# Verona – 14-17 Sept.

, 2009

Renzo L. Ricca
Department of Mathematics & Applications, U. Milano-Bicocca, ITALY E-mail: renzo.ricca@unimib.it – URL: http://www.matapp.unimib.it/~ricca renzo. ricca@ unimib.it http://www.matapp.unimib.it/~ricca

Aims

Theoretical goals: - describe and classify complex morphologies; - study possible relationships between energy and complexity; - understand and predict energy localization and transfer; Applications: - implement new visiometric tools and diagnostics; - develop real-time energy analysis of dynamical processes.

Coherent structures
Leonardo da Vinci (Water studies, 1506)

Werlé, ONERA, 1974 (Van Dyke, 1982)

Vorticity localization in classical and quantum fluids
Kida et al. (Toki-Kyoto, 2002)

Miyazaki et al. (Physica D, 2009)

Geometric approach to vortex filament motion Localized induction approximation (LIA) and intrinsic kinematics Solitons and integrable vortex dynamics Stationary solutions Topological properties and fluid invariants Structural complexity analysis of vortex dynamics

χ

Geometry and topology of fluid flows Measures of morphological complexity Energy/complexity relations Shapefinders and eigenvalue analysis Dynamical properties in terms of graph analysis

T

Localized induction approximation (LIA) homogeneous incompressible inviscid

fluid in

3 :

u = u ( X,t )

ω =∇×u 3 ∇ ⋅ u = 0 in  u = 0 as X → ∞

Space curve

Ct , given by:

Ct :

X(s,t) := Xt (s) ∈ Cν
s ∈ [0,L] → R 3

ˆ t
ˆ n

Intrinsic reference on (Frenet frame)

Ct , given by:
ˆ ˆ {tˆ,n, b}
,

ˆ b Ct

∂X , ˆ t := X ′(s,t) = ∂s

Vortex line on

ˆ Ct : ω = ω 0t

ω 0 = constant

asymptotic theory ( R /a 1) no self-intersections

LIA:

ˆ ˙ (s,t) ≡ ∂X ≈ X ′ × X ′′ = cb ≡ u LIA X ∂t
(Da Rios, 1906; Hama & Arms, 1961)

Intrinsic equations under LIA and NLSE

Intrinsic description: under LIA:

u = (ut ,un ,ub ) , c ( s,t )
.

curvature,

τ ( s,t ) torsion;

ˆ u LIA = cb , ut = un = 0 , ub = c
˙ c = −(cτ )′ − c ′τ
⎛ c ′′ − cτ 2 ⎞′ ˙ τ =⎜ ⎟ + c ′c c ⎝ ⎠

Ct := ϕ t (C)

Da Rios, 1906 Betchov, 1965

NLSE via Madelung transform:

τ

ψ ( s,t ) = c ( s,t ) e
NLSE:

i

∫0 τ (ξ , t ) dξ
s

2
c

ℑm

ℜe

ψ

˙ + ψ ′′ + 1 ψ 2ψ = 0 iψ 2

(Hasimoto, 1972)

Da Rios-Betchov eqs. from NLSE: by taking log-derivative of ψ ( s,t )

Ct := ϕ t (C)

˙ c = f c (s,t) ˙ τ = fτ (s,t)

Higher-order LIAs and integrable geometric dynamics

Higher-order effects:

ˆ ˆ ˆ ˆ ˆ u = (α + βs)cb + (γ + δs)t + µ 1 c 2t + c ′n + cτb 2

(

)

(Lakshmanan & Ganesan, 1985) (Onuki, 1985)

β ,δ : inhomogeneities

µ

γ : non-linear stretching

(Fukumoto & Miyazaki, 1991)

: axial flow and vorticity

Higher-order LIAs:

u ( 0) = u LIA
u ( j +1)
j +1

⎛ ∂ ν u ( 0) ⎞ = ∑ F ⎜ ν ⎟ = R (u ( 0) ) ⎝ ∂s ⎠ ν =1

˙ ψ = P ψ ( un e iΘ ) + Q ψ ( ub e iΘ )
(Nakayama et al., 1992)

(Langer & Perline, 1991)

Integrable geometric dynamics:

u = ( ut ,un ,ub )

Ct := ϕ t (C)

˙ = f s (t ) s ˙ c = f c (s,t) ˙ τ = fτ (s,t)

Germano, 1983 Ricca, 1991

Conserved quantities from integrable dynamics
A = ∫ X × X ′ ds = const.

Arms & ; Hama, 1965

L = ∫ ds = const.

Marsden & Weistein, 1983

Geometric invariants:

P (L) = ∑ I j L j
j= 0

I1 = ∫ c 2 ds
⎛ c4 2 2 2⎞ I 3 = ∫ ⎜ − c ′ − c τ ⎟ ds ⎝4 ⎠

I 0 = ∫ τ ds

I 2 = ∫ c 2τ ds

(Langer & Perline, 1991; Ricca, 1991)

Physical conserved quantities:
kinetic energy: pseudo-helicity: Fukumoto, 1987 Ricca, 1992 enstrophy:

... ...

1 2 K = ∫ u d3 X 2 H = ∫ u ⋅ (∇ × u ) d 3 X

E = ∫ ω d3 X 1 P = ∫ ( X × ω )d 3 X linear momentum: 2 1 M = ∫ X × ( X × ω )d 3 X angular momentum: 3

Structural complexity analysis of 3D vortex tangles

experiment or simulation

domain extraction

projected diagram analysis

Measures of structural complexity: Algebraic information: - average crossing number Topological information: - minimal crossing number - linking numbers - topological changes Geometric information: - tropicity directions - coiling, writhing - alignment - signed area Dynamical systems analysis: - topological entropy - eigenvalue analysis

Energy and helicity of a vortex tangle

2 1 3 3 Kinetic energy: E(T ) ≡ ∫T u d X = ∫T u ⋅ ( X × ω ) d X ; 2 ˆ ˆ ti ⋅ t j Γ2 from Lamb (1932), we have: E(T ) ≈ ∑ ∫∫ X − X dsi ds j , 8π ij χi χ j i j

with total length of vortex filaments given by L(T ) =

∑ ∫ tˆ
i

i

dsi .

χi

ˆ If ω ≡ ∇ × u = ϖ 0t , then the kinetic helicity is given by
(Moffatt, 1969; Moffatt & Ricca, 1992):

H (T ) ≡ ∫ u ⋅ ω d 3 X = ∑ Lki Γ i2 + 2∑ Lkij Γ i Γ j
T
i i≠ j

Lki = Lk ( χ i , χ i ) = Wri + Twi
Lkij = Lk χ i , χ j

Calugareanu-White invariant Gauss linking number i-th vortex circulation

where

(

)

Γi

Tangle analysis by indented projections

ˆ ˆ Let Πi = Π(Ti ) be the “indented” Ti –projection of the oriented tangle
component χ i ; assign the value ε r = ±1 to each apparent crossing in Πi .

writhing:
Wri = Wr( χ i ) =
r ∈χ i

T =  χi

∑ε

r

, Wr = Wr(T ) =

r ∈T

∑ε

r

;

i

Lkij = Lk( χ i , χ j ) = 1 ∑ εr , 2 r ∈χi ∩ χ j
i≠ j

Lktot =

r ∈T
i≠ j

∑ Lk

χ χ ii
;
ˆ Ti

ij

average crossing number:
r ∈χ i ∩ χ j

Cij = C( χ i , χ j ) =

εr ,

C=

r ∈T

∑C

ij

;
Πi

−1 −1

+1

estimated values: ⎛ ⎞ Wr⊥ = ⎜ ∑ε r ⎟ , ⎝ r ∈T ⎠ ⊥

⎛ ⎞ C⊥ = ⎜ ∑ ε r ⎟ . ⎝ r ∈T ⎠⊥

A vortex tangle test case (Barenghi, Ricca & Samuels, 2001)

ABC-type flow field super-imposed on initial vorticity distribution

t =0

t = 0.015

t = 0.050

t = 0.087

Vortex tangle test case: energy-complexity relation

logH

C(t) ∝ [ E(t)]

2

logC⊥
logC
logWr⊥
logLk tot

O(165 s−1 )

O(83 s−1 )

log(E / E 0 )

log(L /L0 )
mature tangle

time

Energy-complexity relation

C(t) ∝ [ E(t)] : since
2

we have

ˆ ˆ Xi − X j ⋅ t i × t j 1 C= dsi ds j ∑ ∫∫ 3 , 4π ij χi χ j Xi − X j
2

(

)

⎛ δs ⎞ δC ∝ ⎜ ⎟ ⎝ Xi − X j ⎠
hence

,

and

δs δ (E / E0 ) ∝ Xi − X j

;

C ∝ (E / E0 )2 .

Helicity-complexity bounds If Lkii = 0 ,

Γi = Γ
2

∀χ i ∈T , then we have (Ricca, 2008):
and

H (t) ≤ 2Γ C(t)

d H (t) 2 dC(t) ≤ 2Γ dt dt

.

Tropicity interpretation of eigenvalues

Eigenvalue analysis: let λ1 ≥ λ2 ≥ λ3 . From (Vilanova et al., 2006), we have:
Cf = 2 ( λ2 − λ 3 ) λ1 − λ2 3λ3 , Cp = , Cs = . ∑ λi ∑ λi ∑ λi
i i i

Shapefinders: let

⎛ 1 1 1⎞ 2 V = ∫ d X , A = ∫ d X , H G = ∫ ⎜ + ⎟ d X, D ∂D 2 ∂D ⎝ R1 R2 ⎠
3 2

1 1 2 χE = ∫ d X. ∂D R R 2 1 2

Then If

HG L= , 4πχ E

A W= , HG

3V T= ; A

(Sahni et al., 1998)

D

is convex and L ≥ W ≥ T > 0 , we can define
L −W , CF = L +W +T 2(W − T ) 3T , CS = , CP = L +W +T L +W +T

so that

λ1 ∝ L , λ2 ∝ W , λ3 ∝ T .

Interpretation of momenta in terms of projected areas

Linear momentum:

1 1 3 P ≡ ∫ X × ω d X = ∑ Γ i  X × X ′ ds , ∫ χi 2 T 2 i

Angular momentum:

1 1 3 M ≡ ∫ X × ( X × ω ) d X = ∑ Γ i  X × ( X × X ′ ) ds . ∫ χi 3 T 3 i

Under Euler (and LIA) equations:

dP =0 , dt

dM =0 . dt

Plane graph by standard projection

Λ ≡ p(T ) :  → 
3

2

Λ xy ≡ pz T

Axy

( ) ≡ A (Λ ) xy
xy

,

...
yz ;

,

...
zx

.

Then (Arms & Hama, 1965; Ricca, 1992):

Pxy = Γ A xy , Pyz = Γ A yz , Pzx = ΓA zx

M xy = Γ dzA xy , M yz = ΓdxA yz , M zx = Γ dyA zx .

Signed area and weight of projected graphs Let Λ j = p( χ j ) and R j ≡ int Λ j

Cauchy index I j = I Λ j : at each

  intersection ρ  Λ j assign ε i = ±1
and take

( )

(  Π) :
.

Γj

χj
p
ˆ ρ

Ij =I Λj =

( ) ∑ε
ˆ ρ

 Λj

i

Π

ˆ t

R j +1

Λj

Example

+1

0 –1 –2 –1

Signed area:

A j ( Λ ) = ∑I jA j (R j )
j

Weighted circulation: γ j =

∑ k Γ k Lk, j Lj

Momenta by signed area interpretation: geometric method (Ricca, 2008) Let T = χ i be a vortex tangle. Then, by considering the associated i principal projected graphs, we have

P = Pxy , Pyz , Pzx : Pxy = ∑ γ jI jAxy R j ,
j

(

)

( )

Pyz = ... , Pzx = ... ;

M = M xy , M yz , M zx : M xy = dz ∑ γ jI jAxy R j ,
j

(

)

( )

M yz = ... , M zx = ... .

Examples

(a)

(b)

Vortex analysis by geomertic method

Following Aref & Zawadzki (Nature, 1991):

+1

:

+1 +2

before

after

weighted areas

Zero-area momentum (P = 0) interaction:
–1 0 +1 –1 + – +1 – +

Head-on collision and breakdown of vortex rings (Lim, Nature, 1992)

Head-on collision of vortex rings (Lim, Nature, 1992): Re=1071

Head-on collision of vortex rings (Lim, Nature, 1992): Re=1573

Selected references