RELAXATION AND ENERGY
DISSIPATION IN TDCDFT
Roberto DAgosta
UPV/EHU and Ikerbasque
San Sebastian
[Link]@[Link]
RELAXATION IN QUANTUM MECHANICS
Common approach (effective non unitary time evolution, e.g.
Caldeira-Leggett)
Time Dependent Density Functional Theory approach
(Unitary Time Evolution)
How to introduce
a time arrow in a
Hamiltonian
framework?
Answer: The
functional depends
on the velocity eld
The Kohn-Sham Hamiltonian for TDCDFT
The Vignale-Kohn exchange-correlation eld
In principle, exact time evolution of particle and current
densities
H
KS
=
1
2m
X
i
[ p
i
+ eA(r, t) + eA
xc
(n, j)]
2
+
Z
V (r, t) + V
ALDA
xc
(r, t)
n(r, t)dr
e
dA
xc
dt
=
xc
n
v =
j
n
xc,ij
=
i
v
j
+
j
v
i
2
3
ij
v
+
ij
v
THE KOHN-SHAM ENERGY FUNCTIONAL
R. DAgosta and G. Vignale, Phys. Rev. Lett. 96, 016405 (2006)
dE
dt
0 t
dE
dt
=
Z
j(r, t) E
xc
(r, t) dr
eE
xc
(r, t) =
dA
xc
dt
E[n,
~
j] = h
KS
|
H
KS
|
KS
i
Z
V
hxc
(r, t)n(r, t) dr +E
ALDA
hxc
[n]
We assume vanishing velocity at the
boundaries
is NOT the total energy but gives a time arrow E
Numerical results
R. DAgosta and G. Vignale, Phys. Rev. Lett. 96, 016405 (2006)
~
>0, t <0
~
=0, t > 0
-L/2
-L/2
L/2
L/2
1D code no memory.
For similar results in the case of
systems with memory look at
H. Wijewardane and C. Ullrich, Phys. Rev. Lett. 95, 086401 (2005)
Quantum Well
E
i
E(t) E
0
e
t
z
y
x
1D Quantum well in
the z direction
2D electron gas in
the (x,y) plane
Energy transfer from the motion in the z direction to
the 2D electron gas
Energy transfer
By using standard time-dependent perturbation theory one can test the
approximations for the xc kernel.
THERMODYNAMICS
The system QW+2DEG is closed and isolated
First principle of Thermodynamics
Q =
Z
T
f
T
i
C
2DEG
V
(T)dT
C
2DEG
V
=T
S = (T
f
T
i
)
T
f
T
i
=
s
2
(E
i
E
0
)
Then:
R. DAgosta and G. Vignale, Phys. Rev. Lett. 96, 016405 (2006)
{
DFT
E
i
E
0
=Q =
Z
0
dE
dt
dt
TIME EVOLUTION
Energy
conservation
dE
dt
+T(t)
dS
dt
= 0
Z
0
dE
dt
+T(t)
dS
dt
dt = 0
quasi-
static
T(t) = T
i
+
s
2
[E
i
E(t)]
dS
dt
=
dS
dT
dT
dt
=C
V
(T)
dT
dt
R. DAgosta and G. Vignale, Phys. Rev. Lett. 96, 016405 (2006)
S(t) =
p
2[E
i
E(t)]
CONCLUSION - 1
TDCDFT is able to capture the relaxation of a many-
particle system to a statistical equilibrium
The corresponding time evolution is unitary and does
not contain ad hoc parameters
At least in a simple case we are able to interpret the
Kohn-Sham energy as the maximum work one can
extract from the system
TDCDFT offers a natural framework where to study the dynamical time
evolution of a many-particle system
OPEN QUANTUM SYSTEMS
Van Kampen, Stochastic Processes in Physics and Chemistry
bath l(t)
The bath induces stochasticity and:
OPEN QUANTUM SYSTEMS
Van Kampen, Stochastic Processes in Physics and Chemistry
bath l(t)
The bath induces stochasticity and:
is macroscopic
OPEN QUANTUM SYSTEMS
Van Kampen, Stochastic Processes in Physics and Chemistry
bath l(t)
The bath induces stochasticity and:
is macroscopic
is unbiased
OPEN QUANTUM SYSTEMS
Van Kampen, Stochastic Processes in Physics and Chemistry
bath l(t)
The bath induces stochasticity and:
l(t) = 0
l(t)l(t
) = (t t
V =
V (t)
is assigned
no feedback
is macroscopic
is unbiased
has no memory
Open Quantum Systems
bath l(t)
The system is in a mixed state and:
is out of equilibrium
is interacting
has memory
Open Quantum Systems
bath l(t)
two formalisms
density matrix
stochastic
Schrdinger eq.
The system is in a mixed state and:
is out of equilibrium
is interacting
has memory
STOCHASTIC
SCHROEDINGER EQ.
t
(r, t) = i
H(r, t)
1
2
V
V (r, t) + l(t)
V (r, t)
STOCHASTIC
SCHROEDINGER EQ.
t
(r, t) = i
H(r, t)
1
2
V
V (r, t) + l(t)
V (r, t)
}
}
dissipation uctuation
STOCHASTIC
SCHROEDINGER EQ.
t
(r, t) = i
H(r, t)
1
2
V
V (r, t) + l(t)
V (r, t)
}
}
dissipation uctuation
(t) =
i
p
i
|
i
i
|
...
p
1
|
1
p
2
|
2
p
M
|
M
STOCHASTIC
SCHROEDINGER EQ.
t
(r, t) = i
H(r, t)
1
2
V
V (r, t) + l(t)
V (r, t)
}
}
dissipation uctuation
(t) =
i
p
i
|
i
i
|
...
p
1
|
1
p
2
|
2
p
M
|
M
BOSE GAS IN A TRAP
Gross - Pitaevskii equation
i
t
(x, t) =
1
2m
d
2
dx
2
+
1
2
m
2
0
x
2
+ gn(x, t)
(x, t)
Davis, et al. PRL (2005),
Anderson, et al. Science (2005)
13
BOSE GAS IN A TRAP
Gross - Pitaevskii equation
How to simulate the dynamics of the condensate?
i
t
(x, t) =
1
2m
d
2
dx
2
+
1
2
m
2
0
x
2
+ gn(x, t)
(x, t)
Davis, et al. PRL (2005),
Anderson, et al. Science (2005)
13
1-D CONFINED BOSE GAS
0 1 1 1 . . .
0 0 0 0 . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0 0 . . .
R. DAgosta and M. Di
Ventra PRB (2008)
.
.
.
E
E
E
E
E
m
0
14
1-D CONFINED BOSE GAS
0 1 1 1 . . .
0 0 0 0 . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0 0 . . .
R. DAgosta and M. Di
Ventra PRB (2008)
.
.
.
E
E
E
E
E
m
0
No interaction,
g=0
14
0 5 10 15 20
0
0.2
0.4
0.6
0.8
1
single run
10 runs
50 runs
100 runs
density matrix
p
0
t!
0
0 5 10 15 20
0
0.04
0.08
t!
0
|p
0
-p
0
|/p
0
dm sse sse
1-D CONFINED BOSE GAS
0 1 1 1 . . .
0 0 0 0 . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0 0 . . .
R. DAgosta and M. Di
Ventra PRB (2008)
.
.
.
E
E
E
E
E
m
0
No interaction,
g=0
14
1-D CONFINED BOSE GAS
0 1 1 1 . . .
0 0 0 0 . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0 0 . . .
R. DAgosta and M. Di
Ventra PRB (2008)
.
.
.
E
E
E
E
E
m
0
No interaction,
g=0
0
0.1
0.2
0.3
0.4
0.5
!
"
#$"
a)
n
x
0
0
0.1
0.2
0.3
!
"
#%
b)
n
x
0
-15 -10 -5 0 5 10 15
0
0.1
0.2
x/x
0
!
"
#"
c)
n
x
0
14
Strong interaction
0 10 20 30 40 50 60
0
0.1
0.2
0.3
0.4
0.5
0.6
O
c
c
u
p
a
t
i
o
n
p
r
o
b
a
b
i
l
i
t
y
p
0
p
1
t
0
p
2
1-D CONFINED BOSE GAS
R. DAgosta and M. Di Ventra
PRB (2008)
15
g/
0
= 5
Strong interaction
0 10 20 30 40 50 60
0
0.1
0.2
0.3
0.4
0.5
0.6
O
c
c
u
p
a
t
i
o
n
p
r
o
b
a
b
i
l
i
t
y
p
0
p
1
t
0
p
2
1-D CONFINED BOSE GAS
R. DAgosta and M. Di Ventra
PRB (2008)
-4 -2 0 2 4
0
0.05
0.1
0.15
0.2
Stochastic
Thomas - Fermi
x/x
0
n
x
0
15
g/
0
= 5
SPIN CHAIN
0 50 100 150 200 250
time
-0.0015
-0.001
-0.0005
0
0.0005
0.001
0.0015
e
n
e
r
g
y
c
u
r
r
e
n
t
j
2,3
j
1,2
T
L
> T
R
EQUATION OF MOTION
average over the stochastic eld
M .Di Ventra and R. DAgosta PRL (2007)
j(r, t)
t
=
F(r, t)
m
+
G(r, t) +
1
m
F
ext
(r, t)
R. DAgosta and M. Di Ventra PRB (2008)
EQUATION OF MOTION
average over the stochastic eld
F(r, t) =
i=j
(r r
i
)
j
U ( r
i
r
j
) + m (r, t)
internal forces
M .Di Ventra and R. DAgosta PRL (2007)
j(r, t)
t
=
F(r, t)
m
+
G(r, t) +
1
m
F
ext
(r, t)
R. DAgosta and M. Di Ventra PRB (2008)
EQUATION OF MOTION
average over the stochastic eld
G(r, t) =
V
j(r, t)
V
1
2
j(r, t)
V
V
1
2
V
V
j(r, t)
bath induced forces
M .Di Ventra and R. DAgosta PRL (2007)
j(r, t)
t
=
F(r, t)
m
+
G(r, t) +
1
m
F
ext
(r, t)
R. DAgosta and M. Di Ventra PRB (2008)
EQUATION OF MOTION
average over the stochastic eld
F
ext
(r, t) =
n(r, t)
m
t
A
ext
(r, t)
j(r, t)
m
[A
ext
(r, t)]
external forces: driving elds
M .Di Ventra and R. DAgosta PRL (2007)
j(r, t)
t
=
F(r, t)
m
+
G(r, t) +
1
m
F
ext
(r, t)
R. DAgosta and M. Di Ventra PRB (2008)
TIME-DEPENDENT
CURRENT-DFT
S. Ghosh and A. Dara, PRA (1988)
G. Vignale and W. Kohn, PRL (1996)
A
ext
(r, t)
In the spirit of Density functional theories
Ideal to study microscopical transport
non-interacting interacting
STOCHASTIC TD-CDFT
Given l(t),
V , (r, 0)
M. Di Ventra and R. DAgosta, PRL (2007)
A
ext
(r, t)
R. DAgosta and M. Di Ventra, PRB (2008)
STOCHASTIC TD-CDFT
Theorem: one-to-one correspondence between the
time dependent ensemble averaged current-density
and the external vector potential
Given l(t),
V , (r, 0)
M. Di Ventra and R. DAgosta, PRL (2007)
A
ext
(r, t)
R. DAgosta and M. Di Ventra, PRB (2008)
STOCHASTIC TD-CDFT
Theorem: one-to-one correspondence between the
time dependent ensemble averaged current-density
and the external vector potential
The bath is not changed in the KS system
Given l(t),
V , (r, 0)
M. Di Ventra and R. DAgosta, PRL (2007)
A
ext
(r, t)
R. DAgosta and M. Di Ventra, PRB (2008)
STOCHASTIC TD-CDFT
Theorem: one-to-one correspondence between the
time dependent ensemble averaged current-density
and the external vector potential
The bath is not changed in the KS system
Given l(t),
V , (r, 0)
M. Di Ventra and R. DAgosta, PRL (2007)
A
ext
(r, t)
A
KS
(r, t) = A
ext
(r, t) + A
xc
[j,
0
,
V ]
R. DAgosta and M. Di Ventra, PRB (2008)
A SURPRISING RESULT
The theorem of STCDFT does admit for the system to be
closed, i.e. we can set for the KS system
V = 0
A
KS
(r, t) = A
ext
(r, t) + A
xc
[
j,
0
, V = 0]
Yuen-Zhou et al., PCCP 2009
Relaxation and dissipation in TDCDFT are described by a KS
vector potential in a closed system!
CONCLUSION - 1I
S-TDCDFT is a theory to study the dynamics of an
interacting open quantum system
The theory can be extended in such a way to map
open quantum system into closed quantum system.
How to connect S-TDCDFT with standard TDCDFT?
THANKS!