---
title: Critical Biconical Vector Model
url: https://www.emergentmind.com/topics/critical-biconical-vector-model
type: topic
---

# Critical Biconical Vector Model

The critical biconical vector model describes multicritical behavior in three-dimensional anisotropic antiferromagnets in a uniform magnetic field \(H\) directed along the \(z\)-axis. In that setting, the staggered magnetization splits into two order parameters: a two-component field \(\phi_\perp=(\phi_x,\phi_y)\), perpendicular to the field and associated with XY-like spin-flop ordering, and a one-component field \(\phi_\parallel=\phi_z\), parallel to the field and associated with Ising-like ordering. The model couples these order parameters to the conserved uniform magnetization \(m\) along \(z\), and its critical dynamics combine relaxational, reversible, and mode-coupling contributions. In \(d=3\) with \((n_\perp,n_\parallel)=(2,1)\), the two-loop renormalization-group analysis identifies the biconical fixed point as the stable static multicritical fixed point and shows that, although asymptotic strong scaling is restored for the two order parameters, experimentally accessible effective exponents remain distinct because the asymptotic regime is reached only extremely slowly [1006.1766].

## 1. Physical setting and multicritical structure

The model is formulated for anisotropic antiferromagnets in an external field, where the field selects a preferred direction and thereby decomposes the staggered magnetization into perpendicular and parallel sectors. The perpendicular order parameter \(\phi_\perp\) has \(n_\perp=2\) components and describes XY-like spin-flop order; the parallel order parameter \(\phi_\parallel\) has \(n_\parallel=1\) component and describes Ising-like antiferromagnetic order along the field direction [1006.1766].

Upon varying temperature and field, or equivalently temperature and anisotropy, two second-order transition lines meet at a multicritical point. One line separates the disordered paramagnet from the spin-flop phase, where \(\phi_\perp \neq 0\) and \(\phi_\parallel=0\). The other separates the disordered phase from the antiferromagnetic phase with \(\phi_\parallel \neq 0\) and \(\phi_\perp=0\). The distinction between tetracriticality and bicriticality is determined by the structure of the ordered sector. Tetracriticality occurs when an intermediate biconical phase exists with \(\phi_\perp \neq 0\) and \(\phi_\parallel \neq 0\), so that four second-order lines meet. Bicriticality occurs when no biconical phase intervenes and the two ordered phases are separated by a first-order line, so that three second-order lines meet [1006.1766].

At the mean-field level, tetracriticality is favored when the quartic cross-coupling satisfies \(u_\times^2 < u_\perp u_\parallel\), while bicriticality and first-order separation occur when \(u_\times^2 \ge u_\perp u_\parallel\). The renormalization-group treatment refines these criteria through fluctuation effects. A plausible implication is that the multicritical topology cannot be inferred reliably from bare couplings alone once the fluctuation-induced flow through the full coupling space is taken into account.

## 2. Static field theory and universality classes

The static critical behavior is described by a coupled Landau–Ginzburg–Wilson functional involving \(\phi_\perp\), \(\phi_\parallel\), and the conserved secondary density \(m\):
\[
\mathcal{H}=\int d^d x \Bigg[
\frac{1}{2} r_\perp \phi_\perp^2
+\frac{1}{2} r_\parallel \phi_\parallel^2
+\frac{1}{2} (\nabla \phi_\perp)^2
+\frac{1}{2} (\nabla \phi_\parallel)^2
+\frac{u_\perp}{4!} (\phi_\perp^2)^2
+\frac{u_\parallel}{4!} (\phi_\parallel^2)^2
+\frac{u_\times}{4!} \phi_\perp^2 \phi_\parallel^2
+\frac{1}{2} m^2
+\frac{1}{2} \gamma_\perp m \phi_\perp^2
+\frac{1}{2} \gamma_\parallel m \phi_\parallel^2
-hm
\Bigg].
\]
Here \(r_\perp\) and \(r_\parallel\) are bare masses, \(u_\perp\) and \(u_\parallel\) are self-couplings, \(u_\times\) couples the two order-parameter sectors, \(\gamma_\perp\) and \(\gamma_\parallel\) couple \(m\) to the order parameters, and \(h\) is the conjugate field to \(m\) [1006.1766].

The \(\gamma\)-couplings are not merely auxiliary. In antiferromagnets in a field they encode the energetics of the interaction between uniform magnetization and staggered order, and they also generate reversible dynamic terms under renormalization. In the perpendicular sector one may equivalently use the complex field
\[
\psi=\phi_x-i\phi_y,
\]
while retaining \(\phi_\parallel\) as a real scalar.

The static renormalization-group flow admits three canonical fixed points. The decoupled fixed point \(D\) has \(u_\times^*=0\), so the two order parameters are effectively independent. The isotropic Heisenberg fixed point \(H\) has \(u_\perp^*=u_\parallel^*=u_\times^*\) and is \(O(3)\)-symmetric. The biconical fixed point \(B\) has \(u_\perp^* \neq u_\parallel^* \neq u_\times^*\) and corresponds to anisotropic multicriticality with simultaneous ordering [1006.1766].

For \(d=3\) and \((n_\perp,n_\parallel)=(2,1)\), the two-loop static renormalization-group calculation with generalized Padé–Borel resummation yields the biconical fixed point as the stable one in the full coupling space, while the isotropic Heisenberg fixed point is unstable. In this regime, the stable static universality class therefore corresponds to tetracritical behavior rather than bicritical behavior. This suggests that, for the physical parameter choice relevant to anisotropic antiferromagnets in a field, biconical criticality is the generic asymptotic static outcome unless initial conditions place the system inside the attraction basin of the Heisenberg fixed point.

## 3. Dynamic field theory and Langevin structure

The dynamics combine relaxational kinetics for the order parameters with reversible precession and diffusion of the conserved density \(m\). The resulting dynamic universality behavior blends ingredients of model \(F\) for the XY-like sector and model \(C\) for the Ising-like sector [1006.1766].

The bare equations of motion are
\[
\partial_t \phi_{\perp 0}^\alpha
=
-\Gamma_\perp' \frac{\delta \mathcal{H}}{\delta \phi_{\perp 0}^\alpha}
+\Gamma_\perp'' \epsilon^{\alpha\beta z} \frac{\delta \mathcal{H}}{\delta \phi_{\perp 0}^\beta}
+g \epsilon^{\alpha\beta z} \phi_{\perp 0}^\beta \frac{\delta \mathcal{H}}{\delta m_0}
+\theta_{\phi_\perp}^\alpha,
\]
\[
\partial_t \phi_{\parallel 0}
=
-\Gamma_\parallel \frac{\delta \mathcal{H}}{\delta \phi_{\parallel 0}}
+\theta_{\phi_\parallel},
\]
\[
\partial_t m_0
=
\lambda \nabla^2 \frac{\delta \mathcal{H}}{\delta m_0}
+g \epsilon^{z\alpha\beta} \phi_{\perp 0}^\alpha \frac{\delta \mathcal{H}}{\delta \phi_{\perp 0}^\beta}
+\theta_m.
\]
Here \(\epsilon^{ijk}\) is the Levi-Civita symbol, \(\alpha,\beta=x,y\), and repeated indices are summed. The kinetic coefficient \(\Gamma_\perp=\Gamma_\perp'+i\Gamma_\perp''\) is complex; \(\Gamma_\perp''\) is the precessional part generated by renormalization when \(\gamma_\perp \neq 0\). By contrast, \(\Gamma_\parallel\) and \(\lambda\) are real. The mode-coupling constant \(g\) couples the conserved density to the transverse order parameter.

The stochastic forces satisfy Einstein relations,
\[
\langle \theta_{\phi_\perp}^\alpha(x,t)\theta_{\phi_\perp}^\beta(x',t')\rangle
=
2\Gamma_\perp' \delta(x-x')\delta(t-t')\delta^{\alpha\beta},
\]
\[
\langle \theta_{\phi_\parallel}(x,t)\theta_{\phi_\parallel}(x',t')\rangle
=
2\Gamma_\parallel \delta(x-x')\delta(t-t'),
\]
\[
\langle \theta_m(x,t)\theta_m(x',t')\rangle
=
-2\lambda \nabla^2 \delta(x-x')\delta(t-t').
\]

Along the individual critical lines, the dynamic universality classes reduce to model \(F\) for the spin-flop line and model \(C\) for the Ising-like antiferromagnetic line. At the multicritical point, however, the coupled dynamics define a distinct universality class governed by the biconical fixed point, with reversible terms and coupling to \(m\) playing an essential role.

## 4. Renormalization-group formulation and two-loop flow variables

The field-theoretic renormalization-group treatment uses dimensional regularization and minimal subtraction. Bare and renormalized quantities are related by
\[
\phi_{\perp 0}=Z_{\phi_\perp}^{1/2}\phi_\perp,\qquad
\phi_{\parallel 0}=Z_{\phi_\parallel}^{1/2}\phi_\parallel,\qquad
m_0=Z_m^{1/2}m,
\]
\[
u_{i0}=\mu^\epsilon Z_{u_i}u_i,\qquad
\gamma_{i0}=Z_{\gamma_i}\gamma_i,\qquad
g_0=\mu^{\epsilon/2}Z_g g,
\]
with \(i\in\{\perp,\parallel,\times\}\), \(\mu\) the renormalization-group scale, and \(\epsilon=4-d\). The kinetic coefficients \(\Gamma_{\perp 0}\), \(\Gamma_{\parallel 0}\), and \(\lambda_0\) renormalize through \(Z_{\Gamma_\perp}\), \(Z_{\Gamma_\parallel}\), and \(Z_\lambda\) [1006.1766].

The dynamic analysis is organized in terms of dimensionless time-scale ratios and mode couplings,
\[
w_\perp=\Gamma_\perp/\lambda,\qquad
w_\parallel=\Gamma_\parallel/\lambda,\qquad
v=\Gamma_\parallel/\Gamma_\perp=w_\parallel/w_\perp,
\]
\[
v_\perp=\Gamma_\perp/\Gamma_\perp^+=w_\perp/w_\perp^+,\qquad
f_\perp=g/\sqrt{\Gamma_\perp' \lambda},\qquad
F=g/\lambda.
\]
Their renormalization-group flows satisfy
\[
l\frac{d w_\perp}{dl}=w_\perp\left[\zeta_{\Gamma_\perp}-\zeta_\lambda\right],\qquad
l\frac{d w_\parallel}{dl}=w_\parallel\left[\zeta_{\Gamma_\parallel}-\zeta_\lambda\right],
\]
\[
l\frac{d f_\perp}{dl}
=
-\frac{f_\perp}{2}
\left[
\epsilon+\zeta_\lambda-2\zeta_m+\operatorname{Re}\!\left(\frac{w_\perp}{w_\perp'}\zeta_{\Gamma_\perp}\right)
\right].
\]
The anomalous dimension of the conserved transport coefficient is
\[
\zeta_\lambda
=
\frac{1}{2}\gamma_\perp^2+\frac{1}{4}\gamma_\parallel^2
-\frac{1}{2}f_\perp^2\left[1+Q(\gamma_\perp,w_\perp,F)\right],
\]
where \(Q\) starts at two-loop order and coincides with the corresponding model \(F\) function.

The two-loop anomalous dimensions of the order-parameter kinetic coefficients contain model-\(A\)-like contributions together with mode-coupling and reversible terms. Writing
\[
D_\perp \equiv w_\perp\gamma_\perp-iF,
\]
and
\[
X_\perp \equiv 1+\ln\!\left[\frac{2v}{1+v}\right]
-\left(1+\frac{2}{v}\right)\ln\!\left[\frac{2(1+v)}{2+v}\right],
\]
the function \(\zeta_{\Gamma_\perp}\) contains the model \(A\) contribution \(\zeta_{\Gamma_\perp}^{(A)}(\{u\},v_\perp,v)\), the term \(D_\perp^2/[w_\perp(1+w_\perp)]\), and further two-loop contributions involving \(A_\perp\), \(B_\perp\), and \(X_\perp\). Similarly, \(\zeta_{\Gamma_\parallel}\) contains \(\bar{\zeta}_{\Gamma_\parallel}^{(C)}\), a model-\(A\) contribution \(\zeta_{\Gamma_\parallel}^{(A)}(\{u\},v_\perp,v)\), and reversible two-loop terms involving \(T_1\) and \(T_2\). The time-scale ratio \(v\) obeys
\[
\frac{dv}{dl}=v\left[\zeta_{\Gamma_\parallel}-\zeta_{\Gamma_\perp}\right].
\]
Because \(\zeta_{\Gamma_\perp}\) can be complex, the imaginary part must flow to zero at the fixed point in order to recover asymptotic power laws.

A central structural point is the appearance of logarithmic dependence on \(v\) in the two-loop functions. This dependence is decisive for the distinction between strict asymptotic scaling and the observable, nonasymptotic regime.

## 5. Fixed points and asymptotic dynamic scaling

Solving the dynamic flow equations at the static fixed points yields two dynamic fixed points in the biconical sector and two in the isotropic Heisenberg sector. Their defining parameters and dynamic exponents are as follows [1006.1766].

| Static sector | Dynamic fixed point | Values |
|---|---|---|
| Biconical | \(s^*=0\) | \(f_\perp^*=1.232\), \(q^*=1.167\times10^{-86}\), \(z_{OP}=2.048\), \(z_m=1.131\) |
| Biconical | \(s^*=0.705\) | \(f_\perp^*=1.232\), \(q^*=2.51\times10^{-782}\), \(z_{OP}=2.048\), \(z_m=1.131\) |
| Isotropic Heisenberg | \(s^*=0\) | \(f_\perp^*=1.211\), \(q^*=3.324\times10^{-8}\), \(z_{OP}=2.003\), \(z_m=1.542\) |
| Isotropic Heisenberg | \(s^*=0.698\) | \(f_\perp^*=1.211\), \(q^*=3.16\times10^{-66}\), \(z_{OP}=2.003\), \(z_m=1.542\) |

Here \(s^*\) denotes the ratio \(w_\perp''/w_\perp'\), and \(q^*=w_\parallel/w_\perp'\). The difference between the two \(s^*\) values in a given static sector does not affect the exponents because \(q^*\) is vanishingly small in both cases. The extreme smallness of \(q^*\), and correspondingly of \(v\) along the full flow, is a defining feature of the dynamic problem.

The dynamic exponents satisfy
\[
z_o=2+\zeta_o^*,\qquad o\in\{\perp,\parallel,m\}.
\]
When \(f_\perp^*\neq 0\) is finite, the multicritical dynamics obey the exact relation
\[
z_\perp+z_m=\frac{2\phi}{\nu},
\]
with \(\nu\) the correlation-length exponent and \(\phi\) the crossover exponent at the static fixed point. Because the asymptotic flow approaches \(w_\perp \to 0\) and \(w_\parallel \to 0\) with finite ratio \(v^*\), the two order-parameter sectors satisfy
\[
z_\perp=z_\parallel \equiv z_{OP},
\]
so asymptotic strong scaling holds within the order-parameter sector, whereas weak scaling persists with respect to the conserved density because \(z_m \neq z_{OP}\).

For the physically stable biconical fixed point in \(d=3\), the asymptotic exponents are therefore \(z_{OP}=2.048\) and \(z_m=1.131\). In the Heisenberg sector, which is reached only within its static attraction basin, the corresponding values are \(z_{OP}=2.003\) and \(z_m=1.542\).

## 6. Effective exponents and the experimentally accessible regime

The asymptotic equalization of \(z_\perp\) and \(z_\parallel\) is not the behavior expected in experimentally accessible scales. The paper defines effective exponents by inserting the running couplings into the dynamic anomalous dimensions,
\[
z_o^{\mathrm{eff}}(l)=2+\zeta_o\big(u(l),w_\perp(l),w_\parallel(l),f_\perp(l),\ldots\big),
\]
and then solving the full flow with the static couplings fixed at their fixed-point values [1006.1766].

In the complete dynamic parameter space, described in the paper as the “background,” the coefficients of the \(\ln v\) terms in \(\zeta_{\Gamma_\perp}\) are reduced relative to one loop, and the flow remains almost one-loop-like. The resulting effective exponents display weak-scaling-like plateaus:
\[
z_\parallel^{\mathrm{eff}} \approx 2.04,\qquad
z_\perp^{\mathrm{eff}} \approx z_m^{\mathrm{eff}} \approx 1.6.
\]
Even for \(\ln l \approx -10^6\), no visible changes occur in these effective exponents, which means that the asymptotic subspace is not reached on realistic scales.

Within the asymptotic subspace itself, where \(w_\perp=w_\parallel=0\) and \(v\) is finite, the approach to the fixed-point values is still extremely slow in the biconical case. Up to \(\ln l \approx -10^4\), one has \(z_\perp^{\mathrm{eff}}<z_{OP}\) and \(z_m^{\mathrm{eff}}>z_m\). By contrast, the isotropic Heisenberg case reaches its fixed-point values more readily.

This distinction between asymptotic and effective scaling is central. A common misunderstanding would be to identify the two-loop restoration of strong scaling in the order-parameter sector with an immediately observable equality of relaxation rates. The calculation implies the opposite: the asymptotic strong-scaling fixed point exists, but the flow toward it is so slow, and the fixed-point value of the relevant time-scale ratio is so small, that observable dynamics retain different apparent exponents over wide scale intervals. The experimentally relevant behavior is therefore nonasymptotic.

The same conclusion extends to dynamic correlation functions and line shapes. The scaling forms
\[
C_\alpha(k,\omega)\sim k^{-2+\eta_\alpha} f_\alpha\!\left(\omega/k^{z_\alpha}\right),\qquad \alpha\in\{\perp,\parallel,m\},
\]
inherit anisotropic effective exponents in the accessible regime, and the evolving time-scale ratios enter the line-shape functions. The paper therefore notes that the line shapes can deviate significantly from simple Lorentzian forms.

## 7. Relation to earlier analyses, experimental significance, and open problems

Earlier one-loop analyses indicated different asymptotic scaling of the order-parameter relaxation times, \(z_\perp \neq z_\parallel\). The complete two-loop calculation changes that conclusion by revealing a dynamic fixed point at which strong scaling in the order-parameter sector is restored, \(z_\perp=z_\parallel\), provided the flow reaches the asymptotic subspace [1006.1766]. What remains unchanged is the practical importance of nonasymptotic behavior: because \(q^*\) ranges from approximately \(10^{-8}\) to approximately \(10^{-782}\), realistic experiments and simulations are expected to observe distinct effective exponents rather than the asymptotic equality.

For experiments, the most direct implications concern probes of the staggered magnetization, such as neutron scattering. The analysis implies distinct apparent relaxation rates for transverse and longitudinal order-parameter fluctuations over broad temperature ranges near the multicritical point, with \(z_\parallel^{\mathrm{eff}} > z_\perp^{\mathrm{eff}}\), and a different critical slowing down for the conserved density \(m\). Transport coefficients, including relaxation rates of the staggered magnetization and the diffusion of \(m\), are therefore expected to exhibit plateau-like effective behavior rather than immediate convergence to asymptotic exponents.

The two-loop treatment also clarifies the role of proximity to a dynamic stability boundary between strong-scaling and weak-scaling fixed points. Near such boundaries, small transient exponents can produce nearly stationary effective behavior over many decades of the flow parameter \(l\). This suggests that careful interpretation of simulations and experiments requires separating fixed-point properties from crossover-controlled effective scaling.

Several limitations remain explicit. Dynamic renormalization-group results beyond two loops are not presently available for this model. While two-loop analysis together with resummation captures the stability interchange in the static sector and the effective weak-scaling behavior in the dynamic sector, higher-order corrections could refine the numerical estimates. The extremely slow approach to the asymptotic regime points to the continuing importance of crossover effects, nonasymptotic scaling functions, and detailed comparison with simulations and experiments, including dynamic structure factors. Real materials may also contain additional anisotropies or couplings; the paper notes that such complications can be incorporated, although they should not alter the universal multicritical behavior near the biconical fixed point.

Source: https://www.emergentmind.com/topics/critical-biconical-vector-model