Flow Crossover in Multiphysics Regimes

• Flow crossover is a context-dependent phenomenon that marks transitions between distinct flow regimes or effective parameter trajectories across various fields.
• It is quantified by specific observables—such as midrapidity slopes in nuclear collisions, diverging correlation lengths in granular media, and parameter flows in RG studies—enabling precise diagnostic insights.
• The concept underscores the importance of control parameters and reparameterisation strategies to avoid misinterpretation of transient signatures in systems ranging from hydrodynamics to spectral statistics.

Flow crossover is a context-dependent term used across several research areas to denote a transition between distinct flow regimes, or, in renormalisation and spectral-statistics settings, a trajectory in an effective parameter space whose endpoints represent different physical phases or fixed points. In the literature surveyed here, the phrase refers variously to the beam-energy dependence of directed flow in relativistic heavy-ion collisions, the disorder-driven motion of fitted spectral parameters between Gaussian orthogonal ensemble and Poisson statistics, the crossover from quasi-static to dense granular flow, the Saffman-induced reorganisation of dipolar membrane hydrodynamics, the no-slip to slip transition near contaminated air–water interfaces, pressure-driven and surface-driven competition in nano-channels, Ohmic-to-viscous electron transport in the Gurzhi regime, crossover and self-discharge in redox flow batteries, and tricritical-to-critical or decoupled-to-first-order renormalisation-group flow (Ivanov et al., 2014, Duarte-Filho et al., 17 Dec 2025, Gimbert et al., 2012, Bhattacharya et al., 26 Apr 2026, Maali et al., 2017, Anand et al., 2023, Li et al., 2021, Ascencio et al., 2019, Gaite, 10 May 2026, Kamiya et al., 2010).

1. Flow crossover as a cross-disciplinary concept

A useful synthesis is that the term has two main meanings. In transport and hydrodynamics, it denotes a change in the dominant mechanism controlling motion: for example, from pressure-dominated to surface-driven transport, from Ohmic to viscous electron flow, from quasi-static to dense granular flow, or from near-field to screened far-field dipolar motion (Gimbert et al., 2012, Bhattacharya et al., 26 Apr 2026, Anand et al., 2023, Li et al., 2021). In statistical and field-theoretic settings, it denotes a trajectory in an effective low-dimensional parameter space, often interpreted as a flow toward a fixed point or toward a line of fixed-point-controlled behaviour (Duarte-Filho et al., 17 Dec 2025, Gaite, 10 May 2026, Kamiya et al., 2010, Kang et al., 2016).

The same phrase therefore covers both observable-space crossovers and parameter-space crossovers. In relativistic nuclear collisions, the relevant observable is the directed-flow coefficient $v_1(y)$ and especially its midrapidity slope $dv_1/dy|_{y=0}$, whose energy dependence discriminates among equations of state (Ivanov et al., 2014). In disordered spin chains, the relevant variables are fitted parameters $(\beta,\gamma)$ entering a ratio-distribution surmise, and disorder strength acts as the progression variable of a phenomenological flow (Duarte-Filho et al., 17 Dec 2025). In renormalisation-group applications, the objects that flow are couplings such as $u=\lambda/m$ and $g$, or phenomenological Binder and correlation-ratio observables, and the crossover is realised by convergence toward a critical manifold or away from an unstable fixed point (Gaite, 10 May 2026, Kamiya et al., 2010).

A concise comparison is useful.

Domain	Flow variables	Crossover description
Heavy-ion collisions	$v_1(y)$, $dv_1/dy|_{y=0}$	Hadronic vs first-order vs crossover EoS (Ivanov et al., 2014)
Spectral statistics	$(\beta,\gamma)$	GOE-to-Poisson trajectory in parameter space (Duarte-Filho et al., 17 Dec 2025)
Granular media	Stress and strain correlation lengths	Quasi-static to dense flow with critical scaling (Gimbert et al., 2012)
Membrane hydrodynamics	$R,\psi$, dipole flow field	Near-field $r^{-1}$ to far-field $dv_1/dy|_{y=0}$0 reorganisation (Bhattacharya et al., 26 Apr 2026)
Transitional turbulence	$dv_1/dy|_{y=0}$1	Structural critical-like crossover and smeared energetic crossover (Rolland, 2018)
RG and QFT	Couplings or disorder distributions	Tricritical-to-critical, decoupled-to-first-order, or ground-to-excited-state flow (Gaite, 10 May 2026, Kamiya et al., 2010, Kang et al., 2016)

This suggests that “flow crossover” is less a single theory than a family of problems in which a control parameter changes the geometry, statistics, or constitutive law of transport.

2. Directed flow in relativistic nuclear collisions

In relativistic heavy-ion physics, the term is closely associated with the question of whether directed flow can discriminate among qualitatively different equations of state. The relevant observable is defined through the Fourier expansion

$dv_1/dy|_{y=0}$2

with directed flow given by the first harmonic $dv_1/dy|_{y=0}$3, and the most diagnostic summary quantity is the midrapidity slope

$dv_1/dy|_{y=0}$4

Because directed flow is generated very early, during the compression stage, it is sensitive to the pressure gradients and hence to the stiffness or softness of the equation of state (Ivanov et al., 2014).

The analysis in the three-fluid dynamics model compares a purely hadronic equation of state, an equation of state with a first-order deconfinement transition, and one with a smooth crossover transition over $dv_1/dy|_{y=0}$5–$dv_1/dy|_{y=0}$6 GeV. The proton results are the main discriminator. The first-order-transition equation of state yields a strong wiggle in the excitation function of the proton $dv_1/dy|_{y=0}$7 slope at midrapidity, mostly in the negative range, while the purely hadronic equation of state gives a smooth and generally positive proton slope. Both disagree with the data. The crossover equation of state produces only a small proton antiflow and gives the best overall description of proton, antiproton, and pion directed flow (Ivanov et al., 2014).

The model interpretation is specific. In three-fluid dynamics, matter is represented by two baryon-rich fluids and a baryon-free fireball fluid, which makes the early nonequilibrium stage directly sensitive to the chosen equation of state. Antiprotons near midrapidity mainly probe the baryon-free fireball fluid and therefore the equation of state near zero net-baryon density, whereas proton flow is dominated by baryon-rich fluids and probes finite baryon density. This leads to the more nuanced conclusion that, although the crossover equation of state is preferred overall, the deconfinement equations of state used in the calculation appear too soft at high baryon densities and should be stiffer in the quark-gluon sector there (Ivanov et al., 2014).

A common misconception in this literature is that any negative midrapidity slope or “antiflow” necessarily signals a first-order phase transition. The comparison performed here argues against that. The pronounced wiggle expected from a first-order softest-point scenario is not seen in the data, whereas a smoother crossover softening is more consistent with the measured excitation function (Ivanov et al., 2014).

3. Parameter-space flows in spectral statistics and renormalisation theory

A second, more abstract use of the term appears in spectral statistics of disordered many-body systems. There, the crossover from Gaussian orthogonal ensemble statistics to Poisson statistics is represented as a trajectory in a two-dimensional parameter space derived from the consecutive-gap ratio distribution. If $dv_1/dy|_{y=0}$8 and

$dv_1/dy|_{y=0}$9

the paper introduces the two-parameter surmise

$(\beta,\gamma)$0

The fitted parameters $(\beta,\gamma)$1 are then treated as flow variables, with disorder strength playing the role of progression variable and the Poisson point $(\beta,\gamma)$2 interpreted as the fixed point corresponding to the many-body-localised phase (Duarte-Filho et al., 17 Dec 2025).

This representation is explicitly phenomenological rather than microscopic. The authors introduce

$(\beta,\gamma)$3

but do not derive $(\beta,\gamma)$4 from the Hamiltonian. In the random-field Heisenberg chain, the fitted points move smoothly from a GOE-like region toward $(\beta,\gamma)$5. In the exchange-disordered $(\beta,\gamma)$6-symmetric chain, by contrast, the flow does not terminate at Poisson and instead fluctuates around an apparent non-Poisson fixed point $(\beta,\gamma)$7, motivating a linearised stochastic differential equation and a discrete Lyapunov analysis for the stationary covariance cloud around that point (Duarte-Filho et al., 17 Dec 2025).

Renormalisation-group studies use related language. In the $(\beta,\gamma)$8 scalar theory describing tricritical-to-critical crossover, the relevant dimensionless couplings are

$(\beta,\gamma)$9

with RG time $u=\lambda/m$0. The tricritical fixed point sits at $u=\lambda/m$1, and the crossover is realised by convergence of trajectories toward the line connecting tricritical and critical fixed points. Near the origin, that connecting line is

$u=\lambda/m$2

The sextic coupling is essential because it creates the larger flow geometry, including a marginal direction and a second separatrix; without it there is only direct quartic critical flow (Gaite, 10 May 2026).

A distinct RG interpretation appears in the two-sublattice XY model with inter-sublattice biquadratic coupling

$u=\lambda/m$3

There the relevant perturbation has RG eigenvalue

$u=\lambda/m$4

so the decoupled 3D-XY fixed point is unstable and the crossover criterion is

$u=\lambda/m$5

Phenomenological couplings drift away from the decoupled fixed point with increasing size, and the flow shows no separatrix or new stable fixed point, supporting a weak first-order transition rather than a new universality class (Kamiya et al., 2010).

A further example is the random Fibonacci chain, where finite energy density is a relevant perturbation to the ground-state infinite-randomness fixed point. The running singlet probability $u=\lambda/m$6 interpolates between $u=\lambda/m$7 and $u=\lambda/m$8, and the crossover is encoded in the RG flow equation

$u=\lambda/m$9

Here the flow is simultaneously an RG flow of bond distributions, a running of $g$0, and a crossover scaling of observables between the $g$1 and $g$2 infinite-randomness regimes (Kang et al., 2016).

4. Hydrodynamic and transport-regime crossovers

Several papers use the term in the more literal sense of a change in flow law or transport regime. In compressed frictional granular media, the transition from quasi-static behaviour to dense flow is analysed through growing correlation lengths in stress redistribution and incremental strain localisation. The control parameter is

$g$3

with $g$4. The stress and strain correlation lengths satisfy

$g$5

and the inertial number jumps from quasi-static values $g$6–$g$7 to dense-flow values $g$8–$g$9. The onset is therefore interpreted not as immediate formation of a shear band of fixed thickness but as a critical transition with diverging spatial correlations (Gimbert et al., 2012).

In viscous fluid membranes coupled to surrounding solvent, the Saffman crossover changes not only the decay exponent of a dipolar flow but the entire phase-space structure of two-dipole dynamics. The single-dipole field crosses from

$v_1(y)$0

to

$v_1(y)$1

with $v_1(y)$2 the Saffman length. In the near field, the leading dipolar flow is purely radial and two identical quenched dipoles have effectively one-dimensional dynamics with $v_1(y)$3 linear in time. In the far field, a component along the dipole axis survives, radial and angular dynamics couple,

$v_1(y)$4

and the collapse law changes from $v_1(y)$5 to $v_1(y)$6 for pullers on the aligned attractive branch (Bhattacharya et al., 26 Apr 2026).

Near contaminated air–water interfaces, the crossover is frequency-driven rather than geometric. A vibrating sphere experiences viscous and elastic drag because trace impurities generate Marangoni stresses. The characteristic crossover frequency is

$v_1(y)$7

and the asymptotic force laws are

$v_1(y)$8

and

$v_1(y)$9

Low frequency therefore gives no-slip-like drag, while high frequency recovers the full-slip free-surface limit (Maali et al., 2017).

A related but distinct near-wall versus core-flow crossover appears in cylindrical nano-channels with traveling wall fluctuations. There the total axial flow is decomposed into pressure-driven Poiseuille flow and a first-order fluctuation-induced contribution. The crossover radius $dv_1/dy|_{y=0}$0 is defined by the condition $dv_1/dy|_{y=0}$1, leading to

$dv_1/dy|_{y=0}$2

This identifies a peripheral region where fluctuation-induced transport can compete with or oppose the bulk pressure-driven core flow (Anand et al., 2023).

In transitional plane Couette flow, the term denotes two consecutive but distinct changes. The first is a structural crossover, in which the order parameter $dv_1/dy|_{y=0}$3 measuring banded modulation decreases continuously and its response $dv_1/dy|_{y=0}$4 grows with system area in a critical-like manner. The second is an energetic crossover, where the kinetic energy $dv_1/dy|_{y=0}$5 rises sharply, $dv_1/dy|_{y=0}$6 peaks, and spatial-temporal coexistence of uniform turbulence and laminar–turbulent bands is observed, but the jump remains rounded and the peak saturates with size, consistent with a first-order transition smeared by finite noise (Rolland, 2018).

5. Electronic, thermal, and electrochemical transport

In electron hydrodynamics through constrictions, the Gurzhi crossover is controlled by the Gurzhi length

$dv_1/dy|_{y=0}$7

which separates Ohmic transport $dv_1/dy|_{y=0}$8 from viscous Stokes-like flow $dv_1/dy|_{y=0}$9. In the viscous limit, the force balance

$(\beta,\gamma)$0

reduces to a Stokes problem and the electric potential becomes the real part of an analytic complex potential. Conformal maps then yield closed-form potential profiles for slit and quantum point contact geometries, while the Gurzhi conductivity in a straight channel scales as

$(\beta,\gamma)$1

The paper also cautions that spatially inhomogeneous profiles usually associated with viscous flow can appear in a nonhydrodynamic nonlocal regime, as seen from the dispersive conductivity

$(\beta,\gamma)$2

so profile shape alone is not a sufficient hydrodynamic diagnostic (Li et al., 2021).

At much smaller scales, collisionless magnetic reconnection reveals another meaning of the phrase. In kinetic particle-in-cell simulations, plasma from one inflow side crosses the midplane and feeds the opposite-side exhaust, rather than simply turning into a same-side outflow. This source-tagged “flow crossover” is tied to field-aligned bulk acceleration and implies that reconnection outflows are more parallel than perpendicular to the magnetic field, especially for electrons. Ions acquire parallel flow mainly within the ion diffusion region, whereas electrons gain it mostly outside the electron diffusion region, near separatrices (Pianpanit et al., 29 Sep 2025).

In thermal transport, graphene provides a clear ballistic-to-diffusive crossover. The effective length-dependent conductivity is written as

$(\beta,\gamma)$3

or equivalently

$(\beta,\gamma)$4

with $(\beta,\gamma)$5 nm at $(\beta,\gamma)$6 K for supported graphene. Short, wide graphene samples of length $(\beta,\gamma)$7 nm reach about $(\beta,\gamma)$8 of the ballistic heat-conductance limit up to room temperature, whereas narrowing comparable samples into graphene nanoribbons drives transport back to an edge-controlled diffusive regime, with thermal conductivity scaling approximately as $(\beta,\gamma)$9 and reaching about $R,\psi$0 in $R,\psi$1-nm ribbons at room temperature (Bae et al., 2013).

Electrochemical flow systems introduce yet another usage. In disproportionation redox flow batteries, crossover refers to transport of active vanadium species through the porous separator, causing self-discharge. A lumped model writes

$R,\psi$2

and therefore

$R,\psi$3

Adaptive or augmented observers then estimate $R,\psi$4, $R,\psi$5, and the unknown crossover flux $R,\psi$6 simultaneously from voltage and operating data, with Lyapunov analysis proving convergence to a bounded residual set (Ascencio et al., 2019, Ascencio et al., 2019).

A different flow-battery architecture uses “crossover” in the literal membrane-transport sense. In a membraneless H$R,\psi$7-Br$R,\psi$8 system, a nanoporous “dispersion blocker” with $R,\psi$9 nm pores is placed between an electrolyte channel and a large-pore porous cathode. The design suppresses hydrodynamic dispersion and increases diffusive resistance, producing tribromide retention up to $r^{-1}$0 and supporting $r^{-1}$1 peak power and $r^{-1}$2 peak current density (Suss et al., 2016).

6. Common structures, misconceptions, and significance

Despite the diversity of applications, several recurrent structures appear. First, each crossover is organised by a small set of control parameters: beam energy in heavy-ion collisions, disorder strength in spectral statistics, stress distance to instability in granular media, $r^{-1}$3 in Saffman screening, $r^{-1}$4 near contaminated interfaces, $r^{-1}$5 in electron hydrodynamics, $r^{-1}$6 or $r^{-1}$7 in graphene, and coupling-space distance to a separatrix in RG problems (Ivanov et al., 2014, Duarte-Filho et al., 17 Dec 2025, Gimbert et al., 2012, Bhattacharya et al., 26 Apr 2026, Maali et al., 2017, Li et al., 2021, Bae et al., 2013, Gaite, 10 May 2026).

Second, the most informative descriptions often require a reparameterisation. Directed flow is reduced to the midrapidity slope $r^{-1}$8 (Ivanov et al., 2014). Spectrum statistics are encoded by $r^{-1}$9 rather than by raw histograms (Duarte-Filho et al., 17 Dec 2025). Granular flow is diagnosed by diverging correlation lengths rather than by a visible shear band (Gimbert et al., 2012). The Gurzhi crossover is better parameterised by $dv_1/dy|_{y=0}$00 than by temperature alone (Li et al., 2021). In batteries, crossover is elevated from an unmodelled disturbance to an estimated state or latent parameter (Ascencio et al., 2019, Ascencio et al., 2019).

Third, several papers explicitly warn against overinterpreting superficially similar signatures. A negative proton directed-flow slope is not, by itself, evidence for a first-order deconfinement transition (Ivanov et al., 2014). A peak in a response function does not, by itself, identify the order of a transition in transitional turbulence; finite-size scaling is required (Rolland, 2018). Spatially inhomogeneous current profiles in constrictions are not, by themselves, proof of hydrodynamic electron flow, because quasiballistic nonlocality can generate similar patterns (Li et al., 2021).

A plausible implication is that “flow crossover” functions as a diagnostic framework precisely because it forces attention onto the mechanism that controls transport or RG evolution in each regime. In some cases the crossover sharpens into critical behaviour, as in granular media or the structural loss of laminar–turbulent order (Gimbert et al., 2012, Rolland, 2018). In others it remains a smooth interpolation, as in the no-slip to slip transition or in GOE-to-Poisson spectral statistics (Maali et al., 2017, Duarte-Filho et al., 17 Dec 2025). In still others it is best viewed as a reorganisation of phase-space geometry, as in Saffman-screened membrane dipoles or tricritical-critical RG flow (Bhattacharya et al., 26 Apr 2026, Gaite, 10 May 2026).

Taken together, these works show that the phrase does not name a single universal phenomenon. It names a family of regime changes in which the dominant constitutive law, effective degrees of freedom, or fixed-point control changes in a way that can be tracked quantitatively by observables, fitted parameters, or RG trajectories.