---
title: Dynamic Virial Theorem
url: https://www.emergentmind.com/topics/dynamic-virial-theorem
type: topic
---

# Dynamic Virial Theorem

The dynamic virial theorem is the time-dependent identity that relates the second derivative of a system’s moment of inertia to its kinetic energy and the virial of the applied forces. In its standard particle form, for \(I(t)=\sum_{i=1}^N m_i r_i^2\), it reads
\[
\frac{1}{2}\frac{d^2 I}{dt^2}=2K+\sum_{i=1}^N \mathbf r_i\cdot \mathbf F_i,
\]
and its steady or long-time-averaged limit yields the familiar virial balance when the inertial term vanishes. Contemporary treatments place this identity in a broader setting: homogeneous power-law interactions, general steady states beyond canonical and microcanonical ensembles, geometric mechanics on manifolds and Lie algebroids, dissipative and open quantum systems, nonequilibrium quantum gases governed by Tan’s contact, continuum and micropolar media, and modified gravitational theories [2507.22624][1410.2032][2205.11731].

## 1. Classical identity and steady-state limit

For an \(N\)-particle system with masses \(m_i\), positions \(\mathbf r_i(t)\), momenta \(\mathbf p_i=m_i\mathbf v_i\), and forces \(\mathbf F_i\), the standard constructions are the scalar moment of inertia
\[
I(t)=\sum_{i=1}^N m_i r_i^2
\]
and the Clausius virial
\[
G(t)=\sum_{i=1}^N \mathbf r_i\cdot \mathbf p_i.
\]
They satisfy \(dI/dt=2G\), hence
\[
\frac{1}{2}\frac{d^2 I}{dt^2}=\frac{dG}{dt}=2T+\sum_{i=1}^N \mathbf r_i\cdot \mathbf F_i,
\]
with \(T=\sum_i p_i^2/(2m_i)\) the instantaneous kinetic energy. For conservative forces \(\mathbf F_i=-\nabla_i V\), the potential form becomes
\[
\frac{1}{2}\frac{d^2 I}{dt^2}=2T-\sum_{i=1}^N \mathbf r_i\cdot \nabla_i V.
\]
The time-averaged theorem follows when \(G\) remains bounded, or more generally when the long-time average of the total derivative vanishes; then
\[
2\langle T\rangle=-\Big\langle\sum_{i=1}^N \mathbf r_i\cdot \mathbf F_i\Big\rangle.
\]
These statements appear in particle, continuum, and stochastic formulations with the same kinematic core [2209.06856].

For homogeneous potentials, Euler-type scaling converts the force virial into a potential-energy relation. If \(V(\lambda q)=\lambda^k V(q)\), then \(\sum_i q^i\partial_i V = kV\), so the averaged theorem becomes
\[
2\langle T\rangle=k\langle V\rangle.
\]
For inverse-power attraction with \(k=-1\), one recovers \(2\langle T\rangle=-\langle V\rangle\), consistent with \(T>0\) and \(V<0\). A persistent source of confusion is the identification of the virial theorem with this averaged relation alone. The dynamic virial theorem is the instantaneous identity; the steady-state form requires the additional condition \(\langle d^2I/dt^2\rangle=0\), and deviations from virial balance during collapse, expansion, or other transients are measured precisely by the inertial term \(\langle \tfrac12 d^2I/dt^2\rangle\) [2507.22624].

## 2. Homogeneous power-law interactions, density of states, and steady states beyond Gibbs ensembles

For pairwise power-law interactions,
\[
\Phi(\mathbf r_1,\ldots,\mathbf r_N)=\frac12\sum_{i=1}^N\sum_{j\neq i}\varphi(|\mathbf r_j-\mathbf r_i|),\qquad \varphi(r)=\varphi_0 r^\gamma,
\]
homogeneity implies
\[
\mathcal W\equiv \sum_{i=1}^N \mathbf r_i\cdot \frac{\partial \Phi}{\partial \mathbf r_i}=\gamma\,\Phi.
\]
Since \(\mathbf F_i=-\partial \Phi/\partial \mathbf r_i\), one has
\[
\sum_{i=1}^N \mathbf r_i\cdot \mathbf F_i=-\gamma\,\Phi,
\]
and therefore in the steady-state limit
\[
2\langle K\rangle=\gamma\langle \Phi\rangle.
\]
With the notation \(s=\gamma\), this is also written \(2\langle K\rangle=s\langle V\rangle\). The relation holds in any spatial dimension \(d\) provided the interaction is a pure power and sufficiently regular for the derivatives to exist.

A notable development is the exact computation of the configurational density of states
\[
D(\phi)=\int d\mathbf r_1\cdots d\mathbf r_N\;\delta\!\big(\Phi(\mathbf r_1,\ldots,\mathbf r_N)-\phi\big)
\]
for homogeneous pair potentials. In three dimensions, using Rugh’s geometrical framework with the vector field \(\omega=(\mathbf r_1,\ldots,\mathbf r_N)\), one obtains
\[
D(\phi)=D_0\,\phi^{\frac{3N}{\gamma}-1}.
\]
Combined with the quadratic kinetic density of states
\[
\Omega_K(k)=W_N\,k^{\frac{3N}{2}-1},
\]
the total density of states follows by convolution,
\[
\Omega(E)=W_ND_0\,B\!\Big(\frac{3N}{2},\frac{3N}{\gamma}\Big)\,E^{\frac{3N}{2}+\frac{3N}{\gamma}-1},
\]
which implies
\[
\frac{1}{k_B T(E)}=\frac{1}{E}\left(\frac{3N}{2}+\frac{3N}{\gamma}-1\right),\qquad
C_E=\left(\frac{3N}{2}+\frac{3N}{\gamma}-1\right)k_B.
\]
Thus the microcanonical heat capacity is constant for this class of systems. For \(\gamma=-1\), the formula gives
\[
C_E=-\left(\frac{3N}{2}+1\right)k_B<0,
\]
while canonical stability requires \(C_E>0\), which is ensured for any positive \(\gamma\) and for negative \(\gamma\) satisfying \(\gamma<-2\).

The same paper shows that the virial theorem can be recovered from the configurational density of states alone, without assuming canonical or microcanonical Gibbs measures. If the steady state has the form
\[
P(\mathbf R,\mathbf P|S)=\rho\!\big(K(\mathbf P)+\Phi(\mathbf R);S\big),
\]
then
\[
P(k,\phi|S)=\rho(k+\phi;S)\,\Omega_K(k;N)\,D(\phi),
\]
and the conjugate variables theorem yields, after choosing \(\omega(K,\Phi)=K\Phi\),
\[
\langle K\rangle_S=\frac{\gamma}{2}\langle \Phi\rangle_S.
\]
This extends the steady-state virial theorem to any stationary distribution that depends on phase-space variables through the conserved energy alone. By contrast, for non-homogeneous interactions such as Lennard–Jones \(v(r)=a r^{-12}-b r^{-6}\), no single degree \(s\) exists; the correct relation is
\[
2\langle K\rangle=s_{12}\langle V_{12}\rangle+s_6\langle V_6\rangle,
\]
not \(2\langle K\rangle=s\langle V\rangle\) with one exponent [2507.22624].

## 3. Geometric formulations on manifolds, in quasi-coordinates, and on Lie algebroids

In geometric mechanics, the dynamic virial theorem is formulated intrinsically for mechanical Lagrangians on a Riemannian configuration manifold \(Q\). With metric \(g\), kinetic energy \(T(q,v)=\tfrac12 g_q(v,v)\), and Lagrangian \(L=T-V\), a vector field \(X\in\mathfrak X(Q)\) defines the linear virial function
\[
G_X(q,v)=g_q(v,X(q)).
\]
Along a solution \(q(t)\) of \(\nabla_t\dot q=-\mathrm{grad}\,V\), differentiation gives
\[
\frac{d}{dt}G_X(q(t),\dot q(t))
=\frac12(\mathcal L_X g)_{q(t)}(\dot q(t),\dot q(t))-X(V)(q(t)).
\]
Under periodicity or boundedness hypotheses ensuring that the average of a total derivative vanishes, the averaged virial relation becomes
\[
\Big\langle \frac12(\mathcal L_X g)(\dot q,\dot q)\Big\rangle=\langle X(V)\rangle.
\]
Special cases follow from the symmetry class of \(X\): for Killing fields \(\mathcal L_X g=0\), one gets \(\langle X(V)\rangle=0\); for homothetic fields \(\mathcal L_X g=2\sigma g\) with constant \(\sigma\), \(\langle 2\sigma T\rangle=\langle X(V)\rangle\); for conformal Killing fields \(\mathcal L_X g=2\sigma(q)g\), the kinetic term is weighted locally by \(\sigma(q)\). The Euclidean dilation field \(X=\sum_i q^i\partial_{q^i}\) recovers
\[
\frac{d}{dt}(q\cdot \dot q)=2T-q\cdot \nabla V,
\]
and hence \(2\langle T\rangle=k\langle V\rangle\) for \(k\)-homogeneous potentials [1410.2032].

A complementary formulation uses quasi-coordinates and Lie algebroids. In quasi-velocities \(v^a\) associated with a moving frame \(\{e_a\}\), Hamel’s coefficients \(C^c_{ab}\) encode nonholonomy through \([e_a,e_b]=C^c_{ab}e_c\). For Hamiltonian systems in quasi-coordinates, the general theorem is
\[
\langle \dot G\rangle=\langle \{G,H\}\rangle=0
\]
for any virial function \(G\) bounded along the motion. Fiberwise-linear virial functions \(G(q,p)=p_a f^a(q)\) are associated with complete lifts of vector fields \(D=f^a e_a\), and the virial identity becomes \(\langle D^c(H)\rangle=0\). In the Lagrangian picture, if \(G=\langle \theta_L,D^c\rangle\), then \(\Gamma_L(G)=D^c(L)\) and therefore
\[
\langle D^c(L)\rangle=0.
\]
The same structure extends to Lie algebroids \(A\to M\), where the virial theorem is written either as \(\langle \Gamma_L(G)\rangle=0\) on \(A\) or \(\langle \{G,H\}_{A^*}\rangle=0\) on \(A^*\). This places the theorem in a setting adapted to systems with symmetry, moving frames, and nontrivial anchor maps [1405.6532].

## 4. Dissipative, stochastic, and open-system formulations

Dissipation modifies the dynamic virial balance by adding explicit force-virial terms from friction and noise. For the damped oscillator
\[
m\ddot x+\mu \dot x+kx=0,\qquad \mu=m\gamma,
\]
the instantaneous relation is
\[
\frac{d}{dt}(xp)=2T-xV'(x)-\gamma xp.
\]
Its long-time average gives
\[
\Big\langle \frac{p^2}{m}\Big\rangle_t=\langle m\omega_0^2x^2\rangle_t+\gamma\langle xp\rangle_t,
\]
so the averaged kinetic and potential energies are generally unequal. For the Brownian oscillator
\[
m\ddot x+\mu\dot x+m\omega_0^2 x=F(t),
\]
the dynamical identity becomes
\[
\frac{d}{dt}\langle xp\rangle
=2\langle T\rangle-\langle xV'(x)\rangle-\gamma\langle xp\rangle+\langle xF(t)\rangle.
\]
In the stationary state, under stationarity and ergodicity, \(\langle xF(t)\rangle=0\) and \(\langle x\dot x\rangle=0\), so equipartition is restored:
\[
\Big\langle \frac{m\dot x^2}{2}\Big\rangle
=\Big\langle \frac{m\omega_0^2x^2}{2}\Big\rangle
=\frac{\Gamma}{4\mu},
\]
with \(K_{\rm th}=V_{\rm th}=k_B T/2\) when \(\Gamma=2\mu k_B T\) [2302.12008].

For a dissipative quantum oscillator coupled to a heat bath in the Caldeira–Leggett form, the virial operator is symmetrized,
\[
G=\frac{xp+px}{2},
\]
and the Heisenberg equation produces bath-induced terms \(I_1\) and \(I_2\):
\[
\frac{d\langle G\rangle}{dt}
=\frac{1}{2i\hbar}\left[
\Big\langle \frac{p^2}{m}\Big\rangle-\langle xV'(x)\rangle+I_1-I_2
\right].
\]
In the stationary state,
\[
\Big\langle \frac{p^2}{m}\Big\rangle=\langle xV'(x)\rangle+I_2-I_1.
\]
For the harmonic oscillator this gives
\[
\langle K\rangle-\langle V\rangle=\frac{I_2-I_1}{2},
\]
so non-Markovian memory and colored quantum noise break equipartition in general. The classical limit \(\hbar\to 0\) and the weak-coupling limit \(\gamma\to 0\) both suppress \(I_1\) and \(I_2\), recovering the standard relation. The same structure has an electrical analogue in noisy RLC circuits, where the virial balance is written in terms of charge \(Q\) and current \(\dot Q\) [2302.12008].

A different extension treats non-differentiable paths in resolution-scale relativity. There the complex velocity is \(V=v-iU\), the scale-covariant derivative is
\[
\hat d\, h=\partial_t h+V\cdot \nabla h-iD\Delta h,
\]
and the virial theorem becomes
\[
2(T_{\rm Tot})=2(T_{\rm class})+2(Q_{\rm Tot})=
\Big(\sum_{k=1}^N \mathbf r_k\cdot \mathbf F_k\Big),
\]
where the quantum-like potential is
\[
Q=-2mD^2\frac{\Delta \sqrt{\rho}}{\sqrt{\rho}}.
\]
Under the identification \(\hbar=2mD\), this reproduces the quantum mechanical virial theorem while retaining a time-average formulation rather than expectation values as the primary averaging device [2209.06856].

## 5. Nonequilibrium quantum gases and Tan’s contact

For short-range interacting quantum gases in three dimensions, the dynamic virial theorem acquires an exact contact term. With particle mass \(m\), scattering length \(a(t)\), harmonic confinement \(V_{ho}=\tfrac12 m\omega(t)^2r^2\), total energy \(E(t)\), trap energy \(E_{ho}(t)\), and
\[
I(t)=mN\langle r^2(t)\rangle,
\]
the theorem reads
\[
E(t)-2E_{ho}(t)=\frac14\frac{d^2I(t)}{dt^2}
-\frac{\hbar^2}{8\pi m\,a(t)}\,C(t),
\]
or equivalently
\[
\frac{d^2}{dt^2}\langle r^2(t)\rangle
=\frac{4}{m}\big[E(t)-2E_{ho}(t)\big]
+\frac{\hbar^2}{2\pi m^2 a(t)}\,C(t),
\]
where \(C(t)\) is Tan’s contact defined by the large-momentum tail \(n(\mathbf k,t)\sim C(t)/k^4\). At unitarity, \(1/a=0\), the contact term drops out, and the cloud-size dynamics becomes identical to that of an ideal gas. The derivation uses the dilation operator \(D=\tfrac12\sum_j (\mathbf r_j\cdot \mathbf p_j+\mathbf p_j\cdot \mathbf r_j)\), the commutator \([D,V_{\rm int}]\), and Tan’s adiabatic relation
\[
\frac{\partial E}{\partial(1/a)}\Big|_{S,N}
=-\frac{\hbar^2}{4\pi m}\,C.
\]
The contact term is therefore the dynamical imprint of short-distance correlations in the virial balance [2205.11731].

This nonequilibrium theorem provides an experimentally accessible formulation of the maximum energy growth theorem. For an initially noninteracting gas with a ramp \(a(t)\propto t^\alpha\), the short-time energy growth obeys
\[
\delta E(t)\propto
\begin{cases}
t^\alpha,& \alpha>1/2,\\
t^{1/2},& \alpha=1/2,\\
t^{1-\alpha},& 0<\alpha<1/2,
\end{cases}
\]
so the square-root ramp \(a(t)\propto \sqrt t\) maximizes the initial growth. Because
\[
E(t)=\frac14\frac{d^2I}{dt^2}
-\frac{\hbar^2}{8\pi m a(t)}\,C(t)
\]
during free expansion, the theorem converts measurements of cloud size and contact into a direct test of the energy-growth bound.

In two-fluid hydrodynamics, the same identity combines with continuity and momentum balance to yield an out-of-equilibrium pressure relation,
\[
\mathcal J(t)=\frac23 E_{\rm internal}(t)
+\frac{\hbar^2}{12\pi m\,a(t)}\,C(t)
+\Gamma(t),
\]
where \(\Gamma(t)\) encodes bulk dissipation. In equilibrium, with no flow and \(\Gamma=0\), this reduces to Tan’s pressure relation. The result is the nonequilibrium analogue of the equilibrium pressure/contact formula, expressed in terms of measurable dynamical quantities [2205.11731].

## 6. Continuum, field-theoretic, and micropolar generalizations

In continuum mechanics, the theorem is obtained by testing momentum balance against the generator of dilatations. For a material region \(\Omega\subset \mathbb R^3\) with density \(\rho(x,t)\), velocity \(v(x,t)\), Cauchy stress \(\sigma_{ij}\), and body-force density \(f_i\), define
\[
I(t)=\int_\Omega \rho\, r^2\, d^3x,\qquad
K=\int_\Omega \frac12 \rho v^2\, d^3x.
\]
Using Reynolds transport and Cauchy’s equation,
\[
\rho \frac{dv_i}{dt}=\partial_j \sigma_{ij}+f_i,
\]
one finds
\[
\frac12\frac{d^2I}{dt^2}
=2K+\int_\Omega x_i f_i\,d^3x
+\oint_{\partial\Omega} x_i\sigma_{ij}n_j\,dS
-\int_\Omega \mathrm{tr}\,\sigma\, d^3x.
\]
For fluids with \(\sigma=-PI\), the stress contribution becomes \(+3\int_\Omega P\,d^3x\). When boundary terms vanish by decay, compact support, or periodicity, the time-averaged continuum theorem reduces to the familiar bulk balance between kinetic, forcing, and trace-of-stress terms [1504.04118].

The same scaling structure appears in field theory through the dilatation Noether current
\[
J_D^\mu=x_\nu T^{\mu\nu}+V^\mu,\qquad
\partial_\mu J_D^\mu=T^\mu{}_\mu+\partial_\mu V^\mu.
\]
In four-dimensional Maxwell theory, the improved energy-momentum tensor is traceless,
\[
T^\mu{}_\mu=0,
\]
which expresses conformal invariance. The continuum virial identity is therefore the nonrelativistic projection of a dilatational balance law. One formulation further identifies the Clausius, Cosserat, Maxwell, and Weyl equations as formal adjoints of the Spencer operator for the conformal group, with the trace term arising from testing against the dilatation generator \(x_i\partial_i\) [1504.04118].

Micropolar media introduce an additional rotational sector. With velocity \(v_i\), microrotation \(\omega_i\), microinertia \(J_{ij}=J_0\delta_{ij}\), force stress \(T_{ji}\), and couple-stress \(M_{ji}\), the translational virial theorem becomes
\[
\Bigg\{
\int_B \rho x_i f_i\,dV
+\int_{\partial B} x_i t_i^{(n)}\,dS
-\int_B T_{ii}\,dV
\Bigg\}_\infty
=-2\{K_u(B)\}_\infty,
\]
with \(K_u(B)=\tfrac12\int_B \rho v_i v_i\,dV\). The rotational theorem is
\[
\Bigg\{
\int_B \rho \omega_i g_i\,dV
+\int_{\partial B}\omega_i m_i^{(n)}\,dS
-\int_B \omega_{i,k}M_{ki}\,dV
+\int_B \omega_i T_i\,dV
\Bigg\}_\infty
=-2\{K_\omega(B)\}_\infty,
\]
with \(K_\omega(B)=\tfrac12\int_B \rho J_0\omega_i\omega_i\,dV\). These relations uncover the virial force-stress
\[
\tau_{ji}=T_{ji}-\rho v_jv_i
\]
and the virial couple-stress
\[
\widetilde M_{ji}=M_{ji}-\rho J_0\omega_j v_i.
\]
The first contains the Reynolds stress through \(-\rho v_j'v_i'\); the second contains the turbulent couple-stress through \(-\rho J_0\omega_j'v_i'\). In the classical Cauchy limit, where microrotation and couple-stress vanish, the rotational theorem collapses and the translational theorem reduces to the standard continuum virial relation [2112.09063].

## 7. Gravitational and astrophysical variants

In Newtonian gravity, the virial theorem is central to the dynamics of self-gravitating systems. For discrete masses,
\[
W_N=-\frac12\sum_{i,j=1;i\neq j}^N \frac{Gm_im_j}{|\mathbf x_i-\mathbf x_j|},
\]
and the time-dependent identity takes the form
\[
\frac12\frac{d^2I}{dt^2}=2T+W_N
\]
or equivalently \(d^2I/dt^2=2T+W_N\) in the notation used for the gravitational pair potential. For collisionless stellar systems, the continuum form is
\[
I(t)=\int \rho r^2\,d^3r,\qquad
W(t)=-\int \rho\,\mathbf r\cdot \nabla\Phi\,d^3r,
\]
and for self-gravity \(W=U\), so
\[
\frac{d^2I}{dt^2}=2T+U.
\]
In spherical collisionless systems with negative total energy, this equation governs undamped finite-amplitude breathing oscillations about virial equilibrium. One treatment maps the breathing mode to an Ermakov–Lewis–Leach structure, where \(x\propto \sqrt{I}\) satisfies a parametric-oscillator equation and the invariant
\[
I_{EL}=\frac12\Big[(\rho \dot x-\dot \rho\,x)^2+(x/\rho)^2\Big]
\]
generates an infinite asymptotic hierarchy \(I(t;\epsilon)=I_0+\epsilon I_1+\epsilon^2 I_2+\cdots\) with \(dI_k/dt=0\) order by order. In that framework, the constants of motion depend on the virialised mass and radius through the universal scalings \(U_0\propto -GM_v^2/R_v\) and \(\Omega_0^2\propto GM_v/R_v^3\), rather than on the detailed choice of potential profile [2306.04435].

Modified gravity changes the virial balance itself. In nonlocal Newtonian gravity, the pair force becomes
\[
F(r)=-Gmm' r^{-2}[1+\mathcal N(r)],
\]
with
\[
\mathcal N(r)=\alpha_0[1-(1+\mu_0 r)e^{-\mu_0 r}]-\epsilon(r),
\]
and the exact virial identity for an isolated \(N\)-body system is
\[
\frac{d^2I}{dt^2}=2T+W_N+D,
\]
where
\[
D=-\frac12\sum_{i,j;i\neq j}\frac{Gm_im_j\,\mathcal N(r_{ij})}{r_{ij}}.
\]
In virial equilibrium,
\[
2\langle T\rangle=-\langle W_N\rangle-\langle D\rangle.
\]
Interpreting the correction as an effective dark component leads to \(\langle D\rangle=f_{DM}\langle W_N\rangle\). For sufficiently isolated nearby galaxies in virial equilibrium at the present epoch, the theory predicts
\[
D_B\ge f_{DM}\lambda_0,\qquad \lambda_0=\frac{2}{\alpha_0\mu_0}\approx 3\pm 2\ \mathrm{kpc},
\]
where \(D_B\) is the baryonic diameter. In this setting the dynamic virial theorem is not merely diagnostic; it is the mechanism by which nonlocality is mapped onto an effective dark-matter fraction [1512.01193].

A further model-specific complication concerns the meaning of stationarity. In a relativistic uniform sphere model with gravitational, electromagnetic, acceleration, and pressure fields, one analysis argues that the partial time derivative of the virial may vanish while the material derivative does not:
\[
\frac{DG}{Dt}=\frac{\partial G}{\partial t}+v\cdot \nabla G,\qquad
\frac{\partial G}{\partial t}\approx 0,\quad v\cdot \nabla G\neq 0.
\]
Within that framework the virial balance becomes
\[
2W_k=\frac{dG}{dt}-(W_g+W_{em}+W_{acc}+W_P),
\]
and the estimated ratio of kinetic to binding energy is \(T/|E_{bind}|\approx 0.6\), rather than the classical \(0.5\), because pressure and acceleration fields contribute to the internal dynamics. This does not alter the standard theorem as a mathematical identity; rather, it shows that the passage from stationarity to vanishing total virial derivative can depend sensitively on the kinematics and constitutive assumptions of the model under study [1801.06453].

Source: https://www.emergentmind.com/topics/dynamic-virial-theorem