $A$ is the magnetic vector potential.
The time component of $J^\mu$ represents the charge density, while the spatial components represent the 3D current density.
From the inhomogeneous Maxwell equations, we can derive the continuity equation or conservation of charge. Conservation of charge is expressed by the continuity equation, which states that the rate of change of charge density equals the negative divergence of current density ($\frac{\partial \rho}{\partial t} = -\nabla \cdot \mathbf{J}$). In covariant notation, this is written as:
$$\partial_\mu J^\mu = 0$$
The fields $\mathbf{E}$ and $\mathbf{B}$ are encapsulated within a rank-2 antisymmetric tensor called the electromagnetic field tensor $F^{\mu\nu}$:
Here, the spatial components $F^{ij}$ correspond to $\mathbf{B}$, and the mixed space-time components $F^{0i}$ correspond to $\mathbf{E}$.
Local charge conservation is a fundamental, inescapable requirement of the antisymmetry of the electromagnetic field tensor. If the field equations are true, the continuity equation must also be true.
The Field Lagrangian Density
The dynamics of the field are determined by the principle of least action. The total Lagrangian density for a field interacting with matter consists of a free-field component and an interaction component.
which works out to $$\mathcal{L} = \frac{1}{2}(E^2 - B^2)$$
The interaction component couples the electromagnetic field to the matter field (the charge and current densities). This interaction term is where the charge density enters the Lagrangian density:
Applying the Euler-Lagrange equations to the total Lagrangian density $\mathcal{L} = \mathcal{L}{\text{field}} + \mathcal{L}{\text{int}}$ yields the inhomogeneous Maxwell equations ($\partial_\mu F^{\mu\nu} = \mu_0 J^\nu$).
How Lagrangians and Actions Vary for Particles & Fields
In classical mechanics, we take the action as the integral of the lagrangian. We solve for the EOM which is essentially how the particle coordinates vary with time. So the degree of freedom that we allow variation in while evaluating the principle of least action is the path i.e. coordinates (and velocities etc) between the start and end time.
For Lagrangians of fields, we are trying to solve for the EOM which is essentiually how the field value (and direction for vectors) varies over time and coordinate dimensions. So the degree of freedom that we allow variation in while evaluating the principle of least action is the actual field value and its various derivatives.
Yes, this is exactly correct. The transition from particle mechanics to field theory requires a fundamental mathematical shift in how a "degree of freedom" is defined and what is being varied.
The Mathematical Mapping
In classical point-particle mechanics, spatial coordinates are the dependent variables (the quantities that change), and time is the sole independent parameter.
In field theory, space is no longer a dependent variable; it joins time as an independent background parameter. The new dependent variable—the degree of freedom—is the field amplitude itself.
Fixed boundary surface: $\delta \phi = 0$ on $\partial\Omega$
The Variation Process
When extremizing the action for a particle, you hold the starting and ending positions fixed at specific times and vary the path $q(t)$ through space. The calculus of variations identifies the specific path where the action is stationary.
When extremizing the action for a field, you evaluate a four-dimensional spacetime volume $\Omega$. You hold the field values rigidly fixed on the boundary surface of this volume ($\partial\Omega$), and you vary the field amplitude $\phi(t,x,y,z)$ at every interior point in spacetime.
You are not varying a spatial trajectory; you are varying the intensity (or vector direction) of the field at every coordinate simultaneously. Because the continuous spatial label $\vec{x}$ effectively replaces the discrete index $i$ found in mechanical coordinates $q_i$, a field possesses an infinite number of degrees of freedom. The resulting generalized Euler-Lagrange equations act as an infinite set of equations of motion—one for every point in space—manifesting collectively as a single partial differential equation.
Deriving the Inhomogeneous Equations from Least Action
The action $S$ of the electromagnetic field interacting with a current source is the four-dimensional integral of the Lagrangian density:
To find the equations of motion for the field, we apply the principle of least action ($\delta S = 0$). This requires the total Lagrangian density to satisfy the Euler-Lagrange equation for the continuous field components $A_\mu$:
Next, evaluate the derivative with respect to the field gradients $\partial_\nu A_\mu$. Only the free-field term contains the gradients (embedded within $F_{\alpha\beta}$):
Because $F^{\alpha\beta} = \partial^\alpha A^\beta - \partial^\beta A^\alpha$, applying the chain rule to the product $F_{\alpha\beta}F^{\alpha\beta}$ isolates the specific indices $\nu$ and $\mu$, yielding:
Rearranging this and recognizing that $F^{\nu\mu} = -F^{\mu\nu}$ (since the electromagnetic tensor is antisymmetric) produces the covariant form of the inhomogeneous Maxwell equations:
$$\partial_\nu F^{\mu\nu} = \mu_0 J^\mu$$
Expanding this into its time ($\mu=0$) and spatial ($\mu=1,2,3$) components recovers Gauss's Law ($\nabla \cdot \mathbf{E} = \frac{\rho}{\epsilon_0}$) and the Ampere-Maxwell Law ($\nabla \times \mathbf{B} - \frac{1}{c^2}\frac{\partial \mathbf{E}}{\partial t} = \mu_0 \mathbf{J}$), respectively.
The Discrete Particle Lagrangian and Lorentz Force
When moving from continuous charge distributions to a discrete point particle of mass $m$ and charge $q$, the action must account for the particle's free kinematics and its specific interaction with the field.
The discrete interaction Lagrangian $L_{\text{int}}$ is obtained by integrating the interaction density over all 3D space. For a point particle at position $\mathbf{r}_p(t)$ with velocity $\mathbf{v}$, the densities are modeled using Dirac delta functions: $\rho = q\delta^3(\mathbf{r} - \mathbf{r}_p)$ and $\mathbf{J} = q\mathbf{v}\delta^3(\mathbf{r} - \mathbf{r}_p)$. Integrating $\int (-\rho\phi + \mathbf{J} \cdot \mathbf{A}) d^3x$ collapses the fields to the particle's location, yielding:
This interaction term acts as a velocity-dependent generalized potential. The kinematic component for the free particle must be relativistically invariant. The correct term is derived by ensuring the action is proportional to the Lorentz-invariant proper time $d\tau = dt/\gamma$:
While it may seem counterintuitive that this term approaches zero as the velocity approaches $c$, it is mathematically correct because physical observables are derived quantities, not equal to the Lagrangian itself. For example, the conjugate momentum $p = \frac{\partial L}{\partial v} = \gamma m v$ and energy $H = pv - L = \gamma m c^2$ both correctly approach infinity as the velocity approaches $c$.
Combining the free particle and interaction terms gives the total Lagrangian for a single charged particle in an external electromagnetic field:
In Lagrangian mechanics, forces are not direct inputs but are derived outputs. Applying the Euler-Lagrange equations to this complete Lagrangian naturally generates the correct equations of motion, yielding the Lorentz force law:
When $\frac{\partial \mathbf{A}}{\partial t} \neq 0$, the system is dynamic (electrodynamics), and $\phi$ undergoes a fundamental shift in its physical meaning. It is no longer the sole source of the electric field, and it ceases to be equivalent to the classical, path-independent "voltage" $V$.
Here is exactly what $\phi$ represents in a time-varying system.
1. The Decomposition of the Electric Field
In electrodynamics, the total electric field is defined as:
This equation splits the electric field into two distinct physical origins:
The Irrotational Component ($-\nabla\phi$): This is the part of the electric field generated directly by the distribution of electric charges ($\rho$).
The Solenoidal Component ($-\frac{\partial \mathbf{A}}{\partial t}$): This is the induced electric field generated by time-varying currents (changing magnetic fields), as dictated by Faraday's Law.
Therefore, when $\frac{\partial \mathbf{A}}{\partial t} \neq 0$, $\phi$ strictly represents the potential arising from the charge distribution, completely stripped of any induced electric effects caused by changing magnetic fields.
2. The Breakdown of Classical "Voltage"
In electrostatics, the electric field is conservative ($\nabla \times \mathbf{E} = 0$). This means the work done moving a charge between two points is independent of the path taken. You can measure a unique "voltage drop" $V$ between point A and point B, which is exactly equal to the difference in $\phi$.
In electrodynamics, the changing magnetic field makes the electric field non-conservative ($\nabla \times \mathbf{E} \neq 0$). If you move a charge from point A to point B, the work done (and thus the measured voltage on a voltmeter) depends entirely on the path the wire takes through the changing magnetic field.
Consequently, a unique "voltage" between two points no longer exists. $\phi$ becomes a mathematical scalar potential used to calculate the charge-driven portion of the field, rather than a directly measurable energy difference.
3. Gauge Dependence of $\phi$
In time-varying fields, the exact physical interpretation of $\phi$ depends on the mathematical gauge you choose to solve Maxwell's equations. Because $\mathbf{E}$ and $\mathbf{B}$ remain unchanged if you simultaneously adjust $\phi$ and $\mathbf{A}$ in specific ways (gauge transformations), you can force $\phi$ to behave differently:
In the Coulomb Gauge ($\nabla \cdot \mathbf{A} = 0$):$\phi$ is forced to act exactly like the electrostatic potential. It is calculated via Poisson's equation ($\nabla^2\phi = -\frac{\rho}{\epsilon_0}$). In this gauge, $\phi$ propagates instantaneously across space based on the current charge distribution, ignoring the speed of light. (The retardation effects ensuring causality are entirely absorbed by the $-\frac{\partial \mathbf{A}}{\partial t}$ term).
In the Lorenz Gauge ($\nabla \cdot \mathbf{A} + \frac{1}{c^2}\frac{\partial \phi}{\partial t} = 0$):$\phi$ becomes a "retarded potential." It satisfies a wave equation ($\nabla^2\phi - \frac{1}{c^2}\frac{\partial^2\phi}{\partial t^2} = -\frac{\rho}{\epsilon_0}$). In this gauge, $\phi$ represents the potential caused by charges, but it accounts for causality—the potential at a distance $r$ depends on what the charges were doing at a time $t - \frac{r}{c}$ in the past.
How the 4-Vector Potential and the Rest of the Theory was Derived
The derivation from the relativistic observation of the Lorentz force to the exact definitions of the electric and magnetic fields via the 4-vector potential is a direct consequence of tensor mathematics.
Here is the exact progression of how physicists deduced the formulas for $\mathbf{E}$ and $\mathbf{B}$ starting entirely from the requirement of reference-frame invariance.
1. Upgrading the Force to a 4-Vector
In a laboratory frame, the classical Lorentz force on a particle with charge $q$ and 3-velocity $\mathbf{v}$ is:
To make this physically consistent across all reference frames, the 3-vectors must be upgraded to 4-vectors. The particle's velocity becomes the 4-velocity $U^\mu = \gamma(c, \mathbf{v})$. The mechanical force becomes the 4-force $K^\mu$, which includes the mechanical power (work done by the field) as its time component:
Physicists needed a relativistically invariant equation linking the 4-velocity of the particle to the resulting 4-force. Because the classical force is linear with respect to velocity ($\mathbf{F} \propto \mathbf{v}$), the relativistic 4-force must be a linear transformation of the 4-velocity.
In linear algebra, mapping one vector to another requires a rank-2 tensor. This defines the electromagnetic tensor $F^{\mu\nu}$:
$$K^\mu = q F^{\mu\nu} U_\nu$$
A fundamental property of magnetic fields is that they do no work ($\mathbf{F}_{\text{mag}} \cdot \mathbf{v} = 0$). For the tensor equation to uphold this physical reality in all frames, the tensor $F^{\mu\nu}$ must be perfectly antisymmetric: $F^{\mu\nu} = -F^{\nu\mu}$.
By expanding $K^\mu = q F^{\mu\nu} U_\nu$ into its spatial and temporal components and comparing it directly to $\mathbf{F} = q(\mathbf{E} + \mathbf{v} \times \mathbf{B})$, physicists identified exactly how $\mathbf{E}$ and $\mathbf{B}$ fit into the tensor (using the metric signature $+ - - -$):
This matrix proves the user's premise: $\mathbf{E}$ and $\mathbf{B}$ are not separate fields, but interwoven components of a single unified tensor $F^{\mu\nu}$. A Lorentz boost (changing reference frames) simply mixes the temporal and spatial components of this matrix, transforming what looks like a pure magnetic field in one frame into a mixture of electric and magnetic fields in another.
3. The Mathematical Guarantee of the 4-Potential
The next step relies on the homogeneous Maxwell equations, which were empirically known: $\nabla \cdot \mathbf{B} = 0$ (no monopoles) and $\nabla \times \mathbf{E} + \frac{\partial \mathbf{B}}{\partial t} = 0$ (Faraday's Law).
When these two equations are written using the tensor $F^{\mu\nu}$, they combine into a single, elegant covariant identity known as the Bianchi identity:
At this point, physicists invoked Poincaré's Lemma from differential geometry. The lemma states that if an antisymmetric tensor satisfies this exact identity (meaning the differential form is "closed"), it can always be expressed as the derivative of a vector field (meaning it is "exact").
Therefore, there is a mathematical guarantee that a 4-vector field $A^\mu = (\phi/c, \mathbf{A})$ must exist such that the tensor is strictly defined by its gradients:
With the relationship between the fields ($F_{\mu\nu}$) and the 4-potential ($A_\mu$) established, the classical formulas for $\mathbf{E}$ and $\mathbf{B}$ are derived simply by unpacking the components.
The covariant 4-potential is $A_\mu = (\phi/c, -A_x, -A_y, -A_z)$, and the covariant derivative operator is $\partial_\mu = (\frac{1}{c}\frac{\partial}{\partial t}, \nabla)$.
For the Electric Field:
The electric field components live in the mixed time-space indices $F_{0i}$. We calculate this directly from the potential definition:
$$F_{0i} = \partial_0 A_i - \partial_i A_0$$
Substituting the definitions of the operators and potentials:
The term inside the parentheses is the exact geometric definition of the components of a curl operator. Mapping this to the spatial block of the electromagnetic tensor yields the exact formula for the magnetic field:
$$\mathbf{B} = \nabla \times \mathbf{A}$$
5. The 4-Force and Zero Work by Magnetic Fields
$$K^\mu = \gamma\left(\frac{\mathbf{F} \cdot \mathbf{v}}{c}, \mathbf{F}\right)$$
The time component must be zero in the rest frame right? Because 4-acceleration in the rest frame has zero time component? Also is this related to magnetic fields doing no work? Because F and v are orthogonal to each other
Both of your deductions are physically and mathematically exact.
1. The Rest Frame and 4-Acceleration
In the particle's rest frame, the 3-velocity is mathematically strictly zero ($\mathbf{v} = 0$), which sets the Lorentz factor to $\gamma = 1$. The time component of the 4-force evaluates directly to zero: $K^0 = \frac{\mathbf{F} \cdot \mathbf{0}}{c} = 0$.
The connection to 4-acceleration is the underlying reason for this. The 4-force is the mass scaled by the 4-acceleration:
$$K^\mu = m a^\mu$$
The 4-acceleration is the derivative of the 4-velocity $U^\mu = \gamma(c, \mathbf{v})$ with respect to proper time $\tau$. The time component of the 4-acceleration is:
$$a^0 = c \frac{d\gamma}{d\tau}$$
By applying the chain rule to the Lorentz factor $\gamma = (1 - \frac{v^2}{c^2})^{-1/2}$, the derivative evaluates to:
where $\mathbf{a}$ is the standard 3-acceleration.
When this is evaluated in the rest frame ($\mathbf{v} = 0$), the entire term collapses to zero. Therefore, $a^0 = 0$, which rigorously enforces $K^0 = m a^0 = 0$. The time component of the 4-force must vanish in the rest frame because the particle's rest mass energy cannot change in its own frame.
2. Magnetic Fields and Work
The term $\mathbf{F} \cdot \mathbf{v}$ represents the mechanical power delivered to the particle (the rate of change of its kinetic energy). The orthogonality of the magnetic force and velocity guarantees that the magnetic field contributes nothing to this component in any reference frame, not just the rest frame.
Substituting the Lorentz force $\mathbf{F} = q(\mathbf{E} + \mathbf{v} \times \mathbf{B})$ into the power term expands to:
By definition, the cross product $\mathbf{v} \times \mathbf{B}$ produces a vector that is strictly orthogonal to both $\mathbf{v}$ and $\mathbf{B}$. The dot product of orthogonal vectors is zero, eliminating the magnetic term completely:
The time component of the 4-force simplifies to $K^0 = \gamma \frac{q(\mathbf{E} \cdot \mathbf{v})}{c}$. The magnetic field does zero work and cannot alter the kinetic energy of the particle. Any change to the time component of the particle's 4-momentum is driven exclusively by the electric field.
The time component of the 4-force, $K^0$, represents the rate at which the electromagnetic field transfers energy to the particle. Specifically, it is the mechanical power delivered to the particle by the electric field, scaled by the relativistic factor $\gamma/c$.
Here is the explicit tensor contraction showing exactly how this emerges from $K^\mu = q F^{\mu\nu} U_\nu$.
3. The Tensor Contraction for $K^0$
To find the time component, we set $\mu = 0$ and sum over the index $\nu$ (from 0 to 3):
Because proper time $d\tau$ relates to coordinate time $dt$ by $dt = \gamma d\tau$, we can divide both sides by $\gamma$ to see the rate of energy change from the perspective of the laboratory frame:
Only the electric field ($\mathbf{E}$) can change the energy of the particle. The magnetic field components ($B_i$) reside purely in the spatial block $F^{ij}$ and simply do not appear in the $K^0$ equation.
The energy transferred is exactly the classical mechanical power (Force $\cdot$ velocity).
The classical Lorentz force is the non-relativistic 3D vector component embedded within the spatial part of the relativistic 4-force, scaled by the Lorentz factor.
Here is the exact mathematical relationship between the standard Lorentz force and its fully covariant 4-vector formulation.
1. The Classical Lorentz Force
In classical electrodynamics, the force $\mathbf{F}$ exerted on a particle with charge $q$ moving with a 3-velocity $\mathbf{v}$ is:
This 3-force dictates the rate of change of the particle's 3-momentum with respect to coordinate time $t$: $\mathbf{F} = \frac{d\mathbf{p}}{dt}$.
2. The General 4-Force
In special relativity, to ensure the laws of physics are invariant across all inertial reference frames, vectors must be upgraded to 4-vectors. The 4-force $K^\mu$ is defined as the derivative of the 4-momentum $p^\mu = (E/c, \mathbf{p})$ with respect to proper time $\tau$.
Because coordinate time $t$ and proper time $\tau$ are related by $dt = \gamma d\tau$ (where $\gamma$ is the Lorentz factor $(1 - v^2/c^2)^{-1/2}$), the general relationship between any 3-force and its corresponding 4-force is:
Since the rate of change of energy $\frac{dE}{dt}$ is the mechanical power $\mathbf{F} \cdot \mathbf{v}$, and $\frac{d\mathbf{p}}{dt}$ is the classical force $\mathbf{F}$, the general 4-force is:
This shows that the classical Lorentz force $\mathbf{F}$ makes up the spatial components of the 4-force, multiplied by $\gamma$.
3. The Covariant Lorentz Force Equation
To calculate this 4-force strictly from electromagnetic fields without relying on non-relativistic 3-vectors, physicists use the covariant formulation. The 4-force is the contraction of the electromagnetic field tensor $F^{\mu\nu}$ with the particle's 4-velocity $U_\nu$:
$$K^\mu = q F^{\mu\nu} U_\nu$$
$F^{\mu\nu}$ is the rank-2 antisymmetric tensor containing all components of $\mathbf{E}$ and $\mathbf{B}$.
$U_\nu$ is the covariant 4-velocity, $U_\nu = \gamma(c, -v_x, -v_y, -v_z)$ (assuming a $+ - - -$ metric signature).
4. Unpacking the Tensor Equation
If you evaluate the components of $K^\mu = q F^{\mu\nu} U_\nu$, you recover the exact relationships defined in Section 2.
The Time Component ($\mu = 0$):
Evaluating the sum over $\nu$ for the zeroth component yields the rate of work done by the electric field:
This matches the time component of the general 4-force, proving that $\mathbf{F} \cdot \mathbf{v} = q(\mathbf{E} \cdot \mathbf{v})$. The magnetic field does no work.
The Spatial Components ($\mu = i$):
Evaluating the sum over $\nu$ for the spatial components yields the relativistic momentum transfer:
$$K^i = q (F^{i0} U_0 + F^{ij} U_j)$$
Substituting the tensor components ($F^{i0} = E^i/c$ and $F^{ij}$ containing the $\mathbf{B}$ field cross-product terms) gives:
Because $K^i = \gamma \mathbf{F}^i$, dividing out the Lorentz factor $\gamma$ leaves you exactly with the classical Lorentz force equation. The 4-force is simply the relativistically complete version that accurately transforms between reference frames.
Stress-Energy-Momentum Tensor
Here is the complete breakdown of the electromagnetic stress-energy tensor $T^{\mu\nu}$. Your intuitions regarding the energy density, momentum density, and the continuity equation are exactly correct.
Throughout this explanation, we will continue using natural units ($c=1$, $\mu_0=1$, $\epsilon_0=1$) and the metric signature $(+,-,-,-)$.
1. The General Covariant Tensor Form
The symmetric, gauge-invariant stress-energy tensor for the electromagnetic field is expressed entirely in terms of the field tensor $F^{\mu\nu}$ and the metric tensor $\eta^{\mu\nu}$:
Regarding the vector potential $A^\mu$:
You asked how to write this in terms of $A$ and $F$. While you can technically substitute $F_{\alpha\beta} = \partial_\alpha A_\beta - \partial_\beta A_\alpha$ into the equation above, it is strictly standard to write the physical stress-energy tensor using only $F^{\mu\nu}$.
If you derive the energy-momentum tensor directly from the Lagrangian $\mathcal{L}$ and the potential $A^\mu$ using Noether's theorem for translational symmetry, you obtain the "canonical" tensor:
However, this canonical tensor is physically problematic: it is not symmetric ($T^{\mu\nu} \neq T^{\nu\mu}$) and it is not gauge invariant (because it explicitly contains the bare potential $A_\alpha$). To obtain the correct physical tensor $T^{\mu\nu}$ shown first, physicists apply the Belinfante-Rosenfeld procedure, which adds a total divergence to symmetrize the tensor and eliminate the gauge-dependent $A^\mu$ terms, leaving only the gauge-invariant $F^{\mu\nu}$.
2. The $T^{00}$ Component (Energy Density)
Your assumption for $T^{00}$ is correct. To evaluate the time-time component, we set $\mu = 0$ and $\nu = 0$:
(In standard SI units, the Poynting vector is $\frac{1}{\mu_0}\mathbf{E} \times \mathbf{B}$ and the momentum density is $\epsilon_0 \mathbf{E} \times \mathbf{B}$. In natural units, they are identical).
4. Poynting's Theorem (The Continuity Equation)
You are exactly right that the relationship between $T^{00}$ and $T^{0i}$ is governed by a continuity equation. In a region of space with no charges or currents ($J^\mu = 0$), the 4-divergence of the stress-energy tensor is strictly zero, representing the local conservation of energy and momentum:
$$\partial_\mu T^{\mu\nu} = 0$$
If we extract the time component ($\nu = 0$) of this 4-divergence, we get:
$$\partial_0 T^{00} + \partial_i T^{i0} = 0$$
Substituting our physical definitions ($T^{00} = u$ and $T^{i0} = \mathbf{S}$), this becomes exactly Poynting's Theorem for a free field:
This explicitly states that any change in the local energy density of the field must be accompanied by an outward or inward flux of energy (the Poynting vector) across the boundary of that region.
(Note: If sources are present, the divergence is not zero. Instead, $\partial_\mu T^{\mu\nu} = -F^{\nu\alpha}J_\alpha$. For $\nu=0$, the right side evaluates to $-\mathbf{J} \cdot \mathbf{E}$, which is the rate at which the field does mechanical work on the charges, thus draining energy from the field).
5. The Spatial Components $T^{ij}$ (The Maxwell Stress Tensor)
The remaining 9 spatial components ($i, j = 1, 2, 3$) form a $3 \times 3$ symmetric matrix describing the flux of momentum. Specifically, $T^{ij}$ represents the flux of the $i$-th component of momentum passing through a surface of constant $j$.
This $3 \times 3$ block is the negative of the classical Maxwell stress tensor ($\sigma^{ij}$).
The diagonal components ($T^{11}, T^{22}, T^{33}$): Represent isotropic pressures exerted by the field.
The off-diagonal components ($T^{12}, T^{13}$, etc.): Represent shear stresses (forces parallel to the surface).
If you evaluate the spatial components ($\nu = i$) of the conservation law $\partial_\mu T^{\mu i} = -F^{i\alpha}J_\alpha$, you obtain the momentum conservation equation, which relates the divergence of this Maxwell stress tensor to the rate of change of field momentum and the mechanical Lorentz force exerted on matter.
1. The First Term and Canonical Momentum
You correctly recognized the second term $\frac{1}{4}\eta^{\mu\nu}F_{\alpha\beta}F^{\alpha\beta}$ as $-\eta^{\mu\nu}\mathcal{L}$. The first term, $-F^{\mu\alpha}F^\nu_{;;\alpha}$, is directly related to the canonical momentum, but it is not exactly $\int(\pi \dot{A})$. It is the gauge-invariant extension of the Hamiltonian density term.
By definition, the canonical momentum density conjugate to the field $A_\alpha$ is:
When you apply Noether's theorem to derive the energy-momentum tensor strictly from spacetime translations, you get the canonical stress-energy tensor:
For the time-time component ($T^{00}{\text{can}}$), the first term is exactly $\pi^\alpha \dot{A}\alpha$ (the canonical momentum multiplied by the time derivative of the generalized coordinate). This forms the standard Legendre transformation into the Hamiltonian.
However, $T^{\mu\nu}{\text{can}}$ is not symmetric and is gauge-dependent (it explicitly contains the bare $\partial^\nu A\alpha$). To get the physical tensor you wrote, the bare derivative $\partial^\nu A_\alpha$ is replaced by the gauge-invariant field tensor $F^\nu_{;;\alpha}$.
2. The Belinfante-Rosenfeld Procedure
The Belinfante-Rosenfeld procedure is not equivalent to choosing a gauge like $A_0 = 0$ (the temporal gauge). It is a completely gauge-independent mathematical operation.
The procedure adds a specific identically conserved quantity (a total four-divergence of an antisymmetric tensor, $\partial_\lambda K^{\lambda\mu\nu}$) to the canonical tensor. Because the divergence of an antisymmetric tensor is automatically zero ($\partial_\mu \partial_\lambda K^{\lambda\mu\nu} = 0$), adding this term does not change the global conservation laws.
Physically, this added term represents the intrinsic spin angular momentum of the electromagnetic field. The procedure absorbs the spin contribution into the orbital contribution, yielding a final tensor $T^{\mu\nu}$ that is perfectly symmetric ($T^{\mu\nu} = T^{\nu\mu}$) and explicitly gauge-invariant (containing only $F$, not $A$).
3. Field Momentum vs. Canonical Momentum
The physical field momentum density $\mathbf{p}_{\text{field}} = \mathbf{E} \times \mathbf{B}$ is fundamentally different from the canonical momentum density $\pi^\alpha$.
Canonical Momentum:$\pi^\alpha = -F^{0\alpha}$. Therefore, $\pi^0 = 0$ and the spatial canonical momentum is strictly the electric field, $\boldsymbol{\pi} = \mathbf{E}$.
Field Momentum:$\mathbf{p}_{\text{field}}$ is extracted from the $T^{0i}$ components of the symmetrized stress-energy tensor. It represents the actual spatial density of linear momentum carried by the propagating electromagnetic wave.
4. The Continuity Equation for All Components
Yes, $\partial_\mu T^{\mu\nu} = 0$ holds for all values of the free index $\nu$. Because it is a tensor equation, it represents four distinct continuity equations:
$\nu = 0$: $\partial_0 T^{00} + \partial_i T^{i0} = 0$. This is the continuity equation for energy (Poynting's Theorem).
$\nu = 1, 2, 3$: $\partial_0 T^{0j} + \partial_i T^{ij} = 0$. These are three continuity equations for the $x$, $y$, and $z$ components of linear momentum.
5. Evaluating the Spatial Components ($T^{ij}$)
Your expansion of the spatial components is exactly correct. $E^2$ and $B^2$ are the scalar products (sum of squares) of the 3-vectors:
$$E^2 = (E^1)^2 + (E^2)^2 + (E^3)^2$$
For the $T^{11}$ component, the Kronecker delta $\delta^{11} = 1$. The equation becomes:
6. Derivation of the Momentum Conservation Equation
Here is the exact mathematical expansion of the spatial components ($\nu = i$) of the conservation law in the presence of sources:
$$\partial_\mu T^{\mu i} = -F^{i\alpha}J_\alpha$$
Step A: Expand the Left-Hand Side
Sum over the index $\mu$ (splitting into time $\mu=0$ and space $\mu=j$):
$$\partial_0 T^{0i} + \partial_j T^{ji}$$
The first term is the time derivative of the field momentum density: $\partial_0 T^{0i} = \frac{\partial}{\partial t}(\mathbf{p}_{\text{field}})^i$.
The second term is the divergence of the spatial block of $T$. We define the classical Maxwell stress tensor as $\sigma^{ji} = -T^{ji}$. Therefore, $\partial_j T^{ji} = -\partial_j \sigma^{ji} = -(\nabla \cdot \sigma)^i$.
Step B: Expand the Right-Hand Side
The source term is $-F^{i\alpha}J_\alpha$. Remember that $J_\alpha = (\rho, -J^1, -J^2, -J^3)$. Expand over $\alpha$:
$$-F^{i0}J_0 - F^{ij}J_j$$
$F^{i0} = E^i$ and $J_0 = \rho$. The first term is $-\rho E^i$.
$F^{ij} = -\epsilon^{ijk}B_k$ (where $\epsilon^{ijk}$ is the Levi-Civita symbol) and $J_j = -J^j$. The second term evaluates exactly to the cross product: $-(-\epsilon^{ijk}B_k)(-J^j) = -\epsilon^{ijk}J^j B_k = -(\mathbf{J} \times \mathbf{B})^i$.
Combining these, the RHS is exactly the negative of the Lorentz force density $\mathbf{f}$:
This states that the mechanical force density $\mathbf{f}$ exerted on the charges, plus the rate of change of the field's momentum, is exactly balanced by the momentum flux (the divergence of the Maxwell stress tensor) flowing into that region of space.
Your proposed simplification, $\nabla \cdot \sigma = \mathbf{f}$, contains a fundamental physical error. It conflates the momentum of the field with the momentum of the particles.
Note that total momentum of the system consists of two distinct reservoirs: mechanical momentum and field momentum.
1. The Two Reservoirs of Momentum
In any region of space containing both matter and electromagnetic fields, the total momentum density $\mathbf{p}_{\text{total}}$ is the sum of two independent components:
Mechanical Momentum Density ($\mathbf{p}_{\text{mech}}$): The actual physical momentum of the particles with mass.
Field Momentum Density ($\mathbf{p}_{\text{field}}$): The momentum stored purely in the electromagnetic field itself, defined by the Poynting vector ($\mathbf{p}_{\text{field}} = \epsilon_0 \mathbf{E} \times \mathbf{B}$ in SI units, or $\mathbf{E} \times \mathbf{B}$ in natural units).
The Lorentz force density $\mathbf{f}$ is strictly the rate of change of the mechanical momentum of the charges:
The fundamental law of conservation of momentum states that the net flux of momentum entering a volume (given by the divergence of the Maxwell stress tensor, $\nabla \cdot \sigma$) must exactly equal the rate of change of the total momentum inside that volume.
If the equation were simply $\nabla \cdot \sigma = \mathbf{f}$, it would physically imply that electromagnetic fields cannot store or transport momentum independently of matter. This would forbid the existence of light carrying momentum through a vacuum (where $\mathbf{f} = 0$).
3. Does the Lorentz Force Come from the Poynting Vector?
The Lorentz force is the mechanism by which momentum is transferred between the field and the particles, but it does not exclusively come from the Poynting vector.
Consider a simple electrostatic system: a stationary point charge in a uniform, constant electric field.
Because the field is entirely static, there is no magnetic field ($\mathbf{B} = 0$).
Because $\mathbf{B} = 0$, the Poynting vector is zero ($\mathbf{p}_{\text{field}} = 0$). The field holds no momentum density.
However, the charge still experiences a continuous Lorentz force ($\mathbf{f} = \rho\mathbf{E}$).
How can the particle gain mechanical momentum if the field has no momentum ($\mathbf{p}_{\text{field}} = 0$) to give?
The momentum is supplied continuously by the spatial flux. The Maxwell stress tensor $\sigma$ is non-zero (composed of $-E^iE^j + \frac{1}{2}\delta^{ij}E^2$). The divergence of this stress tensor ($\nabla \cdot \sigma$) acts as a river of momentum flowing through the vacuum, which deposits that momentum directly into the particle via the Lorentz force.
In this purely static case, $\frac{\partial \mathbf{p}_{\text{field}}}{\partial t} = 0$, so the general equation cleanly reduces to:
$$\nabla \cdot \sigma = \mathbf{f}$$
However, in a dynamic system (like an electromagnetic wave hitting a solar sail), the incoming momentum flux ($\nabla \cdot \sigma$) is partitioned. Some of it pushes the sail ($\mathbf{f}$), and some of it changes the local momentum of the field itself ($\frac{\partial \mathbf{p}_{\text{field}}}{\partial t}$) as the wave reflects or absorbs. Both terms must be accounted for to conserve total momentum.
The Poynting vector is derived from both spatial and time translations, depending on whether you are evaluating it as momentum density or energy flux.
In classical field theory, applying Noether's theorem to the invariance of the Lagrangian under four-dimensional spacetime translations ($x^\mu \rightarrow x^\mu + a^\mu$) generates the complete stress-energy tensor $T^{\mu\nu}$. Because the translation vector $a^\mu$ has four components (one for time, three for space), the derivation splits into two distinct physical conservation laws, both of which contain the Poynting vector.
1. Time Translation (Energy Flux)
If you apply Noether's theorem exclusively for invariance under time translation ($x^0 \rightarrow x^0 + a^0$), you derive the conservation of energy.
The resulting conserved Noether current is the four-vector representing the flow of energy, $T^{0\mu} = (T^{00}, T^{0i})$.
The time component $T^{00}$ is the energy density.
The spatial components $T^{0i}$ represent the flux of energy across spatial boundaries.
In this derivation, the Poynting vector $\mathbf{S}$ emerges strictly as the energy flux.
2. Spatial Translation (Momentum Density)
If you apply Noether's theorem exclusively for invariance under spatial translations ($x^i \rightarrow x^i + a^i$), you derive the conservation of linear momentum.
The resulting conserved Noether currents form the spatial rows of the tensor, $T^{i\mu} = (T^{i0}, T^{ij})$.
The spatial components $T^{ij}$ represent the flux of momentum (the Maxwell stress tensor).
The time component $T^{i0}$ represents the density of momentum stored in the field itself ($\mathbf{p}_{\text{field}}$).
In this derivation, the Poynting vector $\mathbf{S}$ emerges strictly as the momentum density of the field.
3. The Symmetry Requirement
In the canonical derivation from the Lagrangian, the resulting stress-energy tensor is not initially symmetric ($T^{0i} \neq T^{i0}$). The canonical momentum density derived from spatial translation does not initially equal the canonical energy flux derived from time translation.
However, the physical stress-energy tensor must be symmetric to ensure the conservation of angular momentum. Once the Belinfante-Rosenfeld procedure is applied to symmetrize the tensor ($T^{\mu\nu} = T^{\nu\mu}$), it mathematically forces the energy flux to exactly equal the momentum density:
$$T^{0i} = T^{i0}$$
Therefore, applying spatial translation to the Lagrangian does directly yield the Poynting vector (as field momentum density $\mathbf{p}_{\text{field}}$), which is mathematically identical to the energy flux derived from time translation.
Note that the divergence of the stress tensor, $\nabla \cdot \sigma$, does not represent the stress on a single particle. It represents the net flux of electromagnetic momentum into an arbitrary volume of space.
Therefore, what experiences this stress is the total contents of that volume—both the matter and the electromagnetic field combined.
Here is the exact physical interpretation of how this momentum is distributed.
1. The Volume and the Flux
The Maxwell stress tensor $\sigma$ represents the flux of electromagnetic momentum. In continuum mechanics, stress is defined as force per unit area. The components $\sigma^{ij}$ describe the momentum per unit time (which has the dimensions of force) flowing across a spatial surface.
The divergence operator ($\nabla \cdot$) evaluates the net inward or outward flow through the closed boundary of an infinitesimal volume. Therefore, $\nabla \cdot \sigma$ is the net electromagnetic force exerted by the fields outside a given volume upon the entirety of the volume's interior.
2. The Partition of Momentum
Once this momentum flux enters the volume, the conservation equation dictates that it must be partitioned into two distinct physical outcomes. The equation $\nabla \cdot \sigma = \mathbf{f} + \frac{\partial \mathbf{p}_{\text{field}}}{\partial t}$ states:
$\mathbf{f}$ (Mechanical Force Density): If there are charges or currents inside the volume, a portion of the incoming momentum is transferred to them. This manifests as the Lorentz force, physically accelerating the particles. This is the only portion of the total stress that is experienced by matter.
$\frac{\partial \mathbf{p}_{\text{field}}}{\partial t}$ (Field Momentum Density Rate): Any incoming momentum that is not absorbed by matter must remain within the electromagnetic field itself. This locally changes the momentum density of the field (the Poynting vector) inside that specific volume.
If you evaluate a region of perfect vacuum (no charges, meaning $\mathbf{f} = 0$), that volume of space still experiences the stress flux. In this case, $\nabla \cdot \sigma = \frac{\partial \mathbf{p}_{\text{field}}}{\partial t}$, meaning the net flow of momentum into the vacuum strictly alters the local momentum of the passing electromagnetic waves.
