A blow-up confined to one light-cone vertex
The paper studies the defocusing quintic wave equation ∂t²u − Δu + u⁵ = 0 for real-valued functions on R^(1+10), a setting that is energy-supercritical: the critical regularity is s_c = 9/2, well above the energy level. The central object is a discretely self-similar solution built inside a backward light cone K_T = {(t,x) : |x| ≤ T − t}, with the form u(t,x) = (T−t)^(−1/2) W*(ω* log(1/(T−t)), |x|/(T−t)). The profile W* is nonzero, 2π-periodic in its first variable, even in its second, and real-analytic up to and across the cone boundary ρ = 1.
Cutting off the initial data of this self-similar solution produces smooth, compactly supported, radial data whose maximal smooth solution has lifespan exactly T. That solution coincides with the self-similar one inside K_T, stays C^∞ on ([0,T] × R^10) except at the single vertex (T,0), and is bounded on [0,T) × {|x| ≥ ε} for every ε > 0. The singularity is therefore not a spreading front but a point: the singular set at time T is exactly {0}.
Why real-valued solutions were the open case
Finite-time blow-up for defocusing supercritical wave equations had already been proved for complex-valued solutions, by adapting the front-compression mechanism used for defocusing energy-supercritical nonlinear Schrödinger equations. For real-valued solutions of defocusing wave equations, the question had remained open. The authors state that the solutions constructed here appear to be the first real-valued blow-up solutions of defocusing energy-supercritical NLW.
The distinction is not cosmetic. The profile W* is genuinely time-dependent: the authors were unable to find real self-similar (τ-independent) profiles of the equation, and for s_c ≤ 3/2 the maximum principle rules them out. The real profile is instead reached by numerically continuing rotating self-similar profiles of the complex-valued equation, which is why the real and complex settings connect even though the final object is real.
The blow-up rate is (T−t)^(−1/2), not the energy
Inside the cone, u* blows up at the self-similar rate ‖u*(t)‖_{L∞(|x| ≤ T−t)} ≈ (T−t)^(−1/2). The full solution u of the main theorem satisfies ‖u(t)‖_{L∞(R^10)} ≳ (T−t)^(−1/2) for 0 ≤ t < T. The profile bound validated in the paper gives 2.26 ≤ max_{0 ≤ ρ ≤ 1} |W*(θ,ρ)| ≤ 8.25 for all θ, which is what converts the profile estimate into the stated blow-up rate.
Two other quantities behave differently and should not be conflated with the L∞ rate. The critical norm of u*(t) inside the cone is a bounded periodic function of log(T−t), and the energy inside the cone tends to zero like (T−t)^{2(s_c−1)} = (T−t)^7. So the L∞ norm diverges while the critical norm stays bounded and the cone energy vanishes; these are different measures with different denominators and different behavior.
The profile equation degenerates at the light cone
The profile W* solves a degenerate hyperbolic equation whose principal part degenerates at the light cone ρ = 1. Real analyticity of W* across ρ = 1 is essential to the construction. For each Fourier mode in the self-similar time, the equation admits a singular branch behaving like (1−ρ)^{4−iω}, and smoothness excludes these branches.
The linear analysis makes this structure explicit. Writing the profile in Fourier modes V_n(ρ) with n odd, the mode operators L_n reduce, under V(ρ) = Ṽ(ρ²), to hypergeometric operators H_n in z = ρ². In the Chebyshev basis T_k(2ρ²−1), the resulting matrix Λ_n is upper triangular with nonzero diagonal entries, so each finite section is invertible and Λ_n has an upper triangular inverse whose columns have finitely many nonzero entries. The paper's analytic ingredients are this exact mode-operator description and a bound on the inverses that is uniform at high frequencies.
Existence is proved by a computer-assisted Newton–Kantorovich argument
Existence of the real profile is proved by a computer-assisted Newton–Kantorovich argument in a Banach algebra of Fourier–Chebyshev coefficients, with rigorous error control. The profile is expanded in Fourier series in the self-similar time and Chebyshev series in the radial variable; the profile equation becomes a zero-finding problem in a weighted ℓ¹ space of coefficients, which is a Banach algebra under the relevant convolution. A Newton–Kantorovich argument around a numerical approximation then yields an exact solution nearby.
The quantitative output is tight. With ν = 1.3, ν_θ = 1.01, N = (65,200), and r_0 = 7.216·10^(−11), the theorem produces a real solution Ũ = (ω̃, W̃) with |ω̃ − ω̄| ≤ 4.05·10^(−10) and ‖W̃ − W̄‖_{ℓ¹} ≤ 8.42·10^(−10). The approximate frequency satisfies |ω̄ − 10.150514753468228| ≤ 3.4·10^(−16), and the main theorem records |ω* − 10.15051475347| ≤ 4.1·10^(−10). All computations use interval arithmetic, and the codes and approximate objects are available on GitHub.
The numerical continuation is a guide, not part of the proof
The approximate profile is found by numerically continuing rotating self-similar profiles of the complex-valued equation. The authors explicitly state that this step is not essential for the blow-up proof and is left as non-rigorous numerics. The rigorous part is the Newton–Kantorovich verification, which takes the numerical approximation as input and produces a nearby exact solution with controlled error.
The continuation path is specific. The authors start from the rotating profile with (d,p) = (8,5) and ω = 14.8503, computed from the ordinary differential equation for Q, and follow the branch with 6 odd harmonics and 40 Chebyshev modes. The end point ε = p, refined with 16 harmonics and 64 Chebyshev modes, is a numerical approximation to a real DSS profile for (d,p) = (8,5) with ω = 14.3108, checked independently by time evolution in self-similar variables over one period. Continuation in d at p = 5 then gives a collocation profile at d = 10, refined by Newton's method into the approximate solution used in the rigorous step.
The proof covers (d,p) = (10,5), not the full supercritical range
The analytic framework of the paper does not depend on the specific choice (d,p) = (10,5). For ∂t²u − Δu + u^p = 0 on R^(1+d) with d ≥ 3, p ≥ 3 odd, and α = 2/(p−1), the same computation gives L_n = −4H_n with H_n = z(1−z)∂z² + (c − (a_n + b_n + 1)z)∂z − a_n b_n, where c = d/2, a_n = (α + i s_n)/2, b_n = (α + 1 + i s_n)/2, and γ'_n − γ_n = d − 1. The singular exponent at z = 1 is c − a_n − b_n = s_c − 1/2 − i s_n.
However, the computer-assisted existence proof is carried out only for (d,p) = (10,5). The continuation in d and p also produces approximate real DSS profiles for other exponents, for instance (d,p) = (8,7), (7,5), (6,9), and (5,13), but these are numerical approximations, not verified existence results. The authors also note that it is not clear whether the same mechanism can prove blow-up in the full supercritical range s_c > 1, or only for sufficiently supercritical problems, and that stability of the blow-up up to finite codimension remains a natural open question.
Sources used in this answer
Finite-time blow-up for the real-valued defocusing energy-supercritical quintic NLW
Constructs a real-valued discretely self-similar solution of the defocusing quintic wave equation in R^10 that blows up at the vertex of a backward light cone at rate (T−t)^(−1/2), giving the first real-valued blow-up solutions for defocusing energy-supercritical NLW.