Hurdle Theorem for Parallel Flows

• The paper reveals the main contribution: an analytical method that sharply bounds hitting probabilities and instability criteria in coupled parallel systems.
• It employs techniques such as Rayleigh quotient minimization and Sturm–Liouville analysis to derive precise thresholds for instability and rare events.
• Applications span fluid dynamics and queueing theory, offering concrete predictions for flow instability and overflow probabilities in multidimensional models.

The Hurdle Theorem for Parallel Flows provides a sharp analytical framework for understanding instability and rare-event probabilities in systems where two or more parallel processes interact under drift, boundary, and constraint conditions. It encompasses results in hydrodynamic stability, queueing theory, and optical transformations, unifying diverse problems under the common structure of barrier-crossing in multidimensional Markov models and Sturm–Liouville eigenvalue problems. The theorem yields both sufficient criteria for instability in fluid flows and precise asymptotic characterizations of rare event probabilities in parallel queueing models.

1. Mathematical Formulation in Parallel Queue Models

Consider the constrained random walk $X = \{X_k: k\ge 0\}$ on $\mathbb{Z}_+^2$ with increments $I_k \in \{(1,0), (-1,0), (0,1), (0,-1)\}$. The transitions are determined by arrival rates $\lambda_i$ and service rates $\mu_i$ for $i=1,2$, representing the dynamics of two parallel $M/M/1$ queues, with traffic intensities $\rho_i = \lambda_i/\mu_i < 1$.

Define the domain $$A_n = \{ x \in \mathbb{Z}_+^2 : x_1 + x_2 \leq n \}$$ and its boundary $\partial A_n$. The stopping times

$$\tau_n = \inf\{k\ge 0: X_k \in \partial A_n\}, \quad \tau_0 = \inf\{k\ge 0: X_k = (0,0)\}$$

are used to study the overflow probability $p_n(x) = \mathbb{P}_x[\tau_n < \tau_0]$.

Analysis of $p_n(x)$ proceeds by mapping the problem to an auxiliary random walk $Y$ on $\mathbb{Z} \times \mathbb{Z}_+$ with reversed first-coordinate jump probabilities and reflection at $y_2=0$. The hitting time of the diagonal $\partial B = \{ y : y_1 = y_2 \}$, denoted $\tau$, becomes the effective "hurdle" to cross.

The core result—the Hurdle Theorem for Parallel Flows—states that, for scaled starting states $x_n = \lfloor n x \rfloor$ ($x_1 + x_2 \le 1, x_1 > 0$),

$$\left| \frac{ \mathbb{P}_{x_n}[\tau_n<\tau_0] - \mathbb{P}_{y_n}[\tau < \infty] }{ \mathbb{P}_{x_n}[\tau_n<\tau_0] } \right| \leq e^{-Cn}$$

for some $C>0$ and all large $n$, with $y_n = (n - x_{n,1}, x_{n,2})$ (Ünlü et al., 2018).

In the "critical regime" $r^2 = \rho_1 \rho_2$, $r = (\lambda_1+\lambda_2)/(\mu_1+\mu_2)$, the diagonal-hitting probability is given explicitly by

$$\mathbb{P}_y[\tau < \infty] = r^{y_1-y_2} + \frac{r(1-r)}{r-\rho_2}\left( \rho_1^{y_1} - r^{y_1-y_2} \rho_1^{y_2} \right).$$

In the generic regime $r^2 < \rho_2$, $\mathbb{P}_y[\tau < \infty]$ can be approximated arbitrarily well via superpositions of log-linear harmonic functions parameterized on the characteristic surface $p(\beta, \alpha) = 1$.

2. Analytical Criteria in Hydrodynamic Parallel Flows

The original hydrodynamic context for the hurdle theorem is the stability analysis of inviscid, incompressible, 2D parallel shear flows $U(y)$ on a finite interval $y \in [y_1, y_2]$. The Rayleigh eigenvalue problem

$$(U-c)(\varphi'' - k^2 \varphi) - U'' \varphi = 0, \quad \varphi(y_1) = \varphi(y_2) = 0$$

is studied for criteria under which instability arises.

Define the curvature function

$W(y) = -\frac{U''(y)}{U(y) - \alpha}$

at a critical level $y_c$ where $U''(y_c)=0, U'(y_c)\neq 0$, and set $\alpha = U(y_c)$. The domain "hurdle"

$h = \left( \frac{\pi}{y_2 - y_1} \right)^2$

corresponds to the square of the half-wavenumber for Dirichlet conditions. The Hurdle Theorem for parallel flows states: if $W(y) > h$ for all $y \in [y_1, y_2]$, a neutral mode exists at $c=\alpha$ for some $k>0$, and a pair of unstable complex-conjugate modes emerges under perturbation, establishing inviscid instability (Deguchi et al., 25 Jan 2026, Deguchi et al., 2024).

The Rayleigh-quotient criterion asserts that finding a test function $\varphi$ with Rayleigh quotient $R_\alpha[\varphi] < -k^2$ guarantees the existence of a neutral root; a Sturm–Liouville argument and perturbation in $k$ demonstrate transition to instability.

3. Extension to Axisymmetric Annular and Pipe Flows

The hurdle theorem is generalized to axisymmetric flows in cylindrical coordinates. The base flow $U(r) e_z$ and disturbances with axial wavenumber $k$ and azimuthal number $n$ lead to the inviscid stability equation

$$\left\{ \frac{r}{N^2 + r^2}(rG)' \right\}' - k^2 G + \frac{Q'(r)}{U(r)-c} rG = 0$$

where $N = n/k$, $Q(r) = -r U'(r)/(N^2 + r^2)$. The generalized curvature

$W_{\alpha, N}(r) = \frac{r Q'(r)}{U(r) - \alpha}$

serves as the analog of $W(y)$.

For annular domains $\Omega = (r_i, r_o)$, the theorem asserts: if $W_{\alpha, N}(r) > h$ for all $r$ in $\Omega$, with

$$h = \frac{(\pi/\ln \eta)^2}{N^2 + r_i^2} + \frac{1}{N^2},\quad \eta = r_i / r_o$$

and criticality conditions on $Q'(r_c)=0, Q''(r_c)\neq 0, U'(r_c)\neq 0$, then a neutral axisymmetric mode exists, and a nearby branch is unstable for perturbed $k$ (Deguchi et al., 25 Jan 2026).

An analog applies to full pipe geometry, where the hurdle $h(r)$ is replaced with $C r^p$ using domain constants derived from geometric and spectral parameters.

4. Sturm–Liouville and Rayleigh-Quotient Methods

Underlying the hurdle theorem in all settings is the conversion of the linear stability or hitting-time problem to a Sturm–Liouville eigenvalue problem. The Rayleigh quotient

$$R_\alpha(\varphi) = \frac{ \int \left[ |\varphi'|^2 - W \varphi^2 \right] \, dy }{ \int \varphi^2 \, dy }$$

or its axisymmetric generalization is minimized over admissible test functions. Instability or positive hitting probability corresponds to $R_\alpha(\varphi) < -k^2$, i.e., a lowest eigenvalue below zero. The "hurdle" emerges from comparing with the first eigenvalue of a problem with constant potential over a subdomain, yielding the explicit $h$ barrier. The existence of a test function (often the first Dirichlet eigenfunction) for which the Rayleigh quotient crosses this hurdle suffices to guarantee instability or exceeding rare-event probabilities.

In probabilistic parallel-queue systems, similar spectral logic governs the construction of harmonic functions—built from single or conjugate points on the process's characteristic surface—used to approximate hitting probabilities.

5. Physical and Probabilistic Interpretation

In fluid mechanical applications, $W(y)$ plays the role of a reciprocal local Rossby–Mach number; subsonic regions (everywhere $M^{-1}>1$) correspond to instability, while local supersonic patches ($M^{-1}\leq 1$ somewhere) can yield stability, recovering the Kelvin–Arnol'd theorems in various limits (Deguchi et al., 25 Jan 2026, Deguchi et al., 2024). In the annular and pipe extensions, $W_{\alpha,N}(r)$ encodes inviscid centrifugal effects and geometric confinement.

In queueing theory, the diagonal $\partial B$ acts as a "hurdle," with the hitting probability encoding the chance of excessive backlog. The exponential bound for the relative error in the main theorem demonstrates that in the many-server or large-buffer limit, the rare-event behavior is controlled by the much-simpler reflected process hitting a linear barrier (Ünlü et al., 2018).

6. Relations to Broader Stability and Instability Criteria

The hurdle theorem both sharpens and unifies existing sufficient conditions for instability, notably the Kelvin–Arnol'd I/II theorems and the classical Rayleigh and Tollmien criteria. It applies in a broader class of settings, including non-monotonic base flows, stratified and quasi-geostrophic fluids, magnetohydrodynamics (when the eigenproblem reduces to appropriate Sturm–Liouville form), and even alternating jet flows in planetary atmospheres (Deguchi et al., 2024).

Its predictions have been shown, in both queueing and hydrodynamic contexts, to closely bracket numerically computed or exact boundaries for instability and rare-event onset. In practical computations, it provides nearly sharp, analytically tractable regimes separating stable from unstable (or high-probability from rare-event) zones.

7. Illustrative Examples and Applications

• Hydrodynamic flows: For the sinusoidal base profile $U(y)=\cos(2\pi y/L)$ with constant $M^{-1}=a$, the domain-wide hurdle $h=1$ recovers exact agreement between the threshold for instability in $a$ and numerically computed critical values (Deguchi et al., 2024).
• Parallel queue overflow: The critical regime closed-form for diagonal-hitting probability, and robust harmonic-function constructions in non-degenerate regimes, yield sharp asymptotics for overflow events in high-traffic parallel servers (Ünlü et al., 2018).
• Annular and pipe flows: Numerical comparison of the hurdle theorem's thresholds with full eigenvalue computations in model flows confirms that the theorem predicts the instability region with high accuracy; in particular, it outperforms older criteria under confinement or strong curvature (Deguchi et al., 25 Jan 2026).

Application Area	Instability Criterion	Reference
2D parallel shear flow	$W(y)>h$ for all $y$	(Deguchi et al., 25 Jan 2026)
Annular/pipe flows	$W_{\alpha,N}(r)>h(r)$ for all $r$	(Deguchi et al., 25 Jan 2026)
Parallel queues	Hitting $\partial B$ probability approximates $p_n(x)$	(Ünlü et al., 2018)
Alternating jets	$\sup_{y\in [y_1,y_2]} M_\alpha^{-1}(y)>h$ on a subdomain	(Deguchi et al., 2024)

The hurdle theorem thus serves as a cornerstone for modern analysis of instability and rare events in parallel flows of both physical and stochastic origin, connecting spectral theory, probability, and applied mechanics.