(C) The Kolmogorov 4/5 - law

We have focussed so far on the absolute structure functions, that were used to obtain bounds

on the energy flux ⇧` . However, there are other types of structure functions of interest, some

of them more directly related to energy flux and which, in fact, provide alternate definitions of

it. Rather than define “large-scale energy” by

ē` (x, t) = 12 |v̄` (x, t)|2

one can instead make an alternate definition by filtering just one factor
1
e` (x, t) = v(x, t) · v̄` (x, t)
Z2
1
= dd r G` (r) v(x, t) · v(x + r, t)
2
where

er (x, t) = 12 v(x, t) · v(x + r, t)

is the so-called point-split kinetic energy density. The quantities e` (x, t) and er (x, t) are well-

defined whenever v has finite mean energy:

Z Z
1 T 1
dt dd x |v(x, t)|2 < +1.
T 0 V 2
We now derive a balance equation for the point-split kinetic energy density of a Navier-Stokes

solution, as follows


1 1 1 1 1
@t ( v · v ) + r · ( v · v0 )v +
0
(pv0 + p0 v) + |v0 |2 v ⌫r( v · v0 )
2 2 2 4 2
1 0 1 0 0
= rr · [ v | v|2 ] ⌫rv : rv + (f · v + f · v)
4 2

with the notations

v = v(x, t), p = p(x, t)

v0 = v(x + r, t), p0 = p(x + r, t)

v = v0 v = v(x + r, t) v(x, t)

17
Remarks:

#1. This identity was first derived by L. Onsager in the 1940’s in a smoothed and space-

integrated form. The result was communicated to C. C. Lin in a letter in 1945, but never

formally published by Onsager. The space-local form was derived by J. Duchan & R. Robert,

Nonlinearity 13 249-255(2000) in a smoothed version, discussed a bit later.

#2. The relation is analogous to the energy balance equation that we derived in the filtering

approach:

@t ( 12 |v̄` |2 ) +r· 1 2
2 |v̄` | v̄` + p̄` v̄` ⌧ ` · v̄` ⌫r( 12 |v̄` |2 ) = rv̄` : ⌧ ` ⌫|rv̄` |2 + f` · v̄`

Proof of the identity: Take

@t v + r · (vv) ⌫4v + rp = f , r · v = 0

@t v0 + r · (v0 v0 ) ⌫4v0 + rp0 = f 0 , r · v0 = 0

Dot the first by v0 and the second by v and add together, to obtain

@t (v · v0 ) + v · [r · (v0 v0 ) ⌫4v0 ] + v0 · [r · (vv) ⌫4v]

+ r · (p0 v + pv0 ) = f 0 · v + f · v0

Now the viscous term is reorganized as

v · 4v0 + v0 · 4v = r · [vi rvi0 + vi0 rvi ] 2rvi · rvi0

= r · [r(v · v0 )] 2rv : rv0

Lastly we discuss the crucial nonlinear term. Note first that

v · [r · (v0 v0 )] + v0 · [r · (vv)] = vi @j (vi0 vj0 ) + vi0 @j (vi vj )

= vi @j (vi0 vj0 ) + @j (vi0 vi vj ) vi vj (@j vi0 )

= 4 + r · [(v · v0 )v]

with

4 ⌘ vi @j (vi0 vj0 ) vi vj (@j vi0 )

18
 .

By incompressibility,

4 = vi vj0 (@j vi0 ) vi vj (@j vi0 )

= vi (vj0 vj )(@j vi0 )

= vi vj (@j vi0 )

Also by incompressibility rr · ( v) = rr · v0 = 0, so that

rr · [ v | v|2 ] = ( v · r0 )| v|2

= 2( vj @j )vi0 · (vi0 vi )

= 2 vj · vi0 @j vi0 2vi vj (@j vi0 )

= v · r(|v0 |2 ) 24

= r · [|v0 |2 v] 24

Thus,

1
@t (v · v ) + r · (v · v0 )v + (pv0 + p0 v) + |v0 |2 v ⌫r(v · v0 )
0
2
1
rr · [ v | v|2 ] + 2⌫rv : rv0 = (f · v0 + f 0 · v)
2

QED!

Multiplying the point-split identity through by G` (r) and integrating over r gives a correspond-

ing balance equation for the regularized energy density 12 v · v̄` :


1 1 1 1 1 1
@t ( v · v̄` ) + r · ( v · v̄` )v + (pv̄` + p̄` v) + (|v|2 v)` (|v|2 )` v ⌫r( v · v̄` )
2 Z2 2 4 4 2
1 d 2 1
= d r (rG)` (r) · v(r) | v(r)| ⌫rv : rv̄` + (f · v̄` + f̄` · v)
4` 2

The above equation can be shown to be valid even for singular Leray solutions of INS, if the

space-time derivatives are interpreted in the sense of distributions. (See Appendix.)

We now consider the limit of vanishing viscosity ⌫ ! 0. If the Navier-Stokes solution v⌫ ! v

as ⌫ ! 0 in the space-time L2 -sense, i.e.

19
 R R
kv⌫ vk2L2 = dt dd x |v⌫ (x, t) v(x, t)|2 ! 0, as ⌫ ! 0
spacetime

then it is not hard to show that the limiting velocity v is a solution of the incompressible Euler

equations

@t v + r · (vv) = rp + f , r · v = 0

in the sense of space-time distributions. The relevant simple estimate is that

kv⌫ v⌫ vvkL1 = kv⌫ (v⌫ v) + (v⌫ v)vkL1

 kv⌫ kL2 kv⌫ vkL2 + kv⌫ vkL2 kvkL2

The key assumption of convergence of the velocity as ⌫ ! 0 in the strong L2 -sense has never

been proved a priori, but it is consistent with empirical observations, following an argument of

Chen, G.-Q. & Glimm, J., “Kolmogorov’s theory of turbulence and inviscid limit of

the Navier-Stokes equations in R3 ,” Commun. Math. Phys. 310, 267–283 (2012).

as further elaborated in these works:

P. Isett, “Nonuniqueness and existence of continuous, globally dissipative Euler

flows,”’ Arch. Ration. Mech. Anal. 244 12231309 (2022); originally appearing in

[Link]

T. D. Drivas and H. Q. Nguyen. “Remarks on the emergence of weak Euler solutions

in the vanishing viscosity limit,” J. Nonlin. Sci. 29 709–721 (2019).

For example, assume that the 2nd-order structure function S2⌫ (r, t) = h| v⌫ (r)|2 i defined by an

instantaneous space-average satisfies a scaling relation of the form

8
>
< C2 (t)u2 (r/L)⇣2 ,
⌫ rms ⌘2 < r < L
S2 (r, t) ⇠ (i)
>
: C2 (t)Re (2 ⇣2 ) u2 (r/L)2 r < ⌘2
rms

with ⌘2 ⇠ L(Re) for some > 0 and with ⌫-independent constant C2 (t) so that
Z T
1
dt C2 (t) < 1. (ii)
T 0

20
It then follows that the “Besov semi-norm” |||v⌫ (t)|||B 2 ,1 with 2 = ⇣2 /2 is essentially given by
2

the constant C2 (t) :

h| v⌫ (r)|2 i
|||v⌫ (t)|||2B 2 := sup 2 2
= C2 (t)u2rms
2
|r|<L (r/L)

and condition (ii) with an energy bound independent of ⌫ gives v⌫ 2 L2 ([0, T ], B2 2 ,1 (⌦))

uniformly in the viscosity. Because B2 2 ,1 (⌦) is compactly embedded in L2 (⌦) according to a

theorem of Kolmogorov & Riesz, a strong limit v⌫ ! v exists in L2spacetime as ⌫ ! 0 (at least

along a subsequence) by a result known as the Aubin-Lions-Simon lemma. For example, see:

F. Boyer & P. Fabrie, Mathematical Tools for the Study of the Incompressible Navier-

Stokes Equations and Related Models, (Springer, New York, 2013), section 5.3.

This same argument implies also convergence strong in Lpspacetime for p > 2 with an analogous

hypothesis on the absolute structure-function Sp⌫ (r, t).

R R
If it is furthermore true that dt dd x |v(x, t)|3 < +1, then the point-split balance holds

also for the Euler solution in the distribution sense:


1 1 1 1 1
@t ( v · v̄` ) + r · ( v · v̄` )v + (pv̄` + p̄` v) + (|v|2 v)` (|v|2 )` v
2 2Z 2 4 4
1
= dd r (rG)` (r) · v(r) | v(r)|2 (?)
4`

For simplicity, we consider the case with no external force, f = 0. We now consider the limit
R R
` ! 0. Under the same basic assumption, that dt dd x|v(x, t)|3 < +1, it is not hard to

show that the LHS of equation (?)


1 1 1 1 1
@t ( v · v̄` ) + r · ( v · v̄` )v + (pv̄` + p̄` v) + (|v|2 v)` (|v|2 )` v
2 2  2 4 4
1 2 1
! @t ( |v| ) + r · ( |v|2 + p)v ⌘ D(v), as ` ! 0
2 2

in the sense of distributions. Just to consider one typical term,

Z Z ✓ ◆ Z Z
d 1 1
dt d x r'(x, t) · v(x, t) · v̄` (x, t) v(x, t) dt dd x r'(x, t) · |v(x, t)|2 v(x, t)
2 2
1 1
 sup |r'| · k( v · v̄` )v ( |v|2 )vkL1
2 2 spacetime

21
Using 12 (v · v̄` )v 1 2
2 |v| v = 12 [v · (v̄` v)]v and the Hölder inequality

k[v · (v̄` v)]vkL1  kvk2L3 kv̄` vkL3

One can then show that the upper bound ! 0 as ` ! 0. This implies that
⇥ ⇤ ⇥ ⇤
r · ( 12 v · v̄` )v ! r · 12 |v|2 v

in the sense of distributions. The other terms are treated in a similar fashion. But since the

LHS of equation (?) converges to D(v) in the sense of the distributions, so does the RHS!

That is,

Z
1
D(v) = lim dd r (rG)` (r) · v(r)| v(r)|2 (1)
`!0 4`

in the sense of distributions. To summarize, we obtain the above formula for the anomalous

dissipation D(v) that appears in the energy balance relation


1 2 1
@t ( |v| ) + r · ( |v|2 + p)v = D(v). (2)
2 2

for the singular Euler solution v(x, t). This result is quite interesting in its own right and not

just as a step in the proof of the Kolmogorov 4/5-law. It is a precise mathematical formulation

of Onsager’s idea that Euler solutions which arises in the zero-viscosity limit of turbulent flow

may not conserve energy. We could derive the same balance equation (2) by starting from

the balance equation for 12 |v̄` |2 and taking the limit ` ! 0. This would give us another valid

expression

D(v) = lim ⇧` (in the distribution sense)

`!0

for the anomalous dissipation D(v). In fact, the RHS of equation (?)
Z
1
D` (v) = dd r (rG)` (r) · v(r)| v(r)|2 (3)
4`

22
is another way of measuring energy flux to small scales, alternative to ⇧` . We can get from

equation (1) for D(v) Onsager’s bound on energy flux to small scales. For example, if v has

Hölder exponent h, then it follows from equation (3) that

D` (v) = O(`3h 1)

the same bound derived earlier for ⇧` . These bounds imply the assertion of Onsager in his 1949

paper that Euler solutions must conserve energy if the velocity has Hölder exponent h > 1/3.
1
Using Lp norms, one can easily show also that energy is conserved if p > 3 (equivalently,

⇣p > p/3) for p 3. Under any of these regularity assumptions, D(v) ⌘ 0!

Let us now return to our derivation of the 4/5-law, by obtaining a simplified expression for D(v)

for the case of a spherically symmetric filter kernel G that depends upon only the magnitude

r = |r|:

G(r) = G(r)

so that

rG(r) = r̂G0 (r).

In that case, one can go to spherical coordinates in d-dimensions

1
R1 d 1 R
D` (v) = 4` 0 r dr S d 1 d!(r̂)r̂ · v(r)| v(r)|2 (G0 )` (r)

where S d 1 is the unit sphere in d-dimensions and d! is the measure on solid angles. Now

introduce

vL (r) = r̂ · v(r) = longitudinal velocity increment

and
Z
1
h vL (r)| v(r)|2 iang = d!(r̂) vL (r)| v(r)|2
⌦d 1 S d 1
= angular average of vL | v|2

where ⌦d 1 is the (d 1) -dimension volume of S d 1. We thus find that

Z 1
1
D` (v) = ⌦d 1 rd 1
dr (G0 )` (r)h vL (r)| v(r)|2 iang
4` Z 0
1
h vL (r)| v(r)|2 iang
= ⌦d 1 ⇢d d⇢ G0 (⇢)
0 4r r=`⇢

23
where ⇢ = r/`. We know that the limit of the LHS exists as ` ! 0 in the sense of distributions

and gives D(v). Taking the limit on the RHS, we see that
h vL (r)| v(r)|2 iang
4r ! D⇤ (v) , as r ! 0

with
Z 1
⇤
D(v) = D (v) · ⌦d ⇢d d⇢G0 (⇢)
1
0 Z 1
⇤
= D (v) · ( d · ⌦d 1 ⇢d 1 d⇢G(⇢)) by integration by parts
0 Z 1
⇤
= d · D (v) since ⌦d 1 ⇢d 1 d⇢G(⇢) = 1
0

We conclude finally that

h vL (r)| v(r)|2 iang 4

lim = D(v)
r!0 r d

The results in the above form were given in the paper

J. Duchon & R. Robert, “Inertial energy dissipation for weak solution of incom-

pressible Euler and Navier-Stokes equations,” Nonlinearity, 13 249-255(2000).

It is possible, by an elaboration of these arguments, to derive expressions for D(v) that involve

only vL (r), or mixed expressions that involve vL (r) and the transverse velocity increment

vT (r) = v(r) vL (r)r̂

which satisfies r̂ · vT (r) = 0, or

vT2 (r) = | vT (r)|2 /(d 1),

the magnitude of the transverse velocity increment per component. These are, in d-dimensions,

h u3L (r)iang 12
lim = D(v)
r!0 r d(d + 2)
h vL (r) vT2 (r)iang 4
lim = D(v)
r!0 r d(d + 2)

24
For the derivation, see M. Novack, Nonlinearity 37 095002 (2024)

Example: Burgers Equation

The above discussion has been a bit abstract, so that it is useful to consider a concrete example.

All the previous results have exact analogies for singular/distributional solutions of the inviscid

Burgers equation, which can be shown to satisfy the energy balance equation

@t ( 12 u2 ) + @x ( 13 u3 ) = D(u)
1
R +1
with D(u) = lim`!0 12` 1 dr(G0 )` (r) u3 (r) in the sense of distributions. Alternately,
h u3L (r)iang
limr!0 |r| = 12D(u)

where uL (r) := sign(r) u(r) and h u3L (r)iang = 12 [ u3 (+|r|) u3 ( |r|)]. For the Khokhlov

sawtooth solution in the limit ⌫ ! 0 it is straightforward to calculate explicitly that, with r > 0

1h i
h u3L (r)iang = (r/t 4u)3 + (r/t)3 [ r,0] (x)
2 h i
1
+ (r/t)3 ( r/t + 4u)3 [0,r] (x)
⇣2 r ⌘3
+ [ r,r]c (x)
t

where u := 2L/t > 0, so that

h u3L (r)iang
r ! [ 12 (4u)3 + 12 (4u)3 ] (x) = (4u)3 (x) , as r ! 0.

Notice that this is equal to 12"(x), where "(x) = limr!0 ⌫(@x u⌫ )2 is the distributional limit

of the viscous dissipation in u⌫ (x, t) as ⌫ ! 0.

A similar result can be obtained for the ⌫ ! 0 limit of Leray solutions of the Navier-Stokes

equation. These satisfy a local energy balance of the form


1 1 1
@t ( |v⌫ |2 ) + r · ( |v⌫ |2 + p⌫ )v⌫ ⌫r( |v⌫ |2 ) = ⌫|rv⌫ |2 (or,  ⌫|rv⌫ |2 ), (?)
2 2 2

For simplicity, we shall only consider the case where “=” holds above rather than “”. (For the

general case, see Appendix.) Let us assume that v⌫ ! v as ⌫ ! 0 in the L3 -sense in spacetime:
R R d
dt d x |v⌫ (x, t) v(x, t)|3 ! 0.

This is stronger than the L2 -convergence assumed earlier, so that, again, the limiting velocity

25
v is an Euler solution in distribution sense. Furthermore, it is now possible to check that the

LHS of equation (?) above has the limit

⇥ ⇤ ⇥ ⇤
lim⌫!0 @t ( 12 |v⌫ |2 ) + r · ( 12 |v⌫ |2 + p⌫ )v⌫ ⌫r( 12 |v⌫ |2 ) = @t ( 12 |v|2 ) + r · ( 12 |v|2 + p)v

in distribution sense. The argument is very similar to that which we gave earlier for the limit

` ! 0. Furthermore, the limit is exactly the same, i.e. D(v)! Since the limits of the LHS and

the RHS of equation (?) must be the same, we obtain

D(v) = lim⌫!0 ⌫|rv⌫ |2 = lim⌫!0 "⌫ (Duchon & Robert, 2000)

in the sense of distributions. Notice the RHS of the above expression is non-negative, so that

its limit also must be:

D(v) 0
R
More precisely, D(v) is a nonnegative distribution, which satisfies dd x dt '(x, t)D(v)(x, t) 0

for every nonnegative test function ' ( C 1 with compact support). It is known that every

nonnegative distribution is given by a measure, i.e.

R d R RR
d x dt '(x, t)D(v)(x, t) = µ(dx, dt) '(x, t)

This “dissipation measure” has been much studied experimentally and observed to have multi-

fractal scaling properties, as we discuss a bit later! If ' is nonnegative and also normalized
R R d
dt d x '(x, t) = 1,

then we can interpret

R R
dt dd x '(x, t) ⌫|rv⌫ (x, t)|2 ⌘ h⌫|rv⌫ |2 i'

as an average in spacetime over the compact support of ', weighted by '. The above result

then says that

lim⌫!0 h"⌫ i' = hD(v)i'

Our earlier results can be stated in a similar fashion, e.g.

⌫ (r)]3 i
h[ vL ',ang 12
limr!0 lim⌫!0 r = d(d+2) hD(v)i' .

We thus see that, taking first ⌫ ! 0,

26
 12
h u3L (r)i',ang ⇠ d(d+2) h"i' r

12 4
This is the famous Kolmogorov 4/5-law (since the coefficient d(d+2) = 5 for d = 3), derived by

Kolmogorov in the third of his celebrated 1941 papers on turbulence. The related results

4
h vL (r) u2T (r)i',ang ⇠ h"i' r
d(d + 2)
4
h vL (r)| v(r)|2 i',ang ⇠ h✏i' r
d

are called the Kolmogorov 4/15- and 4/3-laws, respectively. These were derived by Kolmogorov

in the statistical sense, averaging over an ensemble of solutions assuming statistical homogeneity

and isotropy. He employed in his derivation an equation derived earlier for the 2-point velocity

correlation hvi (x, t)vj (x + r, t)i by van Kármán and Howarth (1938), so that this is sometimes

called the Kolmogorov-Kármán-Howarth relation. The result presented here is much stronger,

because there is no average over ensembles and no assumption of homogeneity and/or isotropy.

It seems to have been Onsager in the 1940’s who realized that such relations should hold for

individual realizations, without averaging. He derived the formula

1
R d
D` (v) = 4` d r (rG)` (r) · v(r)| v(r)|2

and discussed its limit for ` ! 0. In the statistical framework, the corresponding result

rr · h v(r)| v(r)|2 i ⇠ 4h"i , as r!0

was derived by A. S. Monin (1959) and is sometimes called the Kolmogorov-Monin relation.

It does not assume isotropy. There is another derivation of the 4/5-law by Nie & Tanveer

(1999) without statistical averaging. It uses also space-time averaging and angle-averaging. It

is stronger than the result presented here in that it includes viscous corrections, but it is weaker

than the presented local results, since it requires a global spacetime average.

27
Ensemble Approach to the 4/5th Law

We have focused on the deterministic version of the 4/5th law, but the traditional approach

using ensemble averages has some advantages. For one thing, it provides a simple framework

within which to study the e↵ects of finite Reynolds number. See:

R. A. Antonia et al. et al., “Finite Reynolds number e↵ect and the 4/5 law,” PRF,

4, 084602 (2019)

which extensively reviews this aspect (although we disagree with a great many theoretical claims

in this work!)

Another important contribution of the statistical approach is in the paper:

J. Bedrossian et al. “A sufficient condition for the Kolmogorov 4/5 Law for sta-

tionary martingale solutions to the 3D Navier-Stokes equations,” Commun. Math.

Phys. 367 1045-1075 (2019)

This work involves some (unphysical) mathematical complications because of the assumption

of a continuum Navier-Stokes description, for which only weak solutions are known to exist

(see Appendix on Leray’s theory). However, the paper makes an important contribution by

attempting to derive the the 4/5th-law under the weakest possible hypotheses. The authors

consider the situation with turbulence driven by an external body force which is spatially

homogeneous and white-in-time:

hfi (x, t)fj (x0 , t0 )i = 2Fij (r/L) (t t0 ), r=x x0

as first discussed here:

E. A. Novikov, “Functionals and the random-force method in turbulence theory,”

Sov. Phys. JETP 20 1290-1294 (1965)

28
The useful result with forcing white-in-time obtained by Novikov (and derived rigorously by

Bedrossian et al. for a notion of weak solutions) is that the power input by the force is fixed as

Q = tr(F(0)). Therefore, in the long-time statistical steady-state Q = ", or

tr(F(0)) = ⌫h|rv|2 i.

Thus, energy dissipation rate is trivially independent of Reynolds number! This does not,

however, mean that there is “dissipative anomaly,” which is the requirement that

D(Re) := "/(u3rms /L) ! D⇤ > 0, as Re ! 1.

Hence, there is no dissipative anomaly if

u2rms = h|v|2 i ! 1 as Re ! 1!

Bedrossian et al. introduce the notion of a “weak anomaly”, which occurs when

Re · D(Re) ! 1 as Re ! 1, (⇤)

or, in other words, D(Re) may vanish as Re increases but more slowly than D(Re) / 1/Re.

Equivalently, this means that the Taylor microscale / ⌫u2rms /" satisfies /L ! 0 as Re ! 1.

The interesting result obtained by Bedrossian et al. under the hypothesis (*) is that, for any

lengths `i , `d satisfying

`i /L ! 0 and `d / ! 1 as Re ! 1

then with hh·ii denoting both ensemble-averaging and angle-averaging

1 4
lim lim sup sup hh u3L (`)ii + " = 0.
`i /L!0 Re!1 `2[`d ,`i ] ` 5

Hence, it follows that the 4/5th-law holds to any desired degree of accuracy over the interval

[`d , `i ] for Re 1. An important implication of this result is that the validity of the 4/5th-law

cannot be taken as evidence for a (strict) dissipative anomaly.

29
Experiments and Simulations

? R. A. Antonia et al. et al., PRF, 4, 084602 (2019)

This paper presents data on the 4/5th-law from a compilation of laboratory experiments.

30
? K. R. Sreenivasan & B. Dhruva, Prog. Theor. Phys. Suppl. 130, 103–120 (1998)

This paper presents data from hot-wire measurements in the atmospheric boundary layer,

with Re in the range 10, 000 20, 000. The plot shows the ”Kolmogorov function”

K(r) = h u3L (r)i/"r and its “local slope” d log K(r)/d(log r).

31
? K. P. Iyer, K. R. Sreenivasan, and P. K. Yeung, PRF 5 054605 (2020)

This paper presents data from a 16, 3843 simulation of forced turbulence in a periodic domain,

with Re = 1300. The quantity h u3L (r)i was calculated by averaging over space, time, and

orientation angle of the displacement vector r.

? M. Taylor et al. Phys. Rev. E 68, 026310 (2003)

Another study with a 5123 DNS of homogenous forced turbulence, with Re ⇠

= 249 263.

This paper shows the importance of angle-averaging, obtaining results with such averaging

comparable to those at nearly twice the Reynolds number.

32
 TAYLOR, KURIEN, AND EYINK PHYSICAL REVIEW E 68, 026310 "2003#

FIG. 6. The nondimensional third-order longitudinal structure FIG. 7. The nondimensional third-order longitudinal structure
function, computed from a single snapshot of the stochastic dataset, function computed from a single snapshot of the deterministic
vs the nondimensional scale r/ & . The dots indicate the values of the dataset vs the nondimensional scale r/ & . The various symbols and
structure function computed at various !r j . The thick curve is the lines mean the same as in Fig. 6.
angle average. The horizontal line indicates the 4/5 mark.
are quite different, while the angle-averaged results are quite
reasonable and similar to each other as well as similar to the
aging in time. Excellent agreement is obtained in the inertial
results obtained from long-time averaging of the coordinate
range, with some departure at larger scales.
directions presented in Ref. $5% and shown for our data in
In Fig. 4, we show the second-order isotropy relation for
Sec. III D. Thus, we conclude that angle-averaging the data
our stochastic dataset, and in Fig. 5 we show the third-order
from a single snapshot yields a very reasonable result. Simi-
relation for the deterministic dataset. This data is computed
One last remark:over The derivation lar "not plotted#
resultsapplies are ifobtained
D(v)for⌘the0,4/3
[Link] vanishes
4/15
by angle-averaging a single snapshot ofthat we The
the flow. have given even
laws.
agreement is excellent, both in the inertial range and at the
largest scales. For
everywhere. Forcomparison,
example,thethis figures also show
holds in athesmooth solution ofC. the Euler equations, for which
same relations from the same snapshot but using only a Temporal variance
single
3 (r)icoordinate direction instead2 of angle-averaging.
3 so that In To illustrate the2 variance in time of the third-order longi-
h uthat there⇠areh significant
L case,ang T (r)i ⇠for
vL (r) vdifferences O(r
scales),well into D ` (v)structure
tudinal
) ! 0with
= O(`function, `!
as and 0. Another example
without angle-averaging,
the inertial range. Thus, the angle-averaging technique ap- we plot the peak value as a function of time for each dataset
is 2D
pearsEuler solutions
to be extremely where,
effective under very
in extracting general assumptions, D(v) ⌘ 0 and there is no energy
the isotropic
component of anisotropic data even at large scales, where
anisotropy
cascade to remains after timeE.g.
small scales. averaging over many snap-6 in Duchon & Robert (2000). There is a nontrivial
see Proposition
shots. Similar results were obtained for the second-order
isotropy relation from the deterministic dataset and for the
extension
third-order of the 45relation
isotropy -law to from2D the turbulence, but with h u3L (r)iang positive, corresponding to inverse
stochastic dataset.

energy cascade. E.g. see D. Bernard (1999).

B. Angle-averaging a single snapshot
We now present results using angle-averaging to compute
the third-order
Additional longitudinal structure function in the 4/5 law.
References:
Figures 6 and 7 show the result of the angle-averaging pro-
cedure described above for single snapshots of the stochastic
and deterministic datasets, respectively. The snapshots are
takenT.
aftervon Kármán
the flow & [Link] equilibrate.
has had time Howarth,The“On valuethe
of statistical theory of isotropic turbulence,”
the mean energy dissipation rate ! was calculated from the
snapshot.
Proc. ThisRoy.
is to be contrasted
Soc. [Link] A previous works in (1938).
164, 192-215
which ! is a long-time or ensemble average. We have there-
fore computed a version of the 4/5 relation which is local in FIG. 8. The angle-averaged "solid line# and single-direction
time. The dots represent the data from all 73 directions at all "dotted line# values of the peak of the nondimensionalized third-
[Link] N.
values Kolmogorov,
r that “Dissipation
were computed. The final weightedof energy
angle- inlongitudinal
order locally structure
isotropic
functionturbulence,” Dokl.
for deterministic dataset, as a
average of Eq. "14# is given by the thick curves in both Figs. function of nondimensional time t/T, where T!2E/ ' is the eddy-
6 andAkad.
7. One can see thatSSR
Nauk. the results
32,from different
16-18 directions
(1941). turnover time.

026310-6
33
 A. S. Monin, “Theory of locally isotropic turbulence,” Dokl. Akad. Nauk. SSSR

125 515-518(1959); see also, A.S. Monin & A. M. Yaglom, Statistical Fluid Me-

chanics, vol.2 (MIT, 1975), p.403.

Papers on Deterministic Versions of 45 -law

Q. Nie & S. Tanveer, “A note on the third-order structure functions in turbulence,”

Proc. R. Soc. A 455, 1615-1635(1999).

G. L. Eyink, “Local 45 -law and energy dissipation anomaly in turbulence,” Nonlin-

earity, 16 137-145(2003).

M. Novack, “Scaling laws and exact results in turbulence,” Nonlinearity 37 095002

(2024)

G. L. Eyink, “Onsager and the theory of hydrodynamic turbulence,” Rev. Mod.

Phys. 78 87-135(2006), Section IV, B.

Onsager’s unpublished notes: [Link]

2D Analogues of the 45 -law

D. Bernard, “Three-point velocity correlation functions in two-dimensional forced

turbulence,” Phys. Rev. E 60 6184-6187(1993).

A. M. Polyakov, “The theory of turbulence in two dimensions,” Nucl. Phys. B

396, 367-385(1993). This paper, in particular, discusses the analogy of the 45 -law

with conservation-law anomalies in quantum field theories.

34
The “Onsager Conjecture” and the h-Principle

What is now called the “Onsager conjecture” goes back to the following remark that Onsager

made at the very end of his 1949 paper on fluid turbulence:

“It is of some interest to note that in principle, turbulent dissipation as described

could take place just as readily without the final assistance by viscosity. In the

absence of viscosity, the standard proof of the conservation of energy does not apply,

because the velocity feld does not remain di↵erentiable! In fact it is possible to

show that the velocity field in such ‘ideal’ turbulence cannot obey any LIPSCHITZ

condition of the form

(26) |~v (r~0 +~r) ~v (r~0 )| < (const.)rn

for any order n greater than 1/3; otherwise the energy is conserved. Of course,

under the circumstances, the ordinary formulation of the laws of motion in terms of

di↵erential equations becomes inadequate and must be replaced by a more general

description; for example, the formulation (15) in terms of FOURIER series will do.

The detailed conservation of energy (17) does not imply conservation of the total

energy if the number of steps in the cascade is infinite, as expected and the double

sum of Q(~k, k~0 ) converges only conditionally.”

First, Onsager claims here that “it is possible to show that” that energy is conserved by ideal

(Euler) fluid equations if the Hölder exponent of the velocity is greater than 1/3 and he gives a

brief sketch of a proof using Fourier series. Second, Onsager remarks, after discussing the K41

theory in the preceding paragraphs, that “in principle, turbulent dissipation as described could

take place just as readily without the final assistance by viscosity”, so that he clearly believed

that Euler solutions with Hölder exponents 1/3 (or smaller) could dissipate kinetic energy.

35
 The first published proof of Onsager’s claim of conservation of kinetic energy for exponents

> 1/3 was given in the paper:

G. L. Eyink, “Energy dissipation without viscosity in ideal hydrodynamics, I. Fourier

analysis and local energy transfer,” Physica D 78 222–240 (1994)

which made rigorous Onsager’s heuristic argument using Fourier series but which required a

more restrictive notion than standard Hölder continuity. The paper

P. Constantin, W. E., and E.S. Titi, “Onsager’s conjecture on the energy conser-

vation for solutions of Euler’s equation”, Commun. Math. Phys. 165 207–209

(1994)

very concisely proved Onsager’s original statement on energy conservation for Euler solutions

with velocities in Hölder spaces but also proved conservation for velocities only in Besov spaces.

Their argument employed as a regularization a spatial mollification/low-pass filtering/coarse-

graining operation, which is a very general and powerful technique that we have exploited

throughout these lectures. Another important paper was

J. Duchon and R. Robert, “Inertial energy dissipation for weak solutions of incom-

pressible Euler and Navier-Stokes equations,” Nonlinearity 13 249–255 (2000)

which proved the Besov-space result using a point-splitting regularization that makes a con-

nection with the Kolmogorov 4/5th law. Remarkably, Onsager had performed calculations

identical to those of Duchon-Robert in his unpublished notes:

[Link]

see pp.14-18 in that folder. These results were never formally published but only communicated

in a private letter to T. von Kármán and C. C. Lin in 1945, which is reproduced and discussed

further here:

36
 K. R. Sreenivasan, “Onsager and the theory of hydrodynamic turbulence,” Rev.

Mod. Phys. 78 87–135 (2006)

The second remark of Onsager, that energy dissipation should be possible for Euler solutions

whose velocities have Hölder regularity  1/3, was almost certainly not proved by him in any

mathematical way but instead presumably suggested by the empirical evidence. The modern

form of Onsager’s conjecture in terms of “weak solutions” was stated in Eyink (1994). That

paper also constructed a simple example of a time-independent velocity field which showed

that the paper’s own proof of conservation (and also that of Constantin-E-Titi) could not be

extended to regularity exponents  1/3. Eyink (1994) then noted that

“It must not, of course, be concluded that, simply because our argument fails when

h  1/3, that non-conservation is actually possible for h  1/3. We emphasize that

to demonstrate this it is necessary to construct an appropriate solution v(., .) with

v(., t) 2 C h , 0 < h < 1/3 for t 2 [0, T ], for which the energy indeed decreases or

increases in the interval.”

This issue then lay dormant until 13 years later when preprints appeared by Camillo De Lellis

and László Székelyhidi Jr

C. De Lellis and L. Székelyhidi Jr, “The Euler equations as a di↵erential inclusion,”

arXiv:math/0702079 (2007) [published in Annals of Mathematics, 170 1417–1436

(2009)]

C. De Lellis and L. Székelyhidi Jr, “On admissibility criteria for weak solutions of

the Euler equations”, arXiv 0712.3288 (2007) [published in Arch. Ration. Mech.

Anal. 195 225–260 (2010)]

which initiated a long e↵ort that culminated in full proofs in 2016-2017 that dissipative Euler

solutions exist with any Hölder exponent h < 1/3 :

37
 P. Isett, “A proof of Onsager’s conjecture,” arXiv:1608.08301 (2016) [published in

Annals Math. 188 1–93 (2018)]

T. Buckmaster, C. De Lellis, L. Szkelyhidi Jr. and V. Vicol, “Onsager’s conjecture

for admissible weak solutions,” arXiv:1701.08678 (2017) [Commun. Pure Appl.

Math. 72 229-274 (2019)]

Remarkably, these developments are very closely connected with famous work of the math-

ematician John Nash on C 1 isometric embeddings:

J. Nash, “C 1 isometric imbeddings”, Ann. of Math. 60 383–396 (1954),

a connection which is very lucidly discussed in the following review papers:

C. De Lellis and L. Székelyhidi, Jr., “Continuous Dissipative Euler Flows and a

Conjecture of Onsager,” in Proceedings of the European Congress of Mathematics,

Kraków, 2012, edited by R. Latala, A. Ruciński, P. Strzelecki, J. Światkowski, and

D. Wrzosek (European Mathematical Society, Zurich, 2013), pp. 13–30.

[Link]

C. De Lellis and L. Szkelyhidi, Jr., “On turbulence and geometry: from Nash to

Onsager,” Notices Amer. Math. Soci. 5 677-685 (2019)

We give here a very succinct review, following the discussions in the previous references.

The paper of Nash (1954) addressed a classical problem of di↵erential geometry, whether

a smooth manifold M of dimension n 2 with Riemannian metric g may be isometrically

imbedded in m-dimensional Euclidean space Rm , that is, whether a C 1 embedding map u :

M ! Rm exists so that the Riemannian metric induced by the embedding agrees with g, or

@i u · @j u = gij . (⇤)

38
To answer this question, Nash considered a more general problem of short embeddings which

do not preserve lengths of curves on M but can only decrease lengths, so that

@i u · @j u  gij

in the matrix sense. The startling result obtained by Nash, with some improvement due to

N. H. Kuiper, On C 1 -isometric imbeddings. I, II, Nederl. Akad. Wetensch. Proc.

Ser. A. 58 = Indag. Math. 17 (1955), 545–556, 683–689.

is the following:

Nash-Kuiper Theorem: Let (M, g) be a smooth closed n-dimensional Riemannian

manifold, and let u : M ! Rm be a C 1 short embedding with m n + 1. For any

✏ > 0 there exists a C 1 isometric embedding u : M ! Rm with |u u|C 0 < ✏.

This result is surprising for two reasons. First, the condition (*) is a set of n(n + 1)/2 equations

in m unknowns. A reasonable guess would be that the system is solvable, at least locally, when

m n(n + 1)/2 and this indeed was a classical conjecture of Schläfli (1871). However, for

n 3 and m = n + 1, the system (*) is overdetermined! It is not obvious that there should be

any solutions at all, but the Nash-Kuiper Theorem shows that there exists a huge (C 0 -dense)

set of solutions in C 1 . Moreover, for n = 2 one can compare with classical rigidity results of

Herglotz and Cohn-Vossen for the so-called Weyl problem: if (S, g) is a compact Riemannian

surface with positive Gauss curvature and if u 2 C 2 is an isometric embedding of S into R3 ,

then u is uniquely determined up to a rigid motion! Thus it is clear that isometric embeddings

have very di↵erent qualitative behavior at low and high regularity (i.e. C 1 versus C 2 ). This

type of wild non-uniqueness at low regularity is a central aspect of the h-principle introduced

by mathematician Mikhail Gromov:

M. Gromov, “A topological technique for the construction of solutions of di↵erential

equations and inequalities,” [Link].(Nice 1970) 2 221-225 (1971)

39
 M. Gromov, Partial Di↵erential Relations, Ergebnisse der Mathematik und ihrer

Grenzgebiete (3) [Results in Mathematics and Related Areas (3)], vol. 9 (Springer-

Verlag, Berlin, 1986).

with the isometric embedding problem as a primary example. We shall not discuss here the

details of Nash’s proof of his remarkable result, but just remark that his construction of the

isometry u was in a series of stages, by adding at each stage a new small, high-frequency

perturbation. A brief popular account is here:

[Link]

written on the occasion of the award to Nash of the 2015 Abel Prize in mathematics.

The fundamental contribution of De Lellis and Székelyhidi Jr was to realize that there is

a very deep mathematical analogy between the problem of isometrically embedding a smooth

manifold by a map of low regularity and the problem of solving the Cauchy initial-value problem

for incompressible Euler equations by a velocity field of low regularity and that Nash’s method

of construction can be carried over to the latter. The analog of a “short mapping” for the Euler

system is what De Lellis and Székelyhidi Jr call a smooth subsolution, i.e. a smooth triple

(v, p, ⌧ ) with ⌧ a symmetric, positive-definite tensor such that

@t v + r · (v v + ⌧ ) = rp, r · v = 0.

This exactly corresponds to the incompressible Euler equations with the addition of a positive-

definite “turbulent stress” tensor ⌧ ! The fundamental theorem of Buckmaster et al. (2018) is

then the following:

Theorem: Let (v, p, ⌧ ) be any smooth, strict subsolution of the Euler equations on

T3 ⇥ [0, T ] and let h < 1/3. Then there exists a sequence (vk , pk ) of weak Euler

solutions such that vk 2 C h (T3 ⇥ [0, T ]) satisfy, as k ! 1,

Z Z Z Z
3 3 3
d x f vk ! d x f v, d x f vk vk ! d3 x f (v v + ⌧ )
T3 T3 T3 T3

40
 for all f 2 L1 (T3 ) uniformly in time, and furthermore for all t 2 [0, T ] and all k
Z Z
1 1
d x |vk |2 =
3
d3 x (|v|2 + Tr ⌧ ).
T3 2 T3 2

e ` (x + r), then the convergence property can be restated as

If we take f (r) = G

e` ,
ek,` ! v
v ⌧e` (vk , vk ) ! ⌧e` (v, v) + ⌧e ` (cf. Germano’s identity!)

This result is an h-principle for weak Euler solutions with Hölder regularity h < 1/3. Clearly,

there is a huge number of subsolutions, since one may add any positive definite tensor ⌧

whatsoever. An immediate consquence is therefore

Corollary: Let e : [0, T ] ! R+ be any strictly positive, smooth function. Then for

any 0 < h < 1/3 there exists a weak Euler solution v 2 C h (T3 ⇥ [0, T ]) such that
Z
1 2
d3 x |v| = e(t).
T3 2

In particular, one may take e(t) to be any function strictly decreasing in time and then the

Euler solution of the Corollary (globally) dissipates kinetic energy.

The same “convex integration methods” have also implications for non-uniqueness of the

Cauchy initial-value problem, as discussed already by C. De Lellis and L. Székelyhidi Jr (2010).

The Theorem 1 from that paper shook many previous expectations:

Theorem: Let d 2. There exist compactly-supported divergence-free vector fields

v0 2 L1 for which there are infinitely many weak Euler solutions with that initial

data, satisfying both the strong energy equality

Z Z
2 n
|v(x, t)| d x = |v(x, s)|2 dd x, for all t > s
Rd Rd

and the local energy equality

✓ ◆ ✓ ◆
1 2 1 2
@t |v| +r· |v| + p v = 0.
2 2

41
 Furthermore, there are infinitely many weak Euler solutions with that initial data

satisfying the strong energy inequality:

Z Z
2 d
|v(x, t)| d x < |v(x, s)|2 dd x, for all t > s
Rd Rn

This result showed that one cannot add a local energy inequality
✓ ◆ ✓ ◆
1 2 1 2
@t |v| +r· |v| + p u  0,
2 2

(or even an equality) and obtain a unique weak solution for the Euler equations for certain L1

initial data. Furthermore, non-uniqueness occurs even if total energy is strictly decreasing.

Note that such initial data with non-unique solutions cannot be smooth (say, C 1,✏ or higher

smoothness) because this would violate the following important type of result:

Theorem: Let v 2 L1 ((0, T ); L2 (Td )) be a weak Euler solution, V 2 C 1 (Td ⇥ [0, T ])

a strong solution, and assume that v and V share the same initial datum v0 . Assume

moreover that
Z Z
2 d
|v(x, t)| d x  |v0 (x)|2 dd x (#)
Td Td

for almost every t 2 (0, T ). Then v(x, t) = V(x, t) for almost every (x, t).

Results of this type go by the name of strong-weak uniqueness. For an excellent review, see:

E. Wiedemann, “Weak-strong uniqueness in fluid dynamics,” in: Partial Di↵erential

Equations in Fluid Mechanics, London Mathematical Society Lecture Note Series,

vol. 452, Eds. Fe↵erman, C. L., Robinson, J. C., Rodrigo, J. L., & Diez, J. L. R.

(Cambridge University Press, 2018). [Link]

The conclusion of such results is that any “admissable” weak Euler solution satisfying the weak

energy inequality (#) must coincide with a classical Euler solution, as long as that exists. Note

that strong-weak uniqueness applies also to “measure-valued weak Euler solutions” such as

constructed by DiPerna-Majda (1987): see the Wiedemann (2018) review.

42
 These non-uniqueness results have since been considerably extended and are still currently

under active investigation. Some very important results are contained in this paper:

S. Daneri, E. Runa, and L. Székelyhidi Jr, “Non-uniqueness for the Euler equations

up to Onsager’s critical exponent,” Annals of PDE 7 8 (2021)

who prove:

Theorem: For any h 2 (0, 1/3), there is a set of divergence-free vector fields v0 2

C h (T3 ) which is a dense subset of the divergence-free vector fields in L2 (T3 ) such that
0
infinitely many Euler solutions exist with that initial data for which v(t) 2 C h (T3 )

for all h0 < h, t 2 (0, T ) and for which the weak energy inequality (#) holds.

This theorem shows that the non-uniqueness holds right up to the critical Onsager 1/3 exponent

and for a dense set of initial data. Thus, uniqueness is in some sense “typical”. To our knowl-

edge, it is not yet known if the same result still holds if solutions are not merely “admissable”

but instead satisfy a local energy inequality. A partial result in this direction was proved by

P. Isett, “Nonuniqueness and existence of continuous, globally dissipative Euler

flows,” Arch. Ration. Mech. Anal. 244 1223–1309 (2022)

whose Theorem 6 states that:

Theorem: For any h < 1/15, the set of divergence-free initial data v0 2 C(T3 )

that admit infinitely many incompressible Euler solutions of class C h that obey local

energy conservation
✓ ◆ ✓ ◆
1 2 1 2
@t |v| +r· |v| + p v = 0
2 2

on a common interval of time containing 0 is C 0 dense.

This situation must be contrasted with that for somewhat smoother initial data, for example

v0 2 C h with h > 1, where unique solutions of the incompressible Euler equations exist (at least

43
locally in time) that satisfy local kinetic energy conservation and possess the same regularity

as the initial data. The non-uniqueness of weak solutions to the Cauchy problem with lower

regularity is a “Nash non-rigidity” phenomenon, implying an essential unpredictability.

The constructions of such weak Euler solutions follow a strategy similar to that of Nash, by

a sequence of stages v0 , v1 , v2 , .... At stage n one has, after coarse-graining, a subsolution

@t vn + r · (vn vn + ⌧ n ) = rpn

which is supported on wavenumbers < 2n . By adding a small-scale carefully chosen perturbation

one can (together with other operations, such as evolving under smooth Euler dynamics locally

in time and “gluing” the di↵erent time-segments) succeed to cancel a large part of the stress ⌧ n

so that, in the limit, ⌧ n ! 0 weakly and one obtains a weak limit v which is a distributional

Euler solution. This is therefore a kind of “inverse renormalization group” procedure in physical

terms or a “super-resolution” method in the language of machine learning.

These Euler solutions, one must stress, are not obtained in the physically relevant manner by

taking zero-viscosity limits of smooth solutions of the incompressible Navier-Stokes equations.

One of the outstanding issues is to show that similar weak Euler solutions are obtained as

viscosity tends to 0, at least along suitably selected subsequences of viscosities. There has been

recent progress, but with rather rough forcing:

Bruè, E. & De Lellis, C., “Anomalous dissipation for the forced 3D Navier–Stokes

equations,” Commun. Math. Phys. 400 1507–1533 (2023).

Another open problem is to formulate suitable “admissibility/selection conditions” that inviscid

limits must possess and that uniquely characterize those solutions (or classes of such solutions).

Some relatively simple proposals are ruled out by the previous results. For example, the local

dissipation condition
✓ ◆ ✓ ◆
1 2 1 3
@t u + @x u 0
2 2
selects a unique weak solution of inviscid Burgers equation, under very modest regularity as-

sumptions. This is not true (even with = 0!) for incompressible Euler solutions in C h , h < 1/15.

44
 APPENDIX I: LERAY SOLUTIONS OF INCOMPRESSIBLE NAVIER-STOKES

The purpose of this appendix is to discuss more carefully the issue of energy balance/energy

dissipation for Leray Solutions of the incompressible Navier-Stokes (INS) equations, which are

possibily singular. It was proved by

J. Leray, “Essai sur Le Mouvement d’un fluide visqueux emplissant l’espace,”

Acta. Math. 63 193-248 (1934)

that solutions of the INS

@t v⌫ + r · (v⌫ v⌫ ) = rp⌫ + ⌫4v⌫ , r · v⌫ = 0

always exist in the sense of spacetime distributions, i.e.

Z T Z
dd x[@t '(x, t) · v⌫ (x, t) + r'(x, t) : v⌫ (x, t) ⌦ v⌫ (x, t)
dt
t0 Z
+ (r · '(x, t))p (x, t) ⌫4'(x, t) · v (x, t)] = dd x '(x, t0 ) · v0⌫ (x, t),
⌫ ⌫

Z T Z
dt dd xr (x, t) · v⌫ (x, t) = 0
t0

for any initial condition v0⌫ with finite energy: kv0⌫ kL2 (V ) < +1. Here ' = ('1 , · · · , 'd ), are

C 1 functions on spacetime with compact support. Note that the above rather abstract-looking

formulation is actually very physical! It is equivalent to the following balance equation for the

momentum in a bounded region ⌦ ✓ V with smooth boundary @⌦ and times t > t0 t0

Z Z
v⌫ (x, t)dd x v⌫ (x, t)dd x
⌦Z
t Z  ⌦
@v⌫
= d⌧ v (x, t)(v⌫ (x, t) · n̂) + p⌫ (x, t)n̂
⌫
⌫ (x, t) dA,
Z t 0 @⌦ Z @n
⌫
v (x, t) · n̂ dA = v⌫ (x, t0 ) · n̂ dA = 0
@⌦ @⌦

for all possible choices of ⌦, t, t0 . These equations just state that the momentum change in a

bounded region ⌦ comes from flux of momentum across the surface @⌦ and that there is no net

flux of mass across the surface.

45
In addition, Leray showed that his distributional solutions satisfy the following fundamental

energy inequality for all t t0

1
Rt
⌫ 2
2 kv (t)kL2 (V ) +⌫ t0 d⌧ krv⌫ (⌧ )k2L2 (V )  12 kv0⌫ k2L2 (V )

This inequality states that the total energy at time t plus the integrated energy dissipation

up to time t must be less than or equal to the initial energy 12 kv0⌫ k2L2 (V ) . If the solutions are

everywhere smooth (“strong”), then the inequality “” becomes equality “=”. However, if

the solutions are singular (“weak”), then there may be strict inequality, which corresponds to

“extra dissipation” due to the singularities, in addition to the viscous dissipation. Note that it

is a consequence of this fundamental energy inequality that

sup kv⌫ (t)kL2 (V )  kv0 kL2 (V ) < +1

t0 tT
Z T kv0 k2L2 (V )
dtkrv⌫ (t)k2L2 (V )  < +1
t0 2⌫

for a fixed initial condition v0 with finite energy. From these estimates, some other basic bounds

follow, such as
RT 2
t0 dtkv⌫ (t)k3L3 (V )  (const.) L⌫ kv0 k3L2 (V ) .

This inequality means that Leray solutions are sufficiently regular that one can consider a local

energy balance in spacetime. To construct his solutions, Leray considered the limit as ` ! 0 of

the modified equation

@t v + (v̄` · r)v = rp + ⌫4v

where only the velocity v̄` = G` ⇤ v appearing in the advection term has been smoothed. Leray

showed that the above equation has regular solutions (v`? , p?` ), which lie in a bounded (and thus

weakly compact) subset of L2 ([0, T ], H 1 (V )). Hence, weak limits v`? ! v exist in this space,

along subsequences of `. Such v can be seen to be distributional solutions of the incompressible

Navier-Stokes equation

@t v + r · (vv) = rp + ⌫4v

and, since
R R R R
lim`!0 dd x dt|rv`? (x, t)|2 (x, t) dd x dt|rv(x, t)|2 (x, t)

46
for non-negative test functions , also obeys the energy balance
⇥ ⇤
@t ( 12 |v|2 ) + r · ( 12 |v|2 + p)v ⌫r( 12 |v|2 ) = D(v) ⌫|rv|2

with D(v) 0. By smoothing the solution v one can write an energy balance also for v̄` :

1 1 1
@t ( v · v̄` ) + r · [( v · v̄` )v + (pv̄` + p̄` v)
2 2 2
1 2
1 1
+ ((|v| v)` (|v|2 )v ⌫r v · v̄` ]
4 Z 4 2
1
= d r rG` (r) · v(r)| v(r)|2 ⌫rv : rv̄`
d
4`

Taking the limit ` ! 0, we recover the previous energy balance with an explicit expression for

D(v):
1
R
D(v) = lim`!0 4` dd r(rG)` (r) · v(r)| v(r)|2

This formula makes it clear that D(v) in the Navier-Stokes solution is connected with velocity

increments. Note that, in general, in the presence of such singularities, the total Navier-Stokes

dissipation is

D(v⌫ ) + ⌫|rv⌫ |2

(where we have now added the supperscript ⌫ to indicate the viscosity value) and in the ⌫ ! 0

limit it is this total dissipation which converges to the anomalous dissipation in the Euler

solution

D(v⌫ ) + ⌫|rv⌫ |2 ! D(v).

For more details, see Duchon & Robert (2000).

Note, however, that according to the arguments presented in these lectures, weak solutions

of the incompressible Navier-Stokes equation à la Leray are unphysical, because they assume

that the deterministic, continuum fluid-dynamical description holds down to arbitrarily small

length-scales. This is not a valid idealization, because thermal noise e↵ects appear already at

the Kolmogorov scale and the hydrodynamic description is fundamentally limited to scales `

larger than the mean-free-path length. Attempting to take the limit ` ! 0 leads to difficult

mathematical problems with, however, no relevance to real-world molecular fluids.

47
General References on Leray Solutions of INS

An English translation of Leray’s paper of 1934 has been made by mathematician Bob Terrell:

[Link]

A key paper on “partial regularity” of Leray solutions, improving the earlier work of V. Sche↵er

(1977) is

L. Cafarelli, R. Kohn & L. Nirenberg, “Partial regularity of suitable weak solutions

of the Navier-Stokes equations,” Commun. Pure Appl. Math. 35, 771-831 (1982)

Many good textbooks presentation of the Leray theory are available. I’d recommend

R. Temam, Navier-Stokes Equations: Theory and Numerical Analysis (AMS, 2001)

and

R. Temam, Navier-Stokes Equation & Nonlinear Functional Analysis (SIAM, 1995)

especially for people interested in computational fluid mechanics. A good book for physicists is

G. Gallavotti, Foundations of Fluid Dynamics, Volume 172, Issue 5998 of Theoret-

ical and Mathematical Physics (Springer Science & Business Media, 2002)

48