Two-Dimensional Doi–Onsager Model

• The two-dimensional Doi–Onsager model is a mean-field theory for rod-like molecules using a circular orientation space with head–tail symmetry.
• It demonstrates a continuous phase transition at the critical coupling Kc=3π/4, marking the loss of stability of the isotropic state.
• Fourier analysis, coercivity estimates, and degree theory are employed to rigorously characterize bifurcation structure and uniqueness.

The two-dimensional Doi–Onsager model is a mean-field theory for orientational ordering of rod-like molecules when the orientation is reduced to a single angular variable on the circle. In its planar equilibrium form, the unknown is a probability measure or density on $S^1$, subject to head–tail symmetry, and the competition between entropic spreading and an interaction kernel determines whether the isotropic state remains stable or nonuniform stationary states emerge. Recent phase-transition analysis for the kernel $W_{DO}(\theta)=-|\sin(2\pi\theta)|$ establishes that the critical coupling is exactly $K_c=K_\#=3\pi/4$ and that the transition is continuous, while earlier nonlinear-functional-analytic work on the stationary equation with general even periodic kernels described uniqueness of the isotropic state in a low-parameter regime and the local bifurcation structure of nontrivial branches (Mun et al., 17 Apr 2026).

1. Mean-field formulation on the circle

In the phase-transition formulation, the state space is the set of probability measures $q\in \mathcal P(T)$ on the unit circle

$T=\left[-\frac12,\frac12\right)$

with periodic boundary conditions. The free energy is

$$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$

where $K\ge 0$ is the coupling strength and $W$ is an even interaction potential. For the two-dimensional Doi–Onsager model, the interaction is

$W_{DO}(\theta)=-|\sin(2\pi\theta)|.$

The uniform distribution

$q_u\equiv 1$

is the isotropic state. At $W_{DO}(\theta)=-|\sin(2\pi\theta)|$0, the free energy reduces to relative entropy with respect to $W_{DO}(\theta)=-|\sin(2\pi\theta)|$1, so the isotropic state is uniquely minimizing. As $W_{DO}(\theta)=-|\sin(2\pi\theta)|$2 increases, the interaction term can destabilize $W_{DO}(\theta)=-|\sin(2\pi\theta)|$3, and the resulting change in the structure of minimizers is the phase transition (Mun et al., 17 Apr 2026).

A closely related stationary formulation writes the unknown as an orientation distribution function $W_{DO}(\theta)=-|\sin(2\pi\theta)|$4 on $W_{DO}(\theta)=-|\sin(2\pi\theta)|$5, normalized by

$W_{DO}(\theta)=-|\sin(2\pi\theta)|$6

and constrained by the reversal symmetry

$W_{DO}(\theta)=-|\sin(2\pi\theta)|$7

In that formulation, the constant distribution

$W_{DO}(\theta)=-|\sin(2\pi\theta)|$8

represents the isotropic phase, and equilibrium states are stationary solutions of a nonlinear self-consistency equation (Niksirat et al., 2015).

2. Symmetry, stationary equations, and isotropic equilibrium

The head–tail symmetry of rods is encoded by $W_{DO}(\theta)=-|\sin(2\pi\theta)|$9-periodicity. In the stationary formulation with convolution kernel $K_c=K_\#=3\pi/4$0, the interaction potential is

$K_c=K_\#=3\pi/4$1

where $K_c=K_\#=3\pi/4$2 is the concentration/inverse temperature parameter, and the kernel satisfies

$K_c=K_\#=3\pi/4$3

Stationary solutions satisfy the Doi–Onsager self-consistent equation

$K_c=K_\#=3\pi/4$4

so equilibria are fixed points of a nonlinear integral operator (Niksirat et al., 2015).

For functional-analytic treatment, the stationary problem can be rewritten in terms of the centered potential

$K_c=K_\#=3\pi/4$5

which satisfies

$K_c=K_\#=3\pi/4$6

together with

$K_c=K_\#=3\pi/4$7

The natural Hilbert space is

$K_c=K_\#=3\pi/4$8

and the equation becomes the fixed-point problem $K_c=K_\#=3\pi/4$9 with compact nonlinear operator

$q\in \mathcal P(T)$0

This representation isolates the isotropic state as the trivial solution $q\in \mathcal P(T)$1 and makes degree-theoretic and bifurcation arguments available (Niksirat et al., 2015).

3. Fourier structure and mode selection

The two-dimensional model is especially tractable in Fourier variables. For the phase-transition study, the Doi–Onsager kernel has the explicit expansion

$q\in \mathcal P(T)$2

This shows that the interaction is $q\in \mathcal P(T)$3-periodic, so only the even Fourier modes are active. Its nonzero Fourier coefficients are

$q\in \mathcal P(T)$4

and in particular

$q\in \mathcal P(T)$5

The dominant attractive mode is therefore the $q\in \mathcal P(T)$6 mode, which determines the linear stability threshold of the isotropic state (Mun et al., 17 Apr 2026).

In the stationary-equation literature with general $q\in \mathcal P(T)$7-periodic kernels, one writes

$q\in \mathcal P(T)$8

with coefficients $q\in \mathcal P(T)$9 controlling uniqueness and bifurcation. For the classical Onsager kernel in that normalization,

$T=\left[-\frac12,\frac12\right)$0

The corresponding local bifurcation points are

$T=\left[-\frac12,\frac12\right)$1

and all bifurcations are supercritical. The first bifurcating branch occurs at

$T=\left[-\frac12,\frac12\right)$2

is stable, and all higher bifurcating branches are unstable (Niksirat et al., 2015).

These two Fourier descriptions are written in different normalizations, but both emphasize the same structural fact: the planar Doi–Onsager problem is mode-selective, and the first loss of isotropic stability is governed by a distinguished low Fourier mode.

4. Critical coupling and the continuous phase transition

The phase-transition threshold $T=\left[-\frac12,\frac12\right)$3 is defined by the minimization problem for $T=\left[-\frac12,\frac12\right)$4: if $T=\left[-\frac12,\frac12\right)$5, the isotropic state $T=\left[-\frac12,\frac12\right)$6 is the unique global minimizer; if $T=\left[-\frac12,\frac12\right)$7, $T=\left[-\frac12,\frac12\right)$8 is still a minimizer; and if $T=\left[-\frac12,\frac12\right)$9, there exists a nonuniform minimizer. The transition is continuous when the unique minimizer at criticality is $$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$0, equivalently when minimizers for $$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$1 converge back to the isotropic state. In general,

$$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$2

where $$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$3 is the linear stability threshold determined by the second variation at $$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$4 (Mun et al., 17 Apr 2026).

For a general interaction, the second variation at the uniform state is

$$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$5

Hence $$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$6 is linearly stable as long as

$$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$7

For the Doi–Onsager kernel, the maximal attractive coefficient occurs at $$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$8, so

$$\mathcal F_K(q)=\int_T \log q\,dq(\theta)-K\iint_{T\times T}W(\theta-\theta')\,dq(\theta)\,dq(\theta'),$$9

The main result for the two-dimensional Doi–Onsager model is that the phase transition occurs exactly at this linear threshold and is continuous: $K\ge 0$0 At criticality, the isotropic state is still the unique global minimizer, so nonuniform minimizers do not pre-exist below the stability threshold (Mun et al., 17 Apr 2026).

In the stationary formulation with parameter $K\ge 0$1, a related but differently normalized picture appears. If

$K\ge 0$2

then the only stationary solution is the isotropic one,

$K\ge 0$3

For the classical Onsager kernel, since

$K\ge 0$4

this gives the uniqueness regime

$K\ge 0$5

Because the first bifurcation for that kernel occurs at $K\ge 0$6, there is a gap between the low-$K\ge 0$7 uniqueness theorem and the first local bifurcation point. The phase-transition analysis at $K\ge 0$8 resolves an analogous threshold question in the mean-field minimization setting by proving exact coincidence of onset and loss of linear stability (Niksirat et al., 2015).

5. Proof methods: coercivity, degree theory, and local bifurcation

The recent proof of continuity for the planar Doi–Onsager transition is based on a sharp coercivity estimate derived from the constrained Lebedev–Milin inequality. In the normalized $K\ge 0$9-periodic setting, the key estimate is

$W$0

with equality only for the one-parameter family

$W$1

This yields the decomposition

$W$2

so nonnegativity follows from the entropy inequality and a Fourier decay condition. In the normalized theorem, if $W$3 is $W$4-periodic, normalized by $W$5, and satisfies

$W$6

then

$W$7

the phase transition is continuous, and if $W$8, then $W$9 is the unique critical point. For the Doi–Onsager kernel, the verification reduces to the explicit coefficient inequality

$W_{DO}(\theta)=-|\sin(2\pi\theta)|.$0

equivalently

$W_{DO}(\theta)=-|\sin(2\pi\theta)|.$1

after rescaling by $W_{DO}(\theta)=-|\sin(2\pi\theta)|.$2 (Mun et al., 17 Apr 2026).

Earlier stationary analysis used a different toolkit. The low-parameter uniqueness theorem was obtained by applying Leray–Schauder degree theory to

$W_{DO}(\theta)=-|\sin(2\pi\theta)|.$3

The proof combines compactness of $W_{DO}(\theta)=-|\sin(2\pi\theta)|.$4, an a priori bound for solutions, homotopy invariance of degree,

$W_{DO}(\theta)=-|\sin(2\pi\theta)|.$5

and the fact that every solution has index $W_{DO}(\theta)=-|\sin(2\pi\theta)|.$6. The local bifurcation analysis assumes an ordering of negative Fourier coefficients, written as

$W_{DO}(\theta)=-|\sin(2\pi\theta)|.$7

so that the critical values

$W_{DO}(\theta)=-|\sin(2\pi\theta)|.$8

are distinct. At each $W_{DO}(\theta)=-|\sin(2\pi\theta)|.$9, two nontrivial solutions bifurcate from the trivial branch. Stability of bifurcated branches is determined by the sign of $q_u\equiv 1$0, using Sattinger’s theorem: the branch is supercritical if $q_u\equiv 1$1 and subcritical if $q_u\equiv 1$2; for $q_u\equiv 1$3 all bifurcated solutions are unstable, while for $q_u\equiv 1$4 they are stable if $q_u\equiv 1$5 and unstable if $q_u\equiv 1$6 (Niksirat et al., 2015).

Together, these results separate two analytical layers of the planar Doi–Onsager problem. Degree theory and local bifurcation describe existence and branch structure of stationary solutions, whereas the sharp coercivity argument identifies the exact mean-field phase-transition threshold and proves continuity at criticality.

6. Scope, variants, and the meaning of “two-dimensional”

The expression “two-dimensional Doi–Onsager model” is not uniform across the literature. In the planar equilibrium and stationary works discussed above, two-dimensional means that the orientation of a rod is described by a single angle on $q_u\equiv 1$7, or equivalently by a probability measure on the circle, with $q_u\equiv 1$8-periodicity representing head–tail symmetry (Mun et al., 17 Apr 2026). By contrast, in the small Deborah number analysis without hydrodynamics, the physical space dimension may be $q_u\equiv 1$9 or $W_{DO}(\theta)=-|\sin(2\pi\theta)|$00, but the orientation variable remains

$W_{DO}(\theta)=-|\sin(2\pi\theta)|$01

and the model is a Smoluchowski-type equation on $W_{DO}(\theta)=-|\sin(2\pi\theta)|$02, not a planar orientation model on $W_{DO}(\theta)=-|\sin(2\pi\theta)|$03 (Liu et al., 2017).

The same distinction appears in kinetic-to-continuum derivations. The Bingham-closure derivation of a Q-tensor equation and the subsequent small Deborah number limit to the Ericksen–Leslie system are written in the standard 3D rod-like setting with orientational distribution $W_{DO}(\theta)=-|\sin(2\pi\theta)|$04, $W_{DO}(\theta)=-|\sin(2\pi\theta)|$05, symmetric traceless $W_{DO}(\theta)=-|\sin(2\pi\theta)|$06 Q-tensors, and 3D tensor identities. That framework is part of Doi–Onsager theory, but it is not a literal reduction to planar orientation space (Wang et al., 2013).

A common misconception is therefore to identify every “2D Doi–Onsager” reference with the same mathematical model. The planar circle model, the spatially two-dimensional Smoluchowski equation with $W_{DO}(\theta)=-|\sin(2\pi\theta)|$07-valued orientations, and the Q-tensor or director limits derived from the 3D kinetic theory belong to the same Doi–Onsager hierarchy but address different variables, symmetries, and asymptotic regimes. The planar model distinguished by the kernel $W_{DO}(\theta)=-|\sin(2\pi\theta)|$08 is the one for which the exact threshold

$W_{DO}(\theta)=-|\sin(2\pi\theta)|$09

and the continuity of the phase transition are now rigorously established (Mun et al., 17 Apr 2026).