
==== Front
Commun Math Phys
Commun Math Phys
Communications in Mathematical Physics
0010-3616
1432-0916
Springer Berlin Heidelberg Berlin/Heidelberg

5065
10.1007/s00220-024-05065-9
Article
Non-uniqueness of Admissible Solutions for the 2D Euler Equation with Lp Vortex Data
http://orcid.org/0009-0009-9241-7105
Mengual Francisco fmengual@mis.mpg.de

https://ror.org/00ez2he07 grid.419532.8 0000 0004 0491 7940 Max Planck Institute for Mathematics in the Sciences, 04103 Leipzig, Germany
Communicated by A. Ionescu

20 8 2024
20 8 2024
2024
405 9 2072 6 2023
14 6 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by/4.0/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
For any 2<p<∞ we prove that there exists an initial velocity field v∘∈L2 with vorticity ω∘∈L1∩Lp for which there are infinitely many bounded admissible solutions v∈CtL2 to the 2D Euler equation. This shows sharpness of the weak–strong uniqueness principle, as well as sharpness of Yudovich’s proof of uniqueness in the class of bounded admissible solutions. The initial data are truncated power-law vortices. The construction is based on finding a suitable self-similar subsolution and then applying the convex integration method. In addition, we extend it for 1<p<∞ and show that the energy dissipation rate of the subsolution vanishes at t=0 if and only if p≥32, which is the Onsager critical exponent in terms of Lp control on vorticity in 2D.

http://dx.doi.org/10.13039/100005235 Institute for Advanced Study Universidad de SevillaPostdoc contract Mengual Francisco http://dx.doi.org/10.13039/501100013296 Max-Planck-Institut für Mathematik in den Naturwissenschaften Postdoc contract Mengual Francisco http://dx.doi.org/10.13039/501100004837 Ministerio de Ciencia e Innovación PID2020-114703GB-I00 PID2021-123839OB-I00 MTM2017-85934-C3-2-P EUR2020-112271 http://dx.doi.org/10.13039/100019180 HORIZON EUROPE European Research Council ERC Advanced Grant 834728 Max Planck Institute for Mathematics in the Sciences (2)Open Access funding enabled and organized by Projekt DEAL.

issue-copyright-statement© Springer-Verlag GmbH Germany, part of Springer Nature 2024
==== Body
pmcIntroduction and Main Results

We consider the Cauchy problem for the Euler equation 1a ∂tv+div(v⊗v)+∇p=0,

1b divv=0,

1c v|t=0=v∘,

posed on the domain [0,T]×R2, where p(t, x) is the pressure, v(t, x) is the velocity field, and v∘(x) is the initial datum. In this work we are interested in non-uniqueness of weak (i.e. distributional) solutions to the Euler equation. More precisely, we address the question of what is the threshold regularity at t=0 for which uniqueness of bounded admissible solutions fails. A weak solution v∈Lt∞L2 to the Euler equation is called admissible if it does not increase the (kinematic) energy E:=12‖v‖L222 E(t)≤E(0)for a.e.t∈[0,T].

This admissibility criterion is based on considering weakly convergent sequences of Leray solutions of Navier–Stokes with vanishing viscosity (see e.g. [26]). Before going further, let us present our main result:

Theorem 1.1

For any 2<p<∞ there exists a steady solution vs∈L2 with vorticity ωs∈L1∩Lp to the Euler equation, with the property that there are infinitely many bounded admissible solutions v∈CtL2 to the Euler equation (1) starting from v∘=vs. Moreover, vs is compactly supported and smooth away from the origin.

In the next sections we explain how Theorem 1.1 shows sharpness of the weak–strong uniqueness principle, as well as sharpness of Yudovich’s proof of uniqueness in the class of bounded admissible solutions.

Sharpness of the Weak–Strong Uniqueness Principle

Admissible solutions coincide with strong solutions as long as the latter exist: Suppose v∘ admits a strong solution vs∈C1, and let v be another admissible solution. A straightforward computation shows that the relative energy Erel:=12‖v-vs‖L22 can be bounded by3 Erel(t)≤∫0t∫R2|∇vs||v-vs|2dxdτ.

This estimate combined with the Grönwall inequality allows to conclude that necessarily Erel=0 (v=vs). Indeed, it is enough to assume that ∇vs∈Lt1L∞. This fact is known in the literature as the weak–strong uniqueness principle (see e.g. [74]).

Theorem 1.2

(Weak–strong uniqueness principle) Suppose there exists a strong solution vs∈C1 to the Euler equation. Then, it is unique within the class of admissible solutions.

By the Sobolev embedding W1,p⊂C1-2p for 2<p<∞, Theorem 1.1 implies the following result, which shows sharpness of Theorem 1.2 for Hölder spaces. More precisely, it states that if the C1 assumption is weakened at a single point by Cγ for some 0<γ<1, then uniqueness fails in the class of admissible solutions. As a by-product, it shows existence of wild data above the Onsager critical exponent γ=13 (see Sect. 2.2).

Theorem 1.3

For any 0<γ<1 there exists a steady solution vs∈Cγ to the Euler equation, with the property that there are infinitely many admissible solutions v∈CtL2 to the Euler equation (1) starting from v∘=vs. Moreover, vs is compactly supported and smooth away from the origin.

Notice that Theorem 1.2 concerns uniqueness, while existence is just an assumption. Above C1 regularity, Wolibner [75] and Hölder [49] proved global well-posedness of the 2D Euler equation in C1,γ for any γ>0 (assuming suitable decay as |x|→∞). In this class, uniqueness follows immediately from Theorem 1.2. The proof of global existence exploits the fact that the vorticity ω=rotv is transported by the flow 4a ∂tω+div(vω)=0,

4b v=∇⊥Δ-1ω,

4c ω|t=0=ω∘,

where ω∘=rotv∘, and (4b) is the Biot-Savart law5 v(x)∗=12πi∫R2ω(y)x-ydy.

In (5) we identify R2 with the complex plane C as usual, where i denotes the imaginary unit and ∗ the complex conjugate. It is well known from Harmonic analysis that the map ω↦∇⊥Δ-1ω=v is continuous from Cγ to C1,γ. For C1 vector fields v, the trajectory map X of the flow is well defined by the Cauchy-Lipschitz theory applied to 6a ∂tX=v(t,X),

6b X|t=0=id.

Thus, the Euler equation (4) can be written as ω(t,X)=ω∘(x), where ω and X are related implicitly through v. The rigorous proof of global existence in C1,γ is carried out by a Schauder fixed-point argument. This result is in stark contrast to the 3D case, where Elgindi [31] proved formation of finite-time singularities due to vortex stretching (local well-posedness was known since Lichtenstein [55] and Gunther [47]).

In the borderline case C1, Bourgain and Li [4] and latter Elgindi and Masmoudi [32] proved strong ill-posedness for the Euler equation (see [20] for strong ill-posedness in Hβ). The reason behind is that the map ω↦v sends bounded vorticities to log-Lipschitz velocities. Remarkably, Yudovich [76] showed that the log-Lipschitz modulus of continuity is still valid to define uniquely the Lagrangian map X and prove global well-posedness (see also [56, Chapter 8]). This fact makes the class of bounded vorticities a natural space for the 2D Euler equation. In order to motivate Theorem 1.1 it is convenient to recall Yudovich’s proof of uniqueness, which can be understood as a refinement of the proof of the weak–strong uniqueness principle.

Sharpness of Yudovich’s Proof of Uniqueness

We start by recalling two classical estimates of the Biot-Savart operator (5). The first one is the boundedness of the map ω↦v from L1∩Lp to L∞ for any 2<p≤∞. This follows by splitting R2 into |x-y|≥1 & |x-y|<1, and then applying the Hölder inequality7 ‖v‖L∞≤‖ω‖L1+p-1p-21-1p‖ω‖Lp.

The second estimate is the Lp-boundedness of the map ω↦∇v for any 1<p<∞. Notice that (5) is the Cauchy transform, and thus ∇v can be written in terms of the Beurling transform of ω, a 2D version of the Hilbert transform (see e.g. [3]). Then, it follows from the Calderon-Zygmund theory that8 ‖∇v‖Lp≤Cp2p-1‖ω‖Lp.

We will use C to denote a constant, which may change from line to line but will be universal. While the bound (8) gives the exact growth as p→1,∞, here we just need the inequality ‖∇v‖Lp≤Cp‖ω‖Lp for 2<p<∞.

Next, we recall Yudovich’s energy method. Let vs∈CtL2 with ωs∈Lt∞(L1∩L∞) be a (Yudovich) solution, and let v be another bounded admissible solution with v∘=vs∘. By applying the Hölder inequality and (8), the right hand side of (3) can be bounded by9 ∫R2|∇vs||v-vs|2dx≤Cp‖ωs‖Lp‖v-vs‖L∞2pErel1-1p,

for any 2<p<∞, which plugged into (3) implies that10 Erel(t)≤‖v-vs‖Lt,x∞2(C‖ωs‖Lpt)p.

We recall that ‖ωs‖Lp is independent of time because ωs(t,X)=ωs∘(x) with X volume-preserving by divvs=0. On the one hand, the term ‖v-vs‖Lt,x∞ can be bounded by the Lt,x∞-norm of v and vs separately: the first is bounded by hypothesis, and the latter by (7) for p=∞. On the other hand, by the log-convexity of the Lp-norms we have11 ‖ωs‖Lp≤max{‖ωs‖L1,‖ωs‖L∞}<∞.

Finally, by letting p→∞ in (10), it follows that necessarily Erel=0 (v=vs).

Theorem 1.4

(Yudovich’s well-posedness Theorem) Let v∘∈L2 with ω∘∈L1∩L∞ and divv∘=0. Then, there exists a global solution v∈CtL2 with ω=Lt∞(L1∩L∞) to the Euler equation. Furthermore, it is unique within the class of bounded admissible solutions.

Theorem, 1.1 shows sharpness of Yudovich’s proof of uniqueness for Lp spaces. More precisely, it states that if the L∞ assumption is weakened at a single point by Lp for some 2<p<∞, then uniqueness fails in the class of bounded admissible solutions.

Let us point out where Yudovich’s proof of uniqueness is not working in Theorem 1.1. Notice that the term ‖v-vs‖Lt,x∞ can still be bounded by the Lt,x∞-norm of v and vs separately: the first is bounded by hypothesis, and the latter by (7) for 2<p<∞. However, now the condition (11) fails, namely we have ‖ωs‖Lq=∞ for q>p, which prevents from concluding Erel=0 via (10).

We finish the intro with several remarks, as well as a sketch of the proof of Theorem 1.1. (i) Our solutions have compact support. Hence, the same results hold in any arbitrary open subset of R2, as well as in the periodic domain T2.

(ii) Theorem 1.1 is a corollary of Theorem 1.5. We have chosen to introduce them separately for clarity of presentation. In fact, we take the same vs in all the theorems. They are truncated power-law vortices 12 vs(x)=χ(|x|)|x|-αx⊥,

where χ is a smooth cutoff function, and 0<α<1 is a parameter. With this choice we have vs∈C1-α and ωs∈L2α-. The time of existence depends on the truncation, and it can be made arbitrarily large (see (63)).

(iii) Our solutions v∈CtL2 are obtained via convex integration. They equal vs outside a disc of radius (ct)1α, where 0<c≤Cα are constants that will be specified in Sect. 5. As a result, v|t=0=vs and they are smooth outside {|x|≤(ct)1α}. Inside this region we only know that the vorticity is a distribution. The question of non-uniqueness of vorticities in Lt∞(L1∩Lp) remains open (see Sect. 2.1). In spite of the lack of uniqueness and regularity, these velocities are close in average to a smooth radially symmetric subsolution 13 v¯(t,x)=χ(|x|)h(t,|x|)|x|x⊥,

where h is a self-similar profile 14 h(t,r)=(ct)1-ααH(ξ),ξ=r(ct)1α.

We will declare H(ξ)=ξ1-α for ξ≥1, or equivalently v¯=vs for |x|≥(ct)1α. Our central task will be therefore to find a suitable profile H on [0, 1].

(iv) The construction explained in (iii) is also valid for 1≤α<2. Thus, Theorem 1.1 holds for 1<p≤2, but removing the property “bounded”. The borderline case α→2 corresponds to a point vortex. This will be analyzed in Sect. 6.

(v) We can impose that our solutions conserve the energy. The energy dissipation rate of the subsolution vanishes at t=0 if and only if α<43. This corresponds to the Onsager critical exponent p=32 in terms of Lp control on vorticity in 2D (see Sect. 3).

Sketch of the Proof

The proof of Theorem 1.1 relies on devising a suitable subsolution, a solution to the relaxed Euler equation, and then applying the convex integration method [26].

We start by rewriting the relaxed Euler equation, also known as the Euler-Reynolds equation, for radially symmetric (13) self-similar (14) subsolutions. Remarkably, the use of complex coordinates simplifies the computations and the choice of the Reynolds stress.

Secondly, under these ansatzes, we show that an admissible subsolution exists provided that the profile H satisfies the following conditions15 (4-α)∫01ξ2Hdξ=1,

16 2(2-α)∫01ξH2dξ<1.

The first condition (15) is necessary to guarantee that the Reynolds stress remains concentrated on {|x|≤(ct)1α}. The second condition (16) arises from imposing the admissibility.

Remark 1.1

It is straightforward to check that the above conditions are consistent with α>0 (equiv. p<∞). Namely, by applying the Hölder inequality, we get17 1(4-α)2=∫01ξ2Hdξ2≤∫01ξ3dξ∫01ξH2dξ<18(2-α),

which is satisfied for α>0. In the limiting case α=0 (p=∞) the equality in (17) holds if and only if H(ξ)=ξ. In this case, the subsolution agrees with the steady solution, v¯=vs, and thus [26] can only yield non-admissible solutions.

Thirdly, we construct a profile H. In particular, we make the ansatzH(ξ)=(1-alogξ)ξ1+b,

and we show that it is possible to take the parameters a,b>0 satisfying the conditions (15)(16). In addition, we show that there exits a unique c>0 maximizing the energy dissipation rate of the subsolution (Figs. 1, 2, 3).

Finally, after truncating our solutions to guarantee that the energy is finite, we obtain the following result, from which we deduce Theorem 1.1 as a corollary by applying [26].

Theorem 1.5

There exists a subsolution to the Euler equation which agrees with (12) outside {|x|≤(ct)1α}. Furthermore, the energy dissipation rate equals∂tE¯=-π16c8-α2αt4-3ααwithc=2α4-α2.

In particular, ∂tE¯|t=0=0 if and only if α<43, or equivalently ω∘∈L32.

Fig. 1 From lighter to darker blue, plot of the energy E¯(t) of the subsolution (v¯,σ¯,q¯) for the powers α=k3 with k=1,2,3,4,5. The initial energy E(0) is taken independently of α. For α=13,23 the energy is almost constant. The power α=1 corresponds to the threshold for bounded velocities. The power α=43 corresponds both to the threshold for L32 vorticities and ∂tE¯|t=0=0. For α=53 the energy decreases faster

Fig. 2 From lighter to darker blue, plot of the boundary (ct)1α for the powers α=k3 with k=1,2,3,4,5. This shows how the region {|x|≤(ct)1α} shrinks as α→0 (equiv. p→∞)

Organization of the Paper

In Sect. 2 we review the literature and compare it with the present work. In Sect. 3 we give a heuristic explanation of Theorem 1.5. In Sect. 4 we write the relaxed Euler equation (23) for radially symmetric self-similar subsolutions. In Sect. 5 we derive the conditions under which an admissible subsolution exists, and we construct an example. Then, we prove Theorems 1.1 & 1.5. Finally, we analyze the borderline case α→2 in Sect. 6.

Brief Background

In this section we review the literature on non-uniqueness for the Euler equation and compare it with the present work.

Self-similarity and Symmetry Breakdown

The global existence of solutions was extended to the vorticity class L1∩Lp for 1<p≤∞ by DiPerna and Majda [29] (see also [56, Chapter 10]). In contrast to Yudovich well-posedness Theorem, uniqueness is not expected to hold for p<∞. Roughly speaking, the Sobolev embedding implies continuity of the velocity field, and thus existence of the Lagrangian map X by the Peano Theorem, but the modulus of continuity does not satisfy the Osgood uniqueness criterion. Let us formulate this question as a conjecture, which remains open to the best of our knowledge.

Conjucture 2.1

For any 2<p<∞ there exists v∘∈L2 with ω∘∈L1∩Lp, with the property that there is more than one weak solution v∈CtL2 with ω∈Lt∞(L1∩Lp) to the Euler equation (4).

In the recent groundbreaking work [71, 72] Vishik solved Conjecture 2.1 for the forced Euler equation (see also the book [2]) 18a ∂tω+div(vω)=f,

18b v=∇⊥Δ-1ω,

18c ω|t=0=ω∘.

Theorem 2.1

(Vishik’s non-uniqueness Theorem) For any 2<p<∞ there exists v∘∈L2 with ω∘∈L1∩Lp and a force f=rotg∈Lt1(L1∩Lp) with g∈Lt1L2, with the property that there are infinitely many weak solutions v∈CtL2 with ω∈Lt∞(L1∩Lp) to the forced Euler equation (18).

Remarkably, Albritton, Brué, and Colombo [1] proved recently non-uniqueness of Leray solutions for the forced 3D Navier-Stokes equation by adapting properly Vishik’s construction into the cross section of an axisymmetric vortex ring. This approach is framed within the program of Jia, Šverák, and Guillod [46, 53] on the conjectural non-uniqueness of Leray solutions of the (unforced) 3D Navier–Stokes equation.

In the recent investigation [5, 6] Bressan, Murray, and Shen showed numerical evidence toward the validity of Conjecture 2.1. Their work is also based on self-similarity and symmetry breakdown. Their approach suggests two different ways of regularizing a non-radially symmetric version of ωs, leading to either one or two algebraic spirals. In contrast to Vishik’s spectral analysis, their construction relies on a smart system of adapted coordinates due to Elling [33, 34] (see also the recent work of García and Gómez-Serrano for gSQG [39]).

Relation Between Theorem 1.1 and Vishik’s Non-uniqueness Theorem

Theorem 1.1 has certain connections with Vishik’s non-uniqueness Theorem. Firstly, both theorems share the same initial data (12). Secondly, Vishik’s construction is also based on finding a suitable self-similar velocity v¯. Furthermore, v¯ is a modification of the power-law vortex β|x|-αx⊥ in a disc of radius 2t1α, where β is a sufficiently large constant. Thirdly, v¯ is also truncated by χ to guarantee integrability at infinity.

In spite of these similarities, both the results and the proofs differ significantly. Concerning the results, Vishik’s non-uniqueness Theorem solves Conjecture 2.1 in the natural vorticity class L1∩Lp by introducing a force, while Theorem 1.1 shows non-uniqueness without forcing by considering distributional vorticities. Our original motivation was indeed to explore the possibility of removing the force in Theorem 2.1 by means of convex integration, but paying the price of the low-regularity inherent to these constructions. The first obvious attempt was to absorb Vishik’s force into the Reynolds stress. However, it was not immediate for us that the corresponding subsolution was admissible. More precisely, we first needed to derive the conditions under which a radially symmetric self-similar subsolution yields admissible solutions via convex integration. After this, instead of checking if Vishik’s vortex satisfies these conditions, it becomes easier to construct our own profile H.

Convex Integration

As we mentioned, in this work we deal with a weaker version of Conjecture 2.1: the integrability condition ω∈Lt∞(L1∩Lp) is removed (for t>0) and then necessarily (4) is replaced by (1). The first result in this direction is due to Scheffer [65]: there exist Euler velocities v∈Lt,x2 with compact support in space-time. Latter, this construction was simplified by Shnirelman [66]. In the seminal work [25] De Lellis and Székelyhidi proved the same result in the energy space Lt∞L2, and for any space dimension d≥2.

Observe that these solutions show non-uniqueness for the trivial initial datum v∘=0. Non-uniqueness in Lt∞L2 was generalized by Wiedemann [73] for every divergence-free v∘∈L2 (see [54] for recent improvements of the regularity). For smooth initial data, the aforementioned solutions necessarily increase the energy, as a consequence of the weak–strong uniqueness principle. In this sense, a divergence-free v∘∈L2 is called wild if it admits infinitely many admissible solutions v∈Lt∞L2.

In [26] De Lellis and Székelyhidi initiated the investigation on non-uniqueness of admissible solutions, upon which this work is based. Recall that the property “admissible” includes both conservative (E=E(0)) and dissipative (E<E(0)) solutions. In his famous work [64] Onsager conjectured, in the context of the zeroth law of turbulence, the threshold regularity for the validity of the energy conservation of weak solutions to the Euler equation (in T3). Onsager’s conjecture, which is nowadays a theorem, can be stated as follows: Any weak solution v∈CtCγ to the Euler equation with γ>13 conserves the energy.

For any 0<γ<13 there exist weak solutions v∈CtCγ to the Euler equation which do not conserve the energy.

Part ((a)) was fully proved by Constantin, E, and Titi [17], after a partial result of Eyink [35]. Part (b) was solved more recently by Isett [50], and by Buckmaster, De Lellis, Székelyhidi, and Vicol [9]. The last achievement took a decade of refinements of the convex integration method, and the study of its connection with turbulent flows is still an active research area (see e.g. the recent works [44, 63]). Recently, the dissipative part of the 2D Onsager conjecture was solved by Giri and Radu [45]. The convex integration method has been applied to other 2D active scalar equations (see e.g. [22, 52, 67]) as well as the transport equation (see e.g. [59–61]).

Unfortunately, it seems not possible with the current convex integration techniques to construct solutions, neither with v∈Cγ for 13≤γ<1, nor with ω∈Lp for p≥1 in two dimensions. In [7] Brué and Colombo constructed a Cauchy sequence ωk in the Lorentz space L1,∞, whose velocities vk converge to an anomalous weak solution v. In [8] Buck and Modena adapted the previous construction for the Hardy space Hp for 23<p<1. This space is also weaker than L1 but, in contrast to L1,∞, it already embeds into the space of distributions.

Relation Between Theorem 1.3 and Onsager’s Conjecture

Coming back to the initial value problem, Theorem 1.3 gives presumably the first example of wild data with Hölder regularity above the Onsager critical exponent 13≤γ<1 (see [10] for a convex integration construction in C12- for the forced 3D Euler equation). Below the Onsager critical exponent 0<γ<13, Daneri, Runa, and Székelyhidi [23] proved that the set of wild data v∘∈Cγ is a dense subset of the divergence-free vector fields in L2 (see also [28]). However, to the best of our knowledge, it was not known neither concrete examples of wild data in Cγ, nor even the size of their set of singular points (see [27] for estimates of the singular set of times). In this regard, Theorem 1.3 provides an explicit example with a minimal singular set: a single point {0}. We remark that, although the aforementioned literature on the Onsager conjecture (b) is posed on the 3D periodic domain T3, our solutions can be trivially adjust to this setting. However, in this case the singular set becomes a line {0}×T. On this matter, it would be interesting to explore if there might be other 3D wild data with smaller singular sets.

Remarkably, Székelyhidi [68] provided the first concrete example of a wild datum for the unstable vortex sheet v∘(x)=sgn(x2), which corresponds to the limiting case p→1. This initiated a promising program on modeling hydrodynamical instabilities via convex integration: see [58, 68] for the Kelvin-Helmholtz, [40, 42, 43] for the Rayleigh-Taylor, and [11–13, 19, 37, 48, 57, 62, 69] for the Saffman-Taylor instabilities.

Yudovich Spaces

In [77] Yudovich extended his uniqueness result for unbounded vorticities for which ‖ωs‖Lp has moderate growth as p→∞. Let us recall Taniuchi’s (non-localized) version [70] of this generalization (see also [14]): Given a non-decreasing function Θ:[1,∞)→[1,∞), a vorticity ω belongs to the Yudovich space YΘ if‖ω‖YΘ:=supp∈[1,∞)‖ω‖LpΘ(p)<∞.

Then, the Euler equation (4) is globally well-posed in YΘ if Θ satisfies the Osgood type condition19 ∫3∞dppΘ(p)=∞.

Notice that Theorem 1.4 corresponds to the particular case Y1=L1∩L∞. In this regard, it would be interesting to explore if Theorem 1.1 could be extended to Yudovich spaces for which (19) fails.

Global existence has been proved in other vorticity classes (see e.g. the recent work of Crippa and Stefani [21] and the references therein). In [70] Taniuchi extended it to the Yudovich space YΘ for which the Osgood condition (19) is weakened by20 ∫3∞dppΘ(logp)=∞.

In this regard, Conjecture 2.1 could be also stated for Yudovich spaces satisfying the existence condition (20) but not the uniqueness condition (19).

Energy Conservation/Dissipation

In this section we discuss the energy conservation/dissipation of weak solutions to the Euler equation in terms of the vorticity. For simplicity of presentation we consider the periodic domain. The same results hold in the euclidean space mutatis mutandis.

We start by recalling that Onsager’s conjecture (a) is a corollary of the following Besov type criterion (see e.g. [15, 30]): Any weak solution v to the Euler equation in Td satisfying21 lim|y|→0∫0T∫Td|v(t,x+y)-v(t,x)|3|y|dxdt=0,

conserves the energy. Although this condition is independent of the dimension d≥2, it has stronger implications in the 2D case. As a first easy consequence, by the Sobolev embedding W1,p⊂C1-2p and (a), it follows that the energy is conserved for any weak solution with Lp control on vorticity for p>3. However, the energy conservation can be extended for smaller p’s by taking more advantage of (21). Let us recall the argument from [30, Proposition 6]. Firstly, by applying the Hölder inequality to δyv=v(x+y)-v(x), we get‖δyv‖L3≤‖δyv‖Lpθ‖δyv‖Lq1-θwith13=θp+1-θq.

Secondly, by using the Sobolev embedding W1,p⊂L2p2-p and ‖δyv‖Lp≤|y|‖v‖W1,p, we get‖δyv‖L3≤C|y|θ‖v‖W1,pwithθ=53-2p.

Therefore, the energy conservation criterion (21) is satisfied for θ>13, or equivalently p>32. This result was extended to the borderline case p=32 by Cheskidov, Lopes Filho, Nussenzveig Lopes, and Shvydkoy [16]. Furthermore, they constructed a velocity field with vorticity in L32- exhibiting non-vanishing energy flux. This suggests that there might exist dissipative solutions of the 2D Euler equation with vorticity in L32-. This threshold is somehow related to our construction: the energy dissipation rate of our subsolution vanishes at t=0 if and only if α<43, where recall ω∘=ωs∈L2α-. Let us give a heuristic explanation of this phenomenon.

Heuristic Explanation of Theorem 1.5

First of all, it is convenient to rewrite the Euler equation (1) as 22a ∂tv+div(v⊚v)+∇q=0,

22b divv=0,

22c v|t=0=v∘,

wherev⊚v:=v⊗v-12|v|2I2,

is the projection of v⊗v into the space of symmetric and traceless matrices, and the trace has been absorbed by the Bernoulli pressureq:=p+12|v|2.

Let us analyze the non-uniqueness/energy-dissipation scenario in the Euler equation. Firstly, we introduce some notation. Concerning non-uniqueness: Given v∘∈L2 with divv∘=0 and rotv∘∈L1∩Lp, let us denote S=S(v∘) by the space of admissible solutions to the Euler equation, and Sp=Sp(v∘) by the subset of S formed by velocities with ω∈Lt∞(L1∩Lp). Recall that Sp is non-empty by [29]. With this notation, we can rewrite Yudovich’s well-posedness Theorem: if p=∞, then S=S∞={v} where v is the Yudovich solution. Similarly, we can reformulate Conjecture 2.1: for any 2<p<∞ there exists v∘ satisfying |Sp(v∘)|>1, and Theorem 1.1: |S(v∘)|=∞ for the initial data (12). Concerning energy dissipation: Given v∈S we denote its energy byE:=∫T2edx,e:=12|v|2,

and its dissipation byD:=E(0)-E.

Next, we give a condition for non-uniqueness/energy-dissipation in terms of averaged solutions. Given μ a probability measure on S, we definev¯:=∫Svdμ,σ¯:=∫S(v⊚v)dμ,q¯:=∫Sqdμ.

The triple (v¯,σ¯,q¯) satisfies the relaxed Euler equation 23a ∂tv¯+divσ¯+∇q¯=0,

23b divv¯=0,

23c v¯|t=0=v∘.

The (relaxed) energy was obtained by De Lellis and Székelyhidi in [26, Lemma 3]24 E¯=∫T2e¯dx,e¯:=12|v¯|2+λmax(v¯⊚v¯-σ¯),

where λmax denotes the largest eigenvalue. The crucial facts are that e¯ is convex and agrees with e on S. By applying the Jensen inequality and the Fubini-Tonelli Theorem, we get 25a E¯=∫T2e¯(v¯,σ¯)dx=∫T2e¯∫S(v,v⊚v)dμdx

25b ≤∫T2∫Se¯(v,v⊚v)dμdx=∫S∫T2e(v)dxdμ=E(0)-∫SD(v)dμ.

In particular, E¯≤E(0). The last inequality is strict in two situations: If the Jensen inequality (25a) is strict. In this case, e¯ is strictly convex on (the convex envelope) of spt(μ). In particular, |spt(μ)|>1.

If the dissipative term in (25b) is non-vanishing. In this case, D(v)>0 μ-a.e. v∈S.

In short, (1) concerns non-uniqueness and (2) concerns energy dissipation. On the one hand, (1) is to be expected in the regime 1<p<∞. On the other hand, (2) cannot occur if spt(μ)⊂Sp in the regime 32≤p<∞, while it might be possible if 1<p<32.

We have seen that the existence of a solution (v¯,σ¯,q¯) to (23) with strictly decreasing energy E¯ is linked to either non-uniqueness or energy dissipation. Our third main result Theorem 1.5 shows the existence of this solution (v¯,σ¯,q¯), which is called a subsolution in the convex integration framework, for the initial data (12). Remarkably, the initial energy dissipation of the subsolution is imperceptible (∂tE¯|t=0=0) in the regime 32≤p<∞. Below p=32, the energy dissipation rate is more abrupt, which might reflect the contribution of the dissipation (2) to the Jensen gap (1).

By virtue of the h-principle for the Euler equation discovered by De Lellis and Székelyhidi [26] Theorem 1.5 implies the existence of infinitely many admissible solutions, and thus Theorem 1.1 follows as corollaries. Furthermore, the flexibility of the h-principle allows to prescribe any energy functional e strictly greater than e¯ on the region Ω={|x|≤(ct)1α}. Thus, it is possible to select e making the energy E=∫edx either constant (conservative) or decreasing (dissipative). Here we do not consider increasing energies (non-admissible).

Other Admissibility Criteria

In spite of these non-uniqueness results, there exist other admissibility criteria that rule out some of these solutions. The first criterion is to belong to the natural space Sp. In the regime 32≤p<∞, any weak solution v∈S with decreasing energy cannot belong to Sp because Sp⊂Scon:=S∩{E=E(0)}. In contrast, by taking E decreasing in the regime 1<p<32, Theorem 1.1 shows presumably the first example of dissipative solutions v∈S with Lp vorticity data, although this does not necessarily imply that Sp⊈Scon. The second criterion is the vanishing viscosity limit. It was shown in [16, Theorem 2] that any physically realizable solution conserves the energy for every 1<p≤∞. This scaling gap in the energy conservation between ideal solutions and ideal limits has been observed in other equations of Hydrodynamics (see [18] for SQG and [36] for MHD). By taking E constant in the regime 1<p<∞, Theorem 1.1 shows that |Scon|=∞ for the initial data (12), although this does not necessarily implies that |Sp|>1. For radially symmetric solutions, a third way to rule out non-uniqueness could be the stability of 2D viscous vortices (see e.g. [38]). In such a case, the proof of non-uniqueness of Lp physical solutions would require less symmetric initial data. Another prerequisite satisfied by physicial solutions is the local energy (in)equality (see e.g. [26, 30]). Globally dissipative solutions have been constructed in the last years via convex integration (see e.g. [24, 41, 44, 51]). We do not explore this property here.

H-Principle for Symmetric Subsolutions

In this section we write the h-principle for the Euler equation of De Lellis and Székelyhidi [26] for 2D radially symmetric self-similar subsolutions. We start by recalling the definitions of weak solution and subsolution to the Euler equation.

Definition 4.1

A triple (v¯,σ¯,q¯)∈Ct(L2×L1×L1) where26 v¯=v¯1v¯2,σ¯=σ¯1σ¯2σ¯2-σ¯1,q¯,

is a subsolution to the Euler equation (22) if v¯ is weakly divergence-free and27 ∫0T∫R2(v¯·∂tΦ+σ¯:∇Φ+q¯divΦ)dxdt=-∫R2v∘·Φ|t=0dx,

holds for every test function Φ∈Cc1([0,T)×R2). The pair (v¯,q¯) is a weak solution to the Euler equation if (v¯,v¯⊚v¯,q¯) is a subsolution.

Next, we recall the h-principle for the Euler equation [26, Proposition 2]. Recall the definition of the energy functional e¯ (24).

Theorem 4.1

(H-principle for the Euler equation) Let Ω be a non-empty open subset of (0,T]×R2 and let e∈C(Ω) with e1Ω∈CtL1. Suppose there exists a subsolution (v¯,σ¯,q¯) to the Euler equation satisfying the following properties:{σ¯≠v¯⊚v¯}⊆Ω.

(v¯,σ¯) maps continuously Ω into {e¯<e}.

Then, there exist infinitely many weak solutions (v, q) to the Euler equation with Bernoulli’s pressure q=q¯ and velocity v satisfyingv=v¯outsideΩ,12|v|2=einsideΩ.

By virtue of Theorem 4.1, the proof of non-uniqueness of admissible solutions to the Euler equation is reduced to find a subsolution (v¯,σ¯,q¯) with non-empty Ω and strictly decreasing energy E¯=∫e¯dx. In this case, we will say that (v¯,σ¯,q¯) is an admissible subsolution.

Complex Coordinates

In this section we rewrite the relaxed Euler equation (23) in complex coordinatesx=x1+ix2.

In this setting, for any z,w∈R2≃C we denote as usual|z|=z12+z22,z∗=z1-iz2,z⊥=iz=-z2+iz1,

and alsoz·w=(zw∗)1=z1w1+z2w2,z·w⊥=(zw∗)2=z2w1-z1w2.

By slight abuse of the notation, we identify∇=∂1+i∂2,

and (v¯,σ¯,q¯) in (26) withv¯=v¯1+iv¯2,σ¯=σ¯1+iσ¯2,q¯=q¯+i0.

Proposition 4.1

The relaxed Euler equation (23) is written in complex coordinates as 28a ∂tv¯+∇∗σ¯+∇q¯=0,

28b ∇·v¯=0,

28c v¯|t=0=v∘,

and the energy (24) equals29 e¯=12|v¯|2+12v¯2-σ¯.

Furthermore, the subsolution is a solution to the Euler equation if and only ifσ¯=12v¯2.

Proof

On the one hand,divσ¯1σ¯2σ¯2-σ¯1=∂1σ¯1+∂2σ¯2∂1σ¯2-∂2σ¯1=(∂1-i∂2)(σ¯1+iσ¯2).

On the other hand,v¯⊚v¯=12v¯12-v¯222v¯1v¯2[0.1cm]2v¯1v¯2v¯22-v¯12=12(v¯1+iv¯2)2.

For (29) it is easy to check that any traceless symmetric matrix z satisfiesλmaxz1z2z2-z1=z12+z22=|z|,

where we identify z=z1+iz2. Finally, if (v¯,q¯) is a solution to the Euler equation, it holds∇∗12v¯2-σ¯=0.

Then, since 12v¯2-σ¯ is anti-holomorphic and integrable, necessarily 12v¯2-σ¯=0. □

Radial Symmetry

In this section we write the relaxed Euler equation (28) in polar coordinatesx=reiθ,

for radially symmetric subsolutions. More precisely, we assume that the fluid is rotating around the origin: the velocity v¯ is of the form30 v¯(t,x):=h(t,r)ieiθ,

for some real-valued h, to be determined. Under this choice, v¯ is automatically divergence-free (see (35)) and the vorticity ω¯=rotv¯ is radially symmetric ω¯(t,x)=g(t,r), where h and g are related byrg=∂r(rh).

In order to compare σ¯ with 12v¯2=-12h2e2iθ, it seems convenient to take σ¯ of the form31 σ¯(t,x):=-w(t,r)e2iθ,

for some complex-valued w, to be determined. Finally, we also assume (although it can be deduced from the equation) that the Bernoulli pressure q¯ is radially symmetric, and then (by slight abuse of the notation) we write32 q¯(t,x):=q(t,r),

for some (real-valued) q, to be determined.

Proposition 4.2

Under the choice (30)–(32), the relaxed Euler equation (28) is written as 33a i∂th-∂r(r2w)r2+∂rq=0,

33b h|t=0=h∘,

where v∘(x)=h∘(r)ieiθ, and the energy (29) equals34 e¯=12h2+12h2-w.

Furthermore, the subsolution is a solution to the Euler equation if and only ifw=12h2.

In this case, h is steady and q satisfies∂rq=gh.

Proof

By writing the gradient in polar coordinates∇x=eiθ∂r+ir∂θ,

we deduce that∇x∗v¯=e-iθ∂r-ir∂θ(hieiθ)=i∂r(rh)r,∇x∗σ¯=e-iθ∂r-ir∂θ(-we2iθ)=-eiθ∂r(r2w)r2,

and also∇xq¯=eiθ∂rq.

On the one hand (recall h is real-valued) 35a ∇x·v¯=(∇x∗v¯)1=0,

35b ω¯=∇x⊥·v¯=(∇x∗v¯)2=∂r(rh)r=g.

On the other hand,∂tv¯+∇x∗σ¯+∇xq¯=eiθi∂th-∂r(r2w)r2+∂rq.

We have proved (33). The equality (34) follows from the definitions (30)(31). Finally, by decomposing (33a) into its real and imaginary part respectively, we deduce that w=w1+iw2 and (q, h) are related by 36a r2∂rq=∂r(r2w1),

36b r2∂th=∂r(r2w2).

Hence, if w=12h2 we have ∂th=0, and ∂rq=gh follows from37 12∂r(rh)2=r2gh.

This concludes the proof. □

Corollary 4.1

The energy (34) is minimized in w1 by taking38 w1:=12h2.

Under the choice (38), the relaxed Euler equation (33) is written as 39a ∂rq=gh,

39b ∂r(r2w2)=r2∂th,

39c h|t=0=h∘,

and the energy (34) equals40 e¯=12h2+|w2|.

Furthermore, the subsolution is a solution to the Euler equation if and only ifw2=0.

In this case, h is steady.

Proof

It follows from (34), (36) and (37). □

By virtue of Corollary 4.1, the functions w, q and e¯ are determined by h. Therefore, the construction of an admissible subsolution is reduced to find a profile h with non-vanishing w2 and strictly decreasing energy E¯=∫e¯dx.

Scaling Symmetry

In this section we write the relaxed Euler equation (39) for self-similar subsolutions. The (relaxed) Euler equation possess a two-parameter scaling symmetry (see e.g. [2]): If (v¯,σ¯,q¯) is a (sub)solution and λ,μ>0, then41 v¯λ,μ(t,x)=λμv¯(λt,μx),σ¯λ,μ(t,x)=(λμ)2σ¯(λt,μx),q¯λ,μ(t,x)=(λμ)2q¯(λt,μx),

define another (sub)solution. This corresponds to the physical dimensions[x]=L,[t]=T,[v¯]=LT,[σ¯]=[q¯]=(LT)2.

We say that (v¯,σ¯,q¯) is self-similar if it is invariant under the scaling Lα∼T for some α>0, that is, if (v¯,σ¯,q¯)λ,μ=(v¯,σ¯,q¯) for all λ,μ>0 given by the relationλ=1t=cμα,

for some parameters α,c>0.

We assume that the triple (v¯,σ¯,q¯) given by (30)–(32) and (38) is self-similar for some α,c>0. Then, the profiles (h,w2,q) are of the form42 h(t,r):=(ct)1-ααH(ξ),w2(t,r):=-cα(ct)2(1-α)αW2(ξ),q(t,r):=(ct)2(1-α)αQ(ξ),

in self-similar variablesξ:=r(ct)1α,

for some functions (H,W2,Q), to be determined. Under this choice, the vorticity profile is of the form (see (45))g(t,r):=1ctG(ξ),

where H and G are related byξG=∂ξ(ξH).

Proposition 4.3

Under the choice (42), the relaxed Euler equation (39) is written as 43a ∂ξQ=GH,

43b ∂ξ(ξ2W2)=ξ4-α∂ξ(ξα-1H),

43c limξ→∞ξα-1H(ξ)=β,

where h∘(r)=βr1-α for some β∈R, and the energy (40) equalse¯=(ct)2(1-α)α12H2+cα|W2|.

Furthermore, the subsolution is a solution to the Euler equation if and only ifW2=0.

In this case, H(ξ)=βξ1-α.

Proof

First of all, we compute44 ∂th=-cα(ct)1-2ααξ2-α∂ξ(ξα-1H),∂rh=1ct∂ξH.

On the one hand, since45 g=1r∂r(rh)=1ct1ξ∂ξ(ξH)=1ctG,

the equation (43a) follows from(ct)1-2αα∂ξQ=∂rq=gh=(ct)1-2ααGH.

On the other hand, the equation (43b) follows from-cα(ct)3-2αα∂ξ(ξ2W2)=∂r(r2w2)=r2∂th=-cα(ct)3-2ααξ4-α∂ξ(ξα-1H).

The equation (43c) follows fromh∘(r)=limt→0h(t,r)=r1-αlimξ→∞ξα-1H(ξ).

The rest follows from (42) and (43b). □

By virtue of Proposition 4.3, the functions W2,Q and e¯ are determined by H, which is now time-independent (in contrast to h). Notice that the condition (43c) prevents from constructing subsolutions with finite energy. However, this inconvenient can be easily fixed by truncating the profile H (see Sect. 5.4). Therefore, the construction of an admissible subsolution is reduced to find a profile H with non-vanishing W2 and satisfying ∫∂te¯dx<0.

Admissible Subsolutions

In this section we prove Theorems 1.1 & 1.5 by constructing first admissible subsolutions, and then invoking the h-principle for the 2D Euler equation.

The first step is to construct radially symmetric self-similar subsolutions (v¯,σ¯,q¯). These are given by the choices (30)–(32), (38) and (42) in terms of the parameters α,c>0, and some functions (H,W2,Q) which must satisfy the equation (43).

As it is stated in Proposition 4.3, this subsolution (v¯,σ¯,q¯) is a solution to the Euler equation if and only if W2=0, and so H(ξ)=βξ1-α. In contrast to Vishik’s work [71, 72], here the constant β does not play a crucial role, and thus we will take β=1 for simplicity. The profile H(ξ)=ξ1-α corresponds to the steady power-law vortex v¯=vsvs(x)=|x|-αx⊥.

In this case, the velocity and vorticity profiles are given byhs(r)=r1-α,gs(r)=(2-α)r-α,

and the Bernoulli pressure equals46 qs(r)=2-α2(1-α)r2(1-α),α≠1,[0.1cm]lnr,α=1.

Notice that (vs,qs)∈Lloc2×Lloc1 if and only if α<2, and also ωs∈Llocp for p<2α.

As we mentioned at the end of Sect. 4, we need to find a profile H with non-vanishing W2 and satisfying ∫∂te¯dx<0. Since we want to minimize the contribution of W2 to the energy, we assume that the subsolution agrees with the power-law vortex outside [0, 1]47 H(ξ):=ξ1-α,ξ>1.

We also impose the regularity conditions: H(0)=0, H(1)=1, and H∈C1([0,1]). The condition H(0)=0 is necessary to make v¯ continuous at x=0 for t>0. Hence, it remains to determine H in the interval (0, 1). Next, we need to guarantee that the support of W2 is indeed contained in [0, 1]. This yields the first condition for H.

Lemma 5.1

(1st condition for H) Under the choice (47), the support of the solution W2 to (43b) is contained in [0, 1] if and only if H satisfies48 (4-α)∫01ξ2Hdξ=1.

Proof

The solution W2 to (43b) is given byW2(ξ)=1ξ2∫0ξζ4-α∂ζ(ζα-1H)dζ+C,

for some constant C. Since the profile 1ξ2 is neither continuous nor locally integrable at ξ=0, necessarily C=0. By (47), we have W2(ξ)=0 for ξ≥1 if and only if W2(1)=0. Finally, an integration by parts yields49 W2(ξ)=ξH-4-αξ2∫0ξζ2Hdζ,

from which we deduce (48). □

Proposition 5.1

Suppose (47)(48) hold. Then, the solution to the relaxed Euler equation (43) is given by50 Q(ξ):=qs(1)-∫ξ1GHdζ,0<ξ≤1,[0.4cm]qs(ξ),ξ>1,

with qs as in (46), and51 W2(ξ):=ξH-4-αξ2∫0ξζ2Hdζ,0<ξ≤1,[0.4cm]0,ξ>1.

As a result, (v¯,σ¯,q¯)=(vs,vs⊚vs,qs) outside {|x|≤(ct)1α}.

Proof

The initial condition (43c) is automatically satisfied by (47). The other two equations (43a)(43b) can be integrated (recall (49)). □

The Energy

In this section we compute the energy dissipation rate ∫∂te¯dx, and derive the conditions under which it becomes negative. Recall that the energy e¯ is determined by H through the choices (30)–(32), (38), (42), (47) and (48).

Proposition 5.2

It holds52 ∫R2∂te¯dx=-2παc2(2-α)αA-2(2-α)αBct4-3αα,

where 53a A:=12-(2-α)∫01ξH2dξ,

53b B:=∫01ξ2H-4-αξ∫0ξζ2Hdζdξ.

Proof

First of all, by recalling (40) we write∫R2∂te¯dx=∫R2∂t12h2+|w2|dx.

On the one hand, by applying (42), (44) and (47), we compute12∫R2∂th2dx=2π∫01αh∂thrdr=-2πcα(ct)4-3ααA,

whereA=∫01ξ3-αH∂ξ(ξα-1H)dξ=12∫01ξ2(2-α)∂ξ(ξα-1H)2dξ=12-(2-α)∫01ξH2dξ.

On the other hand, by applying (42) and (51), we compute∫R2|w2|dx=2π∫01α|w2|rdr=2πcα(ct)2(2-α)αB,

whereB=∫01|W2|ξdξ=∫01ξ2H-4-αξ∫0ξζ2Hdζdξ.

Hence,∫R2∂t|w2|dx=∂t∫R2|w2|dx=2πcα2(2-α)α(ct)4-3ααBc.

This concludes the proof □

Corollary 5.1

(2nd condition for H) Suppose ∫∂te¯dx<0. Then, necessarily A>0, or equivalently54 2(2-α)∫01ξH2dξ<1.

The Growth Rate c

In this section we select c maximizing the energy dissipation rate.

Proposition 5.3

Suppose (54) holds. Then, ∫∂te¯dx<0 if and only if55 0<c<α2(2-α)AB.

Furthermore, the energy dissipation rate is maximized56 ∫R2∂te¯dx=-2πA4-α4-αααB2(2-α)αt4-3αα,

by taking57 c:=α4-αAB.

Proof

The first statement (55) follows immediately from (52). For (56), we need to maximize the functionalF(c)=2παc2(2-α)αA-2(2-α)αBc.

Since F is concave withF′(c)=2πα2(2-α)αc4-3ααA-4-ααBc,

it follows that F attains its maximum at (57) withF(c)=2πA4-α4-αααB2(2-α)α.

This concludes the proof. □

The Profile H

In this section we construct profiles H satisfying the requirements from the previous sections. We define H as the space of profiles H∈C1([0,1]) satisfying the conditions H(0)=0, H(1)=1, (48) and (54).

Proposition 5.4

The space H is non-empty and convex.

Proof

First of all, notice that the four conditions H(0)=0, H(1)=1, (48) and (54) are convex. Given a,b>0, we consider the ansatz58 H(ξ):=(1-alogξ)ξ1+b.

It is clear that H∈C1([0,1]) with H(0)=0 and H(1)=1. On the one hand, an integration by parts yields∫01ξ2Hdξ=∫01(1-alogξ)ξ3+bdξ=14+b1+a4+b.

Hence, the condition (48) is equivalent to59 a:=(4+b)(α+b)4-α.

On the other hand, an integration by parts yields60 ∫01ξH2dξ=∫01(1-alogξ)2ξ3+2bdξ=14+2b1+2a4+2b1+a4+2b=14(2+b)1+1+a2+b2=:f(b).

Notice that f is continuous on [0,∞) and the condition (54)2(2-α)f(b)<1,

is open. Then, it is enough to check that it is satisfied at b=0. Since61 f(0)=42+α24(4-α)2,

the condition (54) with b=0 is equivalent to α>0. □

We finish this section by computing the energy dissipation rate for the particular ansatz H given in (58). We consider the case b=0 to simplify the computations. For small b’s the result will be similar by continuity. We remark that, although the profile (58) with b=0 is not differentiable at ξ=0, it still satisfies H(0)=0, which makes v¯ continuous at x=0 for t>0. Moreover, it improves the regularity of the power-law vortex. For b>0 we have H′(0)=0, which makes v¯ differentiable at x=0 for t>0. Similarly, it should be possible to construct profiles H with better regularity.

Proposition 5.5

Let H be the profile (58) with b=0. Then, the energy dissipation rate (56) equals∫R2∂te¯dx=-π162α4-α8-ααt4-3αα,

and the growth rate (57) equals62 c=2α4-α2.

Proof

On the one hand, (53a) equalsA=12-(2-α)f(b),

where f is given in (60). On the other hand,ξ2H-4-αξ∫0ξζ2Hdζ=ξ3+b(1-alogξ)-4-α4+bξ3+b(1-alogξ)+a4+b=(b+α)24-αξ3+b|logξ|,

where we have applied (59). Hence, (53b) equalsB=(b+α)24-α∫01ξ3+b|logξ|dξ=(b+α)2(4-α)(4+b)2.

For b=0, these formulas simplify to (recall (61))A=α34(4-α)2,B=α216(4-α).

Therefore,A4-α4-αααB2(2-α)α=1252α4-α8-αα.

This concludes the proof. □

Fig. 3 Plot of the growth rate c(α)1α

The Truncation

In this section we prove Theorems 1.1 & 1.5. Let us fix 0<α<2. Recall that the inconvenience of considering the self-similar subsolutions (v¯,σ¯,q¯) from the previous sections is that they have infinite energy. This is because their tails are not integrable. In order to make the energy finite, we fix r0>0 and consider the truncated profilehχ:=hχ,

where χ:[0,∞)→[0,1] is a smooth cutoff with χ(r)=1 if r∈[0,r0]. On the one hand, the profile h is determined by H via (42), where we take H as in Proposition 5.4 with b=0. This H determines also the growth rate c by (62), the terms W2,Q by Proposition 5.1, and thus (v¯,σ¯,q¯) by (30)–(32), (38) and (42). On the other hand, hχ determines the terms wχ, qχ and e¯χ by Corollary 4.1. The final subsolution (v¯,σ¯,q¯)χ is defined by (30)–(32), which agrees with (v¯,σ¯,q¯) for |x|≤r0 provided that (ct)1α≤r0. Hence, for all 0≤t≤T, where63 T:=r0αc=4-α2α2r0α,

the energy dissipation rate of the (truncated) subsolution equals (recall Proposition 5.5)∂tE¯χ=∫R2∂te¯χ=-π162α4-α8-ααt4-3αα.

Therefore,E¯χ(t)=Eχ(0)-π32α(2-α)2α4-α8-ααt2(2-α)α,

whereEχ(0)=12∫R2|vsχ|2dx≥π∫00r3-2αdr=π2(2-α)r02(2-α).

We have proved Theorem 1.5. For Theorems 1.1 & 1.3, we invoke the h-principle for the Euler equation (Theorem 4.1) by taking some energy profile e satisfying e¯<e on Ω={|x|<(ct)1α}. It is possible to select e making the energy E=∫edx either constant or decreasing. Finally, notice that these solutions are uniformly bounded on [0,T]×R2 if and only if 0<α≤1 due to (42).

The Case α→2

In this section we analyze the borderline case α→2. This corresponds to the point vortex ωs=2πδ0 for χ=1. Notice that the initial velocity has infinite energy at the origin because h∘(r)=1r. However, our subsolution (v¯,σ¯,q¯) has energy dissipation rate∫R2∂te¯=-π2t,

and therefore it has finite energy for t>0E¯(t)=E¯(1)-π2logt.

Hence, (v¯,σ¯,q¯) is well defined in the Banach space Clogt(L2×L1×L1), which is given by the (weighted) norm

This integrability class is enough to make sense of definition (27). Moreover, the initial datum is attained in L2-. The h-principle for the Euler equation [25] can be easily modified to construct velocities in this class. As a result, we show non-uniqueness of dissipative solutions to the Euler equation for the (truncated) point vortex datum.

Theorem 6.1

Let 0<β<1. Then, there exist infinitely many weak solutions v∈CtL2- to the Euler equation starting fromvs(x)=χ(|x|)x⊥|x|2.

Furthermore, v=vs outside {|x|≤2t}, and v∈ClogtL2 with∂tE=-βπ2t,

for all t∈(0,T], where T is given in (63).

Acknowledgements

The author thanks Ángel Castro, Daniel Faraco, Francisco Gancedo, Antonio Hidalgo and László Székelyhidi for stimulating discussions during the preparation of this work, as well as the anonymous referees for their suggestions to improve the presentation of the article. This research started in Princeton during the Special Year 2021–22: h-Principle and Flexibility in Geometry and PDEs. This work owes a great deal to the Analysis Seminar, as well as financial support provided by the Institute for Advanced Study. Part of the work took place at the University of Sevilla. The author would like to thank its financial support and its friendly atmosphere. This work was finished in Leipzig. The author would like to thank the excellent working conditions and financial support provided by the Max Planck Institute for Mathematics in the Science. The author acknowledge financial support from the Spanish Ministry of Science and Innovation through the Severo Ochoa Programme for Centres of Excellence in R &D (CEX2019-000904-S), the grants PID2020-114703GB-I00, PID2021-123839OB-I00, MTM2017-85934-C3-2-P, and EUR2020-112271, and the ERC Advanced Grant 834728.

Funding

Open Access funding enabled and organized by Projekt DEAL.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Albritton D Brué E Colombo M Non-uniqueness of Leray solutions of the forced Navier-Stokes equations Ann. Math. D 2022 196 1 415 455
Albritton, D., Brué, E., Colombo, M.: Non-uniqueness of Leray solutions of the forced Navier-Stokes equations. Ann. Math. D 196(1), 415–455 (2022)
2. Albritton, D., Brué, E., Colombo, M., De Lellis, C., Giri, V., Janisch, M., Kwon, H.: Instability and non-uniqueness for the 2D Euler equations, after M. Vishik, volume 219 of Annals of Mathematics Studies. Princeton University Press, Princeton, NJ (2024)
3. Astala K Iwaniec T Prause I Saksman E Burkholder integrals, Morrey’s problem and quasiconformal mappings J. Am. Math. Soc. 2012 25 2 507 531
Astala, K., Iwaniec, T., Prause, I., Saksman, E.: Burkholder integrals, Morrey’s problem and quasiconformal mappings. J. Am. Math. Soc. 25(2), 507–531 (2012)
4. Bourgain J Li D Strong illposedness of the incompressible Euler equation in integer Cm spaces Geom. Funct. Anal. 2015 25 1 1 86
Bourgain, J., Li, D.: Strong illposedness of the incompressible Euler equation in integer spaces. Geom. Funct. Anal. 25(1), 1–86 (2015)
5. Bressan A Murray R On self-similar solutions to the incompressible Euler equations J. Differ. Equ. 2020 269 6 5142 5203
Bressan, A., Murray, R.: On self-similar solutions to the incompressible Euler equations. J. Differ. Equ. 269(6), 5142–5203 (2020)
6. Bressan A Shen W A posteriori error estimates for self-similar solutions to the Euler equations Discrete Contin. Dyn. Syst. 2021 41 1 113 130
Bressan, A., Shen, W.: A posteriori error estimates for self-similar solutions to the Euler equations. Discrete Contin. Dyn. Syst. 41(1), 113–130 (2021)
7. Brué E Colombo M Nonuniqueness of solutions to the Euler equations with vorticity in a Lorentz space Commun. Math. Phys. 2023 403 2 1171 1192
Brué, E., Colombo, M.: Nonuniqueness of solutions to the Euler equations with vorticity in a Lorentz space. Commun. Math. Phys. 403(2), 1171–1192 (2023)
8. Buck, M., Modena, S.: Non-uniqueness and energy dissipation for 2D Euler equations with vorticity in Hardy spaces. arXiv:2306.05948 (2023)
9. Buckmaster T De Lellis C Székelyhidi L Jr Vicol V Onsager’s conjecture for admissible weak solutions Commun. Pure Appl. Math. 2019 72 2 229 274
Buckmaster, T., De Lellis, C., Székelyhidi, L., Jr., Vicol, V.: Onsager’s conjecture for admissible weak solutions. Commun. Pure Appl. Math. 72(2), 229–274 (2019)
10. Bulut, A., Huynh, M.K., Palasek, S.: Convex integration above the Onsager exponent for the forced Euler equations. arXiv:2301.00804 (2023)
11. Castro A Córdoba D Faraco D Mixing solutions for the Muskat problem Invent. Math. 2021 226 1 251 348
Castro, A., Córdoba, D., Faraco, D.: Mixing solutions for the Muskat problem. Invent. Math. 226(1), 251–348 (2021)
12. Castro A Faraco D Mengual F Degraded mixing solutions for the Muskat problem Calc. Var. Partial Differ. Equ. 2019 58 2 29
Castro, A., Faraco, D., Mengual, F.: Degraded mixing solutions for the Muskat problem. Calc. Var. Partial Differ. Equ. 58(2), 29 (2019)
13. Castro A Faraco D Mengual F Localized mixing zone for Muskat bubbles and turned interfaces Ann. PDE 2022 8 1 50
Castro, A., Faraco, D., Mengual, F.: Localized mixing zone for Muskat bubbles and turned interfaces. Ann. PDE 8(1), 50 (2022)
14. Chen Q Miao C Zheng X The two-dimensional Euler equation in Yudovich and bmo-type spaces Rev. Mat. Iberoam. 2019 35 1 195 240
Chen, Q., Miao, C., Zheng, X.: The two-dimensional Euler equation in Yudovich and bmo-type spaces. Rev. Mat. Iberoam. 35(1), 195–240 (2019)
15. Cheskidov A Constantin P Friedlander S Shvydkoy R Energy conservation and Onsager’s conjecture for the Euler equations Nonlinearity 2008 21 6 1233 1252
Cheskidov, A., Constantin, P., Friedlander, S., Shvydkoy, R.: Energy conservation and Onsager’s conjecture for the Euler equations. Nonlinearity 21(6), 1233–1252 (2008)
16. Cheskidov A Lopes Filho MC Nussenzveig Lopes HJ Shvydkoy R Energy conservation in two-dimensional incompressible ideal fluids Commun. Math. Phys. 2016 348 1 129 143
Cheskidov, A., Lopes Filho, M.C., Nussenzveig Lopes, H.J., Shvydkoy, R.: Energy conservation in two-dimensional incompressible ideal fluids. Commun. Math. Phys. 348(1), 129–143 (2016)
17. Constantin P E W Titi ES Onsager’s conjecture on the energy conservation for solutions of Euler’s equation Commun. Math. Phys 1994 165 1 207 209
Constantin, P., E, W., Titi, E.S.: Onsager’s conjecture on the energy conservation for solutions of Euler’s equation. Commun. Math. Phys 165(1), 207–209 (1994)
18. Constantin P Ignatova M Nguyen HQ Inviscid limit for SQG in bounded domains SIAM J. Math. Anal. 2018 50 6 6196 6207
Constantin, P., Ignatova, M., Nguyen, H.Q.: Inviscid limit for SQG in bounded domains. SIAM J. Math. Anal. 50(6), 6196–6207 (2018)
19. Córdoba D Faraco D Gancedo F Lack of uniqueness for weak solutions of the incompressible porous media equation Arch. Ration. Mech. Anal. 2011 200 3 725 746
Córdoba, D., Faraco, D., Gancedo, F.: Lack of uniqueness for weak solutions of the incompressible porous media equation. Arch. Ration. Mech. Anal. 200(3), 725–746 (2011)
20. Córdoba, D., Martínez-Zoroa, L., Ożański, W.: Instantaneous gap loss of Sobolev regularity for the 2D incompressible Euler equations. arXiv:2210.17458 (2022)
21. Crippa, G., Stefani, G.: An elementary proof of existence and uniqueness for the Euler flow in localized Yudovich spaces. arXiv:2110.15648 (2021)
22. Dai, M., Friedlander, S.: Non-uniqueness of forced active scalar equations with even drift operators. arXiv:2311.06064 (2023)
23. Daneri S Runa E Székelyhidi L Non-uniqueness for the Euler equations up to Onsager’s critical exponent Ann. PDE 2021 7 1 44
Daneri, S., Runa, E., Székelyhidi, L.: Non-uniqueness for the Euler equations up to Onsager’s critical exponent. Ann. PDE 7(1), 44 (2021)
24. De Lellis C Kwon H On nonuniqueness of Hölder continuous globally dissipative Euler flows Anal. PDE 2022 15 8 2003 2059
De Lellis, C., Kwon, H.: On nonuniqueness of Hölder continuous globally dissipative Euler flows. Anal. PDE 15(8), 2003–2059 (2022)
25. De Lellis C Székelyhidi L Jr The Euler equations as a differential inclusion Ann. Math. 2009 170 3 1417 1436
De Lellis, C., Székelyhidi, L., Jr.: The Euler equations as a differential inclusion. Ann. Math. 170(3), 1417–1436 (2009)
26. De Lellis C Székelyhidi L Jr On admissibility criteria for weak solutions of the Euler equations Arch. Ration. Mech. Anal. 2010 195 1 225 260
De Lellis, C., Székelyhidi, L., Jr.: On admissibility criteria for weak solutions of the Euler equations. Arch. Ration. Mech. Anal. 195(1), 225–260 (2010)
27. De Rosa L Haffter S Dimension of the singular set of wild Hölder solutions of the incompressible Euler equations Nonlinearity 2022 35 10 5150 5192
De Rosa, L., Haffter, S.: Dimension of the singular set of wild Hölder solutions of the incompressible Euler equations. Nonlinearity 35(10), 5150–5192 (2022)
28. De Rosa L Tione R Sharp energy regularity and typicality results for Hölder solutions of incompressible Euler equations Anal. PDE 2022 15 2 405 428
De Rosa, L., Tione, R.: Sharp energy regularity and typicality results for Hölder solutions of incompressible Euler equations. Anal. PDE 15(2), 405–428 (2022)
29. DiPerna RJ Majda AJ Concentrations in regularizations for 2-D incompressible flow Commun. Pure Appl. Math. 1987 40 3 301 345
DiPerna, R.J., Majda, A.J.: Concentrations in regularizations for -D incompressible flow. Commun. Pure Appl. Math. 40(3), 301–345 (1987)
30. Duchon J Robert R Inertial energy dissipation for weak solutions of incompressible Euler and Navier-Stokes equations Nonlinearity 2000 13 1 249 255
Duchon, J., Robert, R.: Inertial energy dissipation for weak solutions of incompressible Euler and Navier-Stokes equations. Nonlinearity 13(1), 249–255 (2000)
31. Elgindi T Finite-time singularity formation for C1,α solutions to the incompressible Euler equations on R3 Ann. Math. 2021 194 3 647 727
Elgindi, T.: Finite-time singularity formation for solutions to the incompressible Euler equations on . Ann. Math. 194(3), 647–727 (2021)
32. Elgindi TM Masmoudi N L∞ ill-posedness for a class of equations arising in hydrodynamics Arch. Ration. Mech. Anal. 2020 235 3 1979 2025
Elgindi, T.M., Masmoudi, N.: ill-posedness for a class of equations arising in hydrodynamics. Arch. Ration. Mech. Anal. 235(3), 1979–2025 (2020)
33. Elling V Algebraic spiral solutions of the 2d incompressible Euler equations Bull. Braz. Math. Soc. (N.S.) 2016 47 1 323 334
Elling, V.: Algebraic spiral solutions of the 2d incompressible Euler equations. Bull. Braz. Math. Soc. (N.S.) 47(1), 323–334 (2016)
34. Elling V Self-similar 2d Euler solutions with mixed-sign vorticity Commun. Math. Phys. 2016 348 1 27 68
Elling, V.: Self-similar 2d Euler solutions with mixed-sign vorticity. Commun. Math. Phys. 348(1), 27–68 (2016)
35. Eyink GL Energy dissipation without viscosity in ideal hydrodynamics. I. Fourier analysis and local energy transfer Phys. D 1994 78 3–4 222 240
Eyink, G.L.: Energy dissipation without viscosity in ideal hydrodynamics. I. Fourier analysis and local energy transfer. Phys. D 78(3–4), 222–240 (1994)
36. Faraco D Lindberg S Proof of Taylor’s conjecture on magnetic helicity conservation Commun. Math. Phys. 2020 373 2 707 738
Faraco, D., Lindberg, S.: Proof of Taylor’s conjecture on magnetic helicity conservation. Commun. Math. Phys. 373(2), 707–738 (2020)
37. Förster C Székelyhidi L Jr Piecewise constant subsolutions for the Muskat problem Commun. Math. Phys. 2018 363 3 1051 1080
Förster, C., Székelyhidi, L., Jr.: Piecewise constant subsolutions for the Muskat problem. Commun. Math. Phys. 363(3), 1051–1080 (2018)
38. Gallay T Wayne CE Global stability of vortex solutions of the two-dimensional Navier-Stokes equation Commun. Math. Phys. 2005 255 1 97 129
Gallay, T., Wayne, C.E.: Global stability of vortex solutions of the two-dimensional Navier-Stokes equation. Commun. Math. Phys. 255(1), 97–129 (2005)
39. García, C., Gómez-Serrano, J.: Self-similar spirals for the generalized surface quasi-geostrophic equations. arXiv:2207.12363 (2022)
40. Gebhard, B., Hirsch, J., Kolumbán, J.J.: On a degenerate elliptic problem arising in the least action principle for Rayleigh-Taylor subsolutions. arXiv:2208.14495 (2022)
41. Gebhard B Kolumbán JJ On bounded two-dimensional globally dissipative Euler flows SIAM J. Math. Anal. 2022 54 3 3457 3479
Gebhard, B., Kolumbán, J.J.: On bounded two-dimensional globally dissipative Euler flows. SIAM J. Math. Anal. 54(3), 3457–3479 (2022)
42. Gebhard B Kolumbán JJ Relaxation of the Boussinesq system and applications to the Rayleigh-Taylor instability NoDEA Nonlinear Differ. Equ. Appl. 2022 29 1 38
Gebhard, B., Kolumbán, J.J.: Relaxation of the Boussinesq system and applications to the Rayleigh-Taylor instability. NoDEA Nonlinear Differ. Equ. Appl. 29(1), 38 (2022)
43. Gebhard B Kolumbán JJ Székelyhidi L A new approach to the Rayleigh-Taylor instability Arch. Ration. Mech. Anal. 2021 241 3 1243 1280 34720113
Gebhard, B., Kolumbán, J.J., Székelyhidi, L.: A new approach to the Rayleigh-Taylor instability. Arch. Ration. Mech. Anal. 241(3), 1243–1280 (2021)34720113
44. Giri, V., Kwon, H., Novack, M.: The -based strong Onsager theorem. arXiv:2305.18509 (2023)
45. Giri, V., Radu, R.-O.: The 2D Onsager conjecture: a Newton-Nash iteration. arXiv:2305.18105 (2023)
46. Guillod J Šverák V Numerical investigations of non-uniqueness for the Navier-Stokes initial value problem in borderline spaces J. Math. Fluid Mech. 2023 25 3 25
Guillod, J., Šverák, V.: Numerical investigations of non-uniqueness for the Navier-Stokes initial value problem in borderline spaces. J. Math. Fluid Mech. 25(3), 25 (2023)
47. Gunther N On the motion of fluid in a moving container. Izvestia Akad. Nauk USSR Ser. Fiz.-Mat. 1927 20 1323 1348
Gunther, N.: On the motion of fluid in a moving container. Izvestia Akad. Nauk USSR. Ser. Fiz.-Mat. 20, 1323–1348 (1927)
48. Hitruhin L Lindberg S Lamination convex hull of stationary incompressible porous media equations SIAM J. Math. Anal. 2021 53 1 491 508
Hitruhin, L., Lindberg, S.: Lamination convex hull of stationary incompressible porous media equations. SIAM J. Math. Anal. 53(1), 491–508 (2021)
49. Hölder E Über die unbeschränkte Fortsetzbarkeit einer stetigen ebenen Bewegung in einer unbegrenzten inkompressiblen Flüssigkeit Math. Z. 1933 37 1 727 738
Hölder, E.: Über die unbeschränkte Fortsetzbarkeit einer stetigen ebenen Bewegung in einer unbegrenzten inkompressiblen Flüssigkeit. Math. Z. 37(1), 727–738 (1933)
50. Isett P A proof of Onsager’s conjecture Ann. Math. 2018 188 3 871 963
Isett, P.: A proof of Onsager’s conjecture. Ann. Math. 188(3), 871–963 (2018)
51. Isett P Nonuniqueness and existence of continuous, globally dissipative Euler flows Arch. Ration. Mech. Anal. 2022 244 3 1223 1309
Isett, P.: Nonuniqueness and existence of continuous, globally dissipative Euler flows. Arch. Ration. Mech. Anal. 244(3), 1223–1309 (2022)
52. Isett P Vicol V Hölder continuous solutions of active scalar equations Ann. PDE 2015 1 1 77
Isett, P., Vicol, V.: Hölder continuous solutions of active scalar equations. Ann. PDE 1(1), 77 (2015)
53. Jia H Šverák V Are the incompressible 3d Navier-Stokes equations locally ill-posed in the natural energy space? J. Funct. Anal. 2015 268 12 3734 3766
Jia, H., Šverák, V.: Are the incompressible 3d Navier-Stokes equations locally ill-posed in the natural energy space? J. Funct. Anal. 268(12), 3734–3766 (2015)
54. Khor, C., Miao, C., Ye, W.: Infinitely many non-conservative solutions for the three-dimensional Euler equations with arbitrary initial data in . arXiv:2204.03344 (2022)
55. Lichtenstein L Über einige Existenzprobleme der Hydrodynamik homogener, unzusammendrückbarer, reibungsloser Flüssigkeiten und die Helmholtzschen Wirbelsätze Math. Z. 1925 23 1 89 154
Lichtenstein, L.: Über einige Existenzprobleme der Hydrodynamik homogener, unzusammendrückbarer, reibungsloser Flüssigkeiten und die Helmholtzschen Wirbelsätze. Math. Z. 23(1), 89–154 (1925)
56. Majda, A.J., Bertozzi, A.L.: Vorticity and incompressible flow. Cambridge Texts in Applied Mathematics, vol. 27. Cambridge University Press, Cambridge (2002)
57. Mengual F H-principle for the 2-dimensional incompressible porous media equation with viscosity jump Anal. PDE 2022 15 2 429 476
Mengual, F.: H-principle for the 2-dimensional incompressible porous media equation with viscosity jump. Anal. PDE 15(2), 429–476 (2022)
58. Mengual F Székelyhidi L Jr Dissipative Euler flows for vortex sheet initial data without distinguished sign Commun. Pure Appl. Math. 2023 76 1 163 221
Mengual, F., Székelyhidi, L., Jr.: Dissipative Euler flows for vortex sheet initial data without distinguished sign. Commun. Pure Appl. Math. 76(1), 163–221 (2023)
59. Modena S Sattig G Convex integration solutions to the transport equation with full dimensional concentration Ann. Inst. H. Poincaré C Anal. Non Linéaire 2020 37 5 1075 1108
Modena, S., Sattig, G.: Convex integration solutions to the transport equation with full dimensional concentration. Ann. Inst. H. Poincaré C Anal. Non Linéaire 37(5), 1075–1108 (2020)
60. Modena S Székelyhidi L Non-uniqueness for the transport equation with Sobolev vector fields Ann. PDE 2018 4 2 38
Modena, S., Székelyhidi, L.: Non-uniqueness for the transport equation with Sobolev vector fields. Ann. PDE 4(2), 38 (2018)
61. Modena S Székelyhidi L Non-renormalized solutions to the continuity equation Calc. Var. Partial Differ. Equ. 2019 58 6 30
Modena, S., Székelyhidi, L.: Non-renormalized solutions to the continuity equation. Calc. Var. Partial Differ. Equ. 58(6), 30 (2019)
62. Noisette F Székelyhidi L Jr Mixing solutions for the Muskat problem with variable speed J. Evol. Equ. 2021 21 3 3289 3312
Noisette, F., Székelyhidi, L., Jr.: Mixing solutions for the Muskat problem with variable speed. J. Evol. Equ. 21(3), 3289–3312 (2021)
63. Novack, M., Vicol, V.: An intermittent Onsager theorem. Invent. Math., (2023)
64. Onsager, L.: Statistical hydrodynamics. Nuovo Cimento (9), 6(Supplemento, 2 (Convegno Internazionale di Meccanica Statistica)):279–287, (1949)
65. Scheffer V An inviscid flow with compact support in space-time J. Geom. Anal. 1993 3 4 343 401
Scheffer, V.: An inviscid flow with compact support in space-time. J. Geom. Anal. 3(4), 343–401 (1993)
66. Shnirelman A On the nonuniqueness of weak solution of the Euler equation Commun. Pure Appl. Math. 1997 50 12 1261 1286
Shnirelman, A.: On the nonuniqueness of weak solution of the Euler equation. Commun. Pure Appl. Math. 50(12), 1261–1286 (1997)
67. Shvydkoy R Convex integration for a class of active scalar equations J. Am. Math. Soc. 2011 24 4 1159 1174
Shvydkoy, R.: Convex integration for a class of active scalar equations. J. Am. Math. Soc. 24(4), 1159–1174 (2011)
68. Székelyhidi L Weak solutions to the incompressible Euler equations with vortex sheet initial data C. R. Math. Acad. Sci. Paris 2011 349 19–20 1063 1066
Székelyhidi, L.: Weak solutions to the incompressible Euler equations with vortex sheet initial data. C. R. Math. Acad. Sci. Paris 349(19–20), 1063–1066 (2011)
69. Székelyhidi L Relaxation of the incompressible porous media equation Ann. Sci. Éc. Norm. Supér. 2012 45 3 491 509
Székelyhidi, L.: Relaxation of the incompressible porous media equation. Ann. Sci. Éc. Norm. Supér. 45(3), 491–509 (2012)
70. Taniuchi Y Uniformly local Lp estimate for 2-D vorticity equation and its application to Euler equations with initial vorticity in bmo Commun. Math. Phys. 2004 248 1 169 186
Taniuchi, Y.: Uniformly local estimate for 2-D vorticity equation and its application to Euler equations with initial vorticity in . Commun. Math. Phys. 248(1), 169–186 (2004)
71. Vishik, M.: Instability and non-uniqueness in the Cauchy problem for the Euler equations of an ideal incompressible fluid. Part I. arXiv:1805.09426 (2018)
72. Vishik, M.: Instability and non-uniqueness in the Cauchy problem for the Euler equations of an ideal incompressible fluid. Part II. arXiv:1805.09440 (2018)
73. Wiedemann E Existence of weak solutions for the incompressible Euler equations Ann. Inst. H. Poincaré C Anal. Non Linéaire 2011 28 5 727 730
Wiedemann, E.: Existence of weak solutions for the incompressible Euler equations. Ann. Inst. H. Poincaré C Anal. Non Linéaire 28(5), 727–730 (2011)
74. Wiedemann, E.: Weak-strong uniqueness in fluid dynamics. In: Partial differential equations in fluid mechanics, volume 452 of London Math. Soc. Lecture Note Ser., pages 289–326. Cambridge Univ. Press, Cambridge (2018)
75. Wolibner W Un theorème sur l’existence du mouvement plan d’un fluide parfait, homogène, incompressible, pendant un temps infiniment long Math. Z. 1933 37 1 698 726
Wolibner, W.: Un theorème sur l’existence du mouvement plan d’un fluide parfait, homogène, incompressible, pendant un temps infiniment long. Math. Z. 37(1), 698–726 (1933)
76. Yudovich V Non-stationary flow of an ideal incompressible liquid USSR Comput. Math. Math. Phys. 1963 3 6 1407 1456
Yudovich, V.: Non-stationary flow of an ideal incompressible liquid. USSR Comput. Math. Math. Phys. 3(6), 1407–1456 (1963)
77. Yudovich VI Uniqueness theorem for the basic nonstationary problem in the dynamics of an ideal incompressible fluid Math. Res. Lett. 1995 2 1 27 38
Yudovich, V.I.: Uniqueness theorem for the basic nonstationary problem in the dynamics of an ideal incompressible fluid. Math. Res. Lett. 2(1), 27–38 (1995)
