Noh Problem in Compressible Hydrodynamics

• The Noh problem is a self-similar stagnation-shock model where a uniform inward flow generates an outward-moving shock with an analytic reference solution.
• It serves as a benchmark to verify numerical methods in compressible hydrodynamics, exposing challenges like strong compression, wall heating, and shock tracking.
• Modern extensions include non-ideal equations of state, multidimensional geometries, and magnetohydrodynamic generalizations for realistic material modeling.

In compressible-flow hydrodynamics, the Noh problem is a self-similar stagnation-shock problem in which a uniform medium with constant inward velocity collapses toward a wall, axis, or center and generates an outward-moving accretion shock, leaving a stagnant shocked region behind it. It is a canonical verification problem because it combines an exact or semi-analytic reference solution with severe numerical stress: strong compression, geometric focusing, and the well-known wall-heating pathology in many shock-capturing schemes. Although the classical form is the ideal-gas problem with vanishing initial pressure and internal energy, the modern literature treats the Noh problem as a broader benchmark family encompassing non-ideal equations of state, multidimensional perturbation theory, adaptive and moving-mesh discretizations, and magnetohydrodynamic generalizations (Yorke et al., 2017).

1. Classical formulation and self-similar structure

A standard formulation uses the symmetry-reduced inviscid Euler equations

$$\frac{\partial \rho}{\partial t} + u \frac{\partial \rho}{\partial r} + \rho \left(\frac{\partial u}{\partial r} + \frac{mu}{r}\right) = 0,$$

$$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$

$$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$

with geometry parameter $m=0,1,2$ for planar, cylindrical, and spherical symmetry, respectively (Burnett et al., 2017). The classical Noh assumptions are a uniform initial state with constant inward velocity $u_0<0$, a shock launched from the wall/origin at $t=0$, and a stagnant post-shock state $u_2=0$.

Under the similarity variable

$\xi=\frac{ar}{t}, \qquad r_s(t)=Dt=\frac{t}{a},$

the solution separates into an unshocked outer region and a shocked inner region. For the classical Noh form,

$$\rho_1 = \rho_0 \left(1-\frac{u_0t}{r} \right)^m, \qquad p_1=p_0, \qquad u_1=u_0,$$

and

$$\rho_2 = \rho_0 \left(1-a u_0 \right)^{m+1}, \qquad p_2 = p_0 - \frac{\rho_0 u_0 }{a} \left(1-a u_0 \right), \qquad u_2=0,$$

with $$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$0 fixed by the energy jump condition (Burnett et al., 2017). In planar geometry, the preshock density remains constant; in cylindrical and spherical geometry it rises ahead of the shock because of geometric convergence.

For the classic ideal-gas strong-shock case, $$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$1 and $$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$2. Then cylindrical and spherical Noh flows retain the familiar closed form

$$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$3

The same structure appears in the stability analysis of the ideal-gas spherical and cylindrical problem, where the unperturbed shock radius is written

$$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$4

with a uniform, stagnant shocked core and cold converging inflow ahead of the shock (Velikovich et al., 2016).

This classical structure explains why the Noh problem is simultaneously simple and difficult: it is piecewise analytic, but it contains a strong shock, a stagnation region, and geometry-driven compression.

2. Equation-of-state generalizations

A persistent misconception is that the Noh problem is intrinsically an ideal-gas benchmark. The modern literature shows that the classical construction extends well beyond the $$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$5-law gas, although the extent of that extension depends strongly on geometry (Burnett et al., 2017).

The key non-ideal EOS generalizations studied in recent work are summarized below.

Extension	Defining relation	Scope
Stiff gas	$$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$6	Planar arbitrary-strength shocks; condensed-matter applications (Burnett et al., 2017)
Noble–Abel	$$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$7	Planar arbitrary strength; curvilinear strong-shock cases (Burnett et al., 2017)
Carnahan–Starling	$$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$8	Planar implicit quasi-analytic solutions (Burnett et al., 2017)
Arbitrary isentropic EOS	$$\frac{\partial u}{\partial t} + u \frac{\partial u}{\partial r} + \frac{1}{\rho} \frac{\partial p}{\partial r} = 0,$$9	Planar classical Noh with implicit shocked-density equation (Giron et al., 2019)
Black-box EOS	$$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$0 or $$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$1	Semi-analytic Newton solution of RH closure (Gerberding et al., 20 Sep 2025)

For arbitrary isentropic EOS $$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$2, the planar Noh problem remains self-similar, but the post-shock state is no longer generally available in closed form. Instead, the shocked density is determined implicitly by

$$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$3

with shock speed then recovered from

$$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$4

This result shows that exact solvability of the Noh structure does not imply a closed-form shocked state (Giron et al., 2019).

For condensed matter, the stiff-gas EOS has been especially important because it preserves much of the mathematical tractability of the ideal-gas case while introducing a realistic bulk modulus. In the planar condensed-matter extension, the EOS

$$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$5

admits exact planar Noh formulas for shock speed, shocked density, and pressure, and was calibrated to experimental principal Hugoniot data for Al, Fe, Cu, and W (Yorke et al., 2017).

A second misconception is that once one leaves the ideal gas, the Noh problem ceases to be useful for verification. The black-box EOS framework contradicts that view: by solving the Rankine–Hugoniot closure numerically with Newton iteration, semi-analytic Noh solutions can be constructed for stiffened gas, Noble–Abel, Carnahan–Starling, Steinberg/Mie–Grüneisen form, and tabulated SESAME/EOSPAC data (Gerberding et al., 20 Sep 2025).

3. Geometry restrictions and admissibility

The generalization beyond ideal gas is not geometry-neutral. For the classical two-region Noh structure, planar geometry is much less restrictive than cylindrical or spherical geometry.

The structural restriction follows from the unshocked-region energy equation. In the self-similar derivation,

$$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$6

Hence either $$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$7 or $$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$8. This means that planar geometry allows arbitrary initial pressure and finite-strength shocks, while cylindrical and spherical geometry require zero initial bulk modulus, equivalently zero initial sound speed, under the classical Noh assumptions (Burnett et al., 2017). The black-box formulation restates this for nonplanar classical Noh solutions as the requirement

$$\frac{\partial p}{\partial t} + u \frac{\partial p}{\partial r} + K \left(\frac{\partial u}{\partial r}+ \frac{mu}{r} \right) = 0,$$9

and notes that many realistic EOSs therefore admit only planar classical Noh solutions in that framework (Gerberding et al., 20 Sep 2025).

This restriction is central to interpreting the literature. It explains why planar Noh has become the main vehicle for finite-strength and non-ideal EOS studies, while cylindrical and spherical formulations are often retained in the classic strong-shock ideal-gas form. It also clarifies that “extension of the Noh problem” can mean different things: extension of the EOS class, extension of the admissible geometry, or extension of the similarity structure are not equivalent operations.

4. Verification benchmark and characteristic numerical pathologies

The Noh problem is attractive as a verification test because it has an exact analytical solution, is easy to set up conceptually, is numerically demanding, and exposes wall heating and shock smearing (Yorke et al., 2017). In the non-ideal EOS literature, it is used explicitly to verify shock location, density jump, pressure jump, and internal-energy jump against exact or quasi-exact reference states.

For the stiff-gas condensed-matter extension, 16 planar test cases were run with Los Alamos hydrocodes Flag and xRage for Al, Fe, Cu, and W. Both codes accurately predicted shock location and jump magnitudes, and both exhibited approximately first-order spatial convergence; the reported density-convergence fits were $m=0,1,2$0 for Flag and $m=0,1,2$1 for xRage (Yorke et al., 2017). In the broader non-ideal EOS study with FLAG, near first-order convergence was likewise observed across stiff gas, Noble–Abel, and Carnahan–Starling cases, while wall heating remained the dominant discrepancy and became more severe from planar to spherical geometry (Burnett et al., 2017).

The black-box EOS formulation extends this verification logic to production-style material models. In a stiffened-gas planar xRAGE test with $m=0,1,2$2, the semi-analytic closure produced

$m=0,1,2$3

and the numerical solution showed near first-order $m=0,1,2$4 convergence in density and pressure on four uniform meshes (Gerberding et al., 20 Sep 2025).

These studies establish the Noh problem as more than a pedagogical ideal-gas exercise. It is a reusable verification template in which the difficult part is the shock-state closure, not the PDE structure itself.

5. Behavior across numerical methods

Because the Noh problem is dominated by strong compression and a stagnant post-shock core, it is frequently used less as an accuracy test than as a robustness discriminator. In a one-dimensional comparison of high-order finite-volume methods, the 1D symmetric planar Noh setup

$m=0,1,2$5

caused the WENO5-AO-HLLC scheme to blow up, whereas WENO5-AO-GKS remained stable and produced a smaller central density dip than WENO5-AO-LF (Yang et al., 2020).

Several modern methods have used the Noh problem to target the wall-heating anomaly directly. In two dimensions, the WENO-$m=0,1,2$6-$m=0,1,2$7 formulation combines isotropic $m=0,1,2$8-method artificial viscosity with wavelet-based noise removal and was reported to eliminate the wall-heating phenomenon for the Noh problem; the stand-alone simplified WENO solver could not be run to $m=0,1,2$9 even with CFL $u_0<0$0 because noise in the cold gas caused loss of positivity of density (Ramani et al., 2018). In one-dimensional space-time DG, the MDG+ICE method treated geometry as an unknown and enforced interface conservation explicitly; for the 1D Noh problem on four quadrilateral space-time cells, the computed MDG(P1) solution was reported to be virtually identical to the exact solution and to exhibit no anomalous overheating (Luo et al., 2021).

Moving-mesh formulations use the Noh problem to test whether mesh motion improves symmetry preservation and shock tracking. A three-dimensional ALE compact gas-kinetic scheme on the Cartesian octant $u_0<0$1 used the spherically symmetric Noh inflow $u_0<0$2 with $u_0<0$3, $u_0<0$4, and compared radial density against the exact profile

$u_0<0$5

Good agreement was reported, with finer moving meshes yielding less oscillation (Zhang et al., 10 Feb 2026). In a separate SAM-ALE study of the 2D Noh implosion, a $u_0<0$6 adaptive run achieved $u_0<0$7 density error $u_0<0$8, compared with $u_0<0$9 for a $t=0$0 uniform run, while using 45.6 s versus 289 s CPU time and 7.8 MB versus 43.7 MB memory (Ramani et al., 2022).

A plausible implication is that the Noh problem has evolved into a solver-class discriminator: fixed-grid shock-capturing, moving-mesh ALE, interface-fitted DG, and kinetic-flux methods can all be compared against a benchmark whose main difficulty is not smooth accuracy but controlled treatment of extreme compression.

6. Perturbation theory, MHD extensions, and benchmark scope

The classical ideal-gas Noh solution is not merely a 1D verification profile; it also supports a nontrivial multidimensional perturbation theory. For spherical geometry and cylindrical $t=0$1 filamentation modes, linear perturbations scale as

$t=0$2

not exponentially. The exact theory provides a dispersion equation for $t=0$3, hypergeometric eigenfunctions, and shows that

$t=0$4

for all modes and all adiabatic indices $t=0$5: perturbations decay as power laws, either oscillatory or monotonic (Velikovich et al., 2016). This result matters for verification because it distinguishes physical decay from numerically induced perturbations on Cartesian grids.

The benchmark family has also been extended to ideal MHD. The “Mag Noh problem” generalizes the classic cylindrical Noh flow to an axisymmetric rotating magnetized plasma with azimuthal field $t=0$6, axial field $t=0$7, and rotation $t=0$8, retaining a constant-speed outward shock in self-similar form (Beresnyak et al., 2021). The resulting family is five-parametric, includes both stable and unstable regions in parameter space, and even admits a zero-initial-velocity branch in which magnetic tension, rather than imposed implosion velocity, launches the shock at $t=0$9 (Beresnyak et al., 2021). That extension makes explicit that the Noh problem is better understood as a paradigm—cold converging flow, self-similar stagnation, and an expanding accretion shock—than as a single ideal-gas formula.

This broader view also clarifies current limits. Some papers use Noh chiefly as a robustness test and provide only qualitative profile comparisons; others derive exact or semi-analytic closures suitable for formal verification. Some extensions preserve closed form, others reduce to transcendental or Newton-solved algebraic systems. The unifying feature is the same: the Noh problem supplies a controlled setting in which shock formation, stagnation, and numerical pathology can be studied against a known self-similar structure.