---
title: 'Flux-Balance Laws: Principles & Applications'
url: https://www.emergentmind.com/topics/flux-balance-laws
type: topic
---

# Flux-Balance Laws: Principles & Applications

Searching arXiv for recent and foundational papers on flux-balance laws across self-force, asymptotic gravity, PDEs, and constraint-based modeling.
Searching arXiv: "flux-balance laws self-force Kerr Carter constant"
Flux-balance laws are relations that connect the change of a density, charge, or orbital invariant to flux through the relevant boundary and, when present, to source or production terms. In one common local form they appear as $\partial_t u+\nabla\cdot F(u)=S(u,x,t)$, while in asymptotic gravity they relate variations of charges on cuts of $\mathscr I^+$ to radiative fluxes, and in self-force theory they relate the secular drift of Kerr invariants to radiation through future null infinity and the future event horizon [2109.08491, 2505.00077, 2209.13829]. A recurrent theme is that the same system may admit a true conservation law only in special symmetry or source-free regimes; otherwise one obtains a balance law.

## 1. Definition and structural principles

In continuum formulations, a balance law expresses the temporal change of a density plus the divergence of a flux as a source or production term. In Rational Extended Thermodynamics, a balance system is written as
$$
(F_i^\mu(s(x)))_{,\mu}=\Pi_i(s(x)),
$$
or equivalently
$$
F_i^0{}_{,t}+F_i^A{}_{,x_A}=\Pi_i,
$$
with $F_i^0$ the densities, $F_i^A$ the flux components, and $\Pi_i$ the production terms [1008.0211]. Within that framework, “supplementary balance laws” are additional laws $(K^\mu)_{,\mu}=Q$ satisfied by all solutions; they include the original balances, the entropy balance, and symmetry-induced balances, and they are characterized by the Lagrange–Liu relations
$$
d_y K^\mu=\lambda^i\,d_y F_i^\mu.
$$
The same formalism links entropy density, main fields, and production through
$$
\frac{\partial h^0}{\partial w^j}=\lambda^j,\qquad E=\lambda^i\Pi_i\ge 0
$$
[1008.0211].

This broad viewpoint already shows why the term “flux-balance law” is not restricted to one discipline. In some settings the law is local and differential, in others it is global and boundary-based, and in others it is an emergent law for averaged quantities. A common misconception is to treat “balance law” and “conservation law” as interchangeable. The papers considered here consistently distinguish them: sources, explicit coordinate dependence, radiation leakage through asymptotic boundaries, hyperstress terms, or broken scaling symmetry convert conservation into balance [1302.5561, 1008.0211].

## 2. Hyperbolic balance laws, entropy, and weak formulations

In PDE theory, balance laws are commonly studied in the form
$$
\partial_t u+\nabla_x\cdot f(t,x,u)=g(t,x,u),
$$
or, in one space dimension,
$$
u_t+\big(f(u)\big)_x=\alpha(t)\,u
$$
on the quarter-plane [2503.20370, 2206.04986]. For the latter problem, introducing
$$
\beta(t):=\int_0^t \alpha(\theta)\,d\theta,\qquad w:=e^{-\beta(t)}u
$$
transforms the equation into a conservation law for $w$, and the solution admits a Lax–Oleinik type representation
$$
W(x,t)=\min\{A(x,t),B(x,t)\},
$$
with admissible sets $H_A$ and $H_B$ controlling whether minimizing curves reach the initial line or the boundary $x=0$ [2206.04986]. The boundary analysis is nontrivial because minimizing curves may spend time on the boundary, and the paper classifies boundary points into Types I–III in order to verify the Bardos–le Roux–Nédélec condition.

For general scalar balance laws, the entropy-production viewpoint packages all entropy inequalities into a single operator. Given a bounded function $u$, the entropy production $\mathcal M_u$ acts on an entropy $E$; for distributional solutions and $E\in C^2(\mathbb R)$,
$$
\mathcal M_u(E)=\frac12\int_{J_u} E''(k)\,\mu_k\,dk,
$$
where $\mu_k$ are the Kružkov entropy productions [2503.20370]. The same work proves that positivity of all $\mu_k$ is equivalent to positivity of $\mathcal M_u(E)$ for every convex entropy, and extends the framework to space- and time-dependent or complex-valued entropies. This places entropy admissibility, weak solvability, and Fourier representations of entropy production in a single operator-theoretic framework.

The integral balance interpretation of weak solutions is also sharpened by flux regularity results. For nonlinear hyperbolic conservation laws, the time-integrated boundary flux is shown to be Lipschitz continuous with respect to suitable perturbations of the boundary, so that a weak solution indeed satisfies the integral balance law on smooth bounded domains [2006.09253]. In one-dimensional strictly hyperbolic systems with source, vanishing-viscosity solutions are SBV for all but at most countably many times under genuine nonlinearity, and this extends to the broader non-degeneracy condition; when linear degeneracy is present, full SBV regularity fails but an SBV-like statement survives at the level of the eigenvalue functions [2409.06087]. Stability theory also extends beyond local fluxes: for systems with spatially nonlocal interactions and temporal memory, entropy solutions satisfy quantitative $L^1$ stability estimates with respect to perturbations in fluxes, kernels, sources, and initial data [2605.01709].

## 3. Self-force theory and Kerr flux-balance laws

In scalar self-force theory on Kerr, a radiating point particle of scalar charge $q$ and dynamical mass $m(\tau)$ moves on a worldline $z^\mu(\tau)$, and the field satisfies
$$
\Box\Phi=-4\pi \rho,\qquad \rho(x)=q\int d\tau\,\frac{\delta^{(4)}(x-z(\tau))}{\sqrt{-g}}.
$$
Using the Detweiler–Whiting decomposition $\Phi^+=\Phi^R+\Phi^S$, the first-order scalar self-force is
$$
f^\mu=q(g^{\mu\nu}+u^\mu u^\nu)\nabla_\nu\Phi^R,
$$
and the instantaneous drift of the symmetry-generated integrals obeys
$$
\frac{dE}{d\tau}=-q\,\partial_t\Phi^R,\qquad \frac{dL_z}{d\tau}=-q\,\partial_\phi\Phi^R,
$$
while for the Carter constant
$$
\frac{dQ}{d\tau}=-2q\,\mathcal L_{K\cdot p}\Phi^R
$$
[2209.13829]. The central difficulty is that the Carter constant is not directly associated with a Noether current built from a spacetime isometry.

The resolution proceeds through symplectic currents, symmetry operators, and action-angle variables. The scalar-field symplectic current is
$$
\omega_{abc}(\Phi_1,\Phi_2)=\epsilon_{dabc}\left[\Phi_2\nabla^d\Phi_1-\Phi_1\nabla^d\Phi_2\right],
$$
and bound Kerr geodesics admit action-angle coordinates $(q^\alpha,J_\alpha)$ with $\Omega=dJ_\alpha\wedge dq^\alpha$ and $dq^\alpha/d\tau=\nu_\alpha(J)$ [2209.13829]. This yields exact integrated balance laws for the actions and, after orbit averaging and replacing $\Phi^R$ by the globally defined radiative field $\Phi^{\mathrm{rad}}$, explicit asymptotic mode sums at $\mathscr I^+$ and $\mathscr H^+$. With $\Lambda=(l,m,k,n)$ and
$$
\omega_{mkn}=m\Omega_\phi+k\Omega_\theta+n\Omega_r,
$$
the averaged law for the Carter constant becomes
$$
\left\langle\frac{dQ}{d\tau}\right\rangle=\frac{\partial Q}{\partial J_\alpha}\left\langle\frac{dJ_\alpha}{d\tau}\right\rangle,
$$
which supplies a flux-balance law for $Q$ even though no direct stress-energy Noether current reproduces $\Delta Q$ [2209.13829].

The gravitational spinning-body extension uses the regular metric perturbation and the gravitational symplectic current instead of a scalar field. In the pole–dipole approximation with the Tulczyjew–Dixon spin supplementary condition, the paper derives a quasi-local averaged law
$$
\langle\dot\Gamma^A\rangle=\Pi^{AB}\!\left[F_B+\nabla_B\langle H\rangle\right],
$$
and, off resonance, an invariant flux law
$$
\langle\dot P_\alpha\rangle=-A^\beta{}_\alpha(\partial_{\theta^\beta})^A F_A
$$
for the four geodesic-like invariants $\{m^2,E,L_z,K\}$ [2406.10343]. Here $K$ is the generalized Carter constant including the Killing–Yano correction. The fifth invariant, the Rüdiger constant, remains coupled to constraint variables rather than yielding a practical flux law. This suggests that, at linear order in spin, the generalized Carter constant is accessible through asymptotic radiation, whereas the full five-invariant system is not yet closed in a practical boundary-flux form.

## 4. Asymptotic gravitational charges, memory, and extensions beyond flat space

At future null infinity of asymptotically flat spacetimes, flux-balance laws govern the Bondi mass aspect $m$, the angular momentum aspect $N_A$, and the higher BMS charges built from the shear $C_{AB}$ and news $N_{AB}=\partial_u C_{AB}$. In a Carrollian-holographic derivation, local Carroll, Weyl, and diffeomorphism invariance at $\mathscr I^+$ imply the Bondi evolution equations
$$
\partial_u m=-\frac18 N_{AB}N^{AB}+\frac14 D_A D_B N^{AB},
$$
together with the corresponding angular-momentum balance equation, and the cut charge
$$
Q_\Sigma(f,Y)=\int_\Sigma \left(4 f M+2Y^A N_A+16\,Y^A D_A(C_{BC}C^{BC})\right)d^2\Omega
$$
obeys a flux law sourced by the news [2505.00077]. This is a boundary-intrinsic derivation of the same balance structure usually obtained from bulk Einstein equations.

Within the standard Bondi framework, the Poincaré and proper BMS flux-balance laws distinguish supermomentum, super-angular momentum, and super-center-of-mass charges, and they encode the displacement, spin, and center-of-mass memory effects defined from the shear [1912.03164]. Their multipolar realization is exact in terms of radiative symmetric tracefree moments. In that formulation, fluxes of energy, angular momentum, and octupole super-angular momentum arise at $2.5\mathrm{PN}$, fluxes of quadrupole supermomentum arise at $3\mathrm{PN}$, and fluxes of momentum, center-of-mass, and octupole super-center-of-mass arise at $3.5\mathrm{PN}$ [1912.03164]. The same work also scrutinizes the prescriptions for angular momentum and center-of-mass and singles out the covariant choice that aligns with the symplectic framework.

The covariant phase-space and Wald–Zoupas formalism extends these laws beyond general relativity. In diffeomorphism-invariant extensions such as luminal Horndeski theory, charges on cuts $\mathcal S\subset\mathscr I^+$ satisfy
$$
\mathcal Q_\xi[\mathcal S_2]-\mathcal Q_\xi[\mathcal S_1]=\int_{\Delta\mathscr I^+}\boldsymbol F_\xi,
$$
and the flux receives both the general-relativistic news contribution and a scalar contribution proportional to $(L_\xi\varphi_1+K\varphi_1)\dot\varphi_1$ [2601.07091]. In Brans–Dicke theory this reduces to an explicit scalar-flux correction to the supertranslation law. A related cosmological deformation occurs in de Sitter spacetime: at quadrupolar order the $SO(1,4)$ flux-balance laws govern dilatations, rotations, spatial translations, and cosmological boosts, and the flat limit recovers the standard Poincaré laws at future null infinity [2411.16215]. The de Sitter dilatation flux can be written in two distinct negative-definite forms, making precise how curvature deforms the notion of radiative energy.

## 5. Constraint-based metabolic formulations

In systems biology, the phrase “flux-balance” refers to stoichiometric mass balance under metabolic constraints. Standard flux balance analysis imposes
$$
S\,v=0,\qquad \ell_i\le v_i\le u_i,
$$
and typically maximizes the biomass-reaction flux $\lambda=v_{\text{biomass}}$ [1607.00128]. Constrained Allocation Flux Balance Analysis adds a proteome-wide constraint derived from growth-law data,
$$
w_C\,v_C+\sum_i w_i\,|v_i|+w_R\,\lambda=\phi_{\max},
$$
where the coefficients encode catabolic, enzymatic, and ribosomal proteome costs [1607.00128]. In this usage, flux-balance laws are not asymptotic charge laws but steady-state mass-balance constraints augmented by resource-allocation laws. The model predicts a respiratory-to-fermentative crossover and quantitatively relates growth rate, uptake, secretion, and yield.

Dynamic enzyme-cost Flux Balance Analysis extends this setting by adding slow-time balances for external species and macromolecules while maintaining quasi-steady internal metabolism:
$$
\dot Y=S_y^y V_y,\qquad \dot{\tilde P}=S_p^p\tilde V_p,\qquad
S_y^x V_y+S_x^x V_x+\alpha^{-1}S_p^x\tilde V_p=0.
$$
Enzyme capacities appear as linear constraints
$$
\tilde H_c\tilde V-H_E\tilde P\le 0,
$$
and robust deFBA handles uncertainty through a scenario tree with non-anticipativity in the first interval and a robust horizon $r=1$ [1609.08961]. This suggests a distinct but mathematically recognizable interpretation of flux-balance laws: balance of stoichiometric fluxes, enzyme capacities, and biomass synthesis under uncertainty rather than boundary-flux accounting.

## 6. Continuum-mechanical and numerical realizations

In micromorphic elasticity, translational, rotational, and dilatational symmetries generate fluxes whose divergences are the configurational balance laws of the medium. The translational flux is the Eshelby stress tensor
$$
P_{ki}=W\,\delta_{ki}-u_{a,k}\,t_{ai}-\varphi_{\alpha\beta,k}\,m_{\alpha\beta i},
$$
the rotational flux is the angular momentum tensor $M_{ki}$, and the scaling flux is the dilatational vector $Y_i$ [1302.5561]. Their surface integrals define the $J$-, $L$-, and $M$-integrals. In homogeneous isotropic media without body forces or couples, the $J$- and $L$-integrals become conservation laws, whereas the dilatational law is generally broken by the hyperstress term $K_{\alpha\beta i}m_{\alpha\beta i}$ [1302.5561]. This is an explicit example of the general principle that internal length scales or constitutive structure convert a conservation law into a balance law.

Numerical analysis recasts the same principle at the discrete level. For multidimensional linear hyperbolic balance laws, the Global Flux formulation rewrites the spatial operator through one-dimensional and mixed derivatives of global primitives, for example
$$
\partial_x u+\partial_y v-S_p=\partial_{xy}(U+V-K_p),
$$
and the resulting finite-element schemes are stationarity preserving because the stabilizations act on the same global-flux combinations that define the steady kernel [2510.02928]. With Gauss–Lobatto nodes, the line and row integrators are Lobatto IIIA tables and yield nodal steady-state errors $O(h^{K+2})$ for $K\ge 2$ [2510.02928]. Residual distribution schemes implement the same balance at the element level through
$$
\Phi^K(u^h)=\int_K\left(\nabla\cdot F^h-S^h\right)\,dx
$$
or through boundary numerical fluxes, and the framework encompasses SUPG, DG, finite-volume, and well-balanced shallow-water discretizations [2109.08491]. A plausible implication is that, in numerical settings as in analytic ones, a flux-balance law is most effective when the discrete operator, the source treatment, and the steady-state kernel are made exactly compatible.

Source: https://www.emergentmind.com/topics/flux-balance-laws