ThesisPDF Available

Vergleich von Methoden zur Rekonstruktion von Quantenzuständen

Authors:

Figures

Content may be subject to copyright.
414 3·108
pi
|Ψii
ρ=X
i
pi|ΨiihΨi|
{Mm}
X
m
MmMm=1
m
ρ
m
p(m) = tr(MmMmρ)
m
ρ0=MmρMm
tr(MmMmρ)
Em:= MmMm
Em
{Mm}
p(m) = tr(Emρ)
ρ
ρ
x
x
0,1
Sd
Hd
Sd=ρ:Hd→ Hd|ρ , ρ =ρ
hρ, σi:= tr(ρ·σ)
RSd
kρk2=hρ, ρi=tr(ρ2)
Sdd2
d
Sd≡ S
N
H2H2..H22N
dimR(S) = (2N)2= 4N
4N4N1
N
M V
a, b M0λ1
λa + (1 λ)bM
S
N
N:= {ρ∈ S | tr(ρ)=1}⊂S
N
S
P
P:= {ρ∈ S | ρ0} ⊂ S
P
S
D
D:= {ρ∈ S | tr(ρ)=1, ρ 0}=N ∩ P
N P P =N ∩ P
S
ρtomo
ρ
A B
tr(A·B) = hA, Bi ≥ 0
A0A=A
A|ali=al|ali,hal|aki=δlk, al0
tr(A·B) = tr(B·A) = tr BX
l
al|alihal|!
=X
l
altr(B|alihal|) = X
l
alX
jhaj|B|alihal|aji
=X
lX
j
alhaj|B|aliδlj =X
l
alhal|B|ali
| {z }
0
0
ρ∈ D E0
tr(ρE) = hρ, Ei ≥ 0
S
ηE:SdRkρ
E
ηE(ρ) :=
tr(E1ρ)
..
tr(Eiρ)
..
tr(Ekρ)
=
hE1, ρi
..
hEi, ρi
..
hEk, ρi
ηEη
~
P= (p1, p2, .., pk)Rk
Θ := (~
P= (p1, p2, .., pk)Rk|
k
X
i=1
pi= 1)Rk
Θ
Ω := n~
P= (p1, p2, .., pk)Rk|pi0ioRk
W:= (~
P= (p1, p2, .., pk)Rk|
k
X
i=1
pi= 1, pi0)= Θ Rk
W
Θ Ω Rk
W= Θ
η(N)=Θ
η(P)
η(D)⊆ W
ρ0η(ρ)i=tr(Eiρ)0
~
P(ρ) = ηE(ρ)=(p1(ρ), .., pk(ρ))
ρ E
k
X
i=1
pi(ρ) = tr(ρ)
tr(ρ) = 1 1
k
X
i=1
pi=
k
X
i=1
tr(Eiρ) = tr k
X
i=1
Eiρ!=tr ρ
k
X
i=1
Ei!=tr(ρ1) = tr(ρ)
E={E1, E2, .., Ek}Ei0PiEi=1
ME=hE1, E2, .., EkiR⊆ S
ES
ME≡ M
S
S=M⊕M,M∩M={0}
MM
M:= {ρ∈ S | ∀ρM∈ M :hρ, ρMi= 0}
ηESd
ηE(λρ1+ρ2)i=tr(Ei(λρ1+ρ2)) = tr(λEiρ1+Eiρ2)
=λ tr(Eiρ1) + tr(Eiρ2)
=ληE(ρ1)i+ηE(ρ2)iρ1, ρ2∈ Sd,λR
ηEM
E0Rk
ρ∈ M
E
ηE(ρ)i=hEi, ρi= 0
Ei∈ MEρ∈ M
EηE(ρ)=0
M
E
tr(ρ)=0
~
P=ηE(ρ)=0Pipi= 0
tr(ρ)=0
M
ρ0∈ D ρ1=ρ0+ ∆ρ, ρ∈ M
tr(ρ1) = tr(ρ0) + tr(∆ρ) = tr(ρ0)=1
η(ρ1) = η(ρ0+ ∆ρ) = η(ρ0) + η(∆ρ) = η(ρ0)
ρ ρ10ρ0ρ1
E
E
ρ∈ M\{0}
tr(ρ)=0 ρ
ρ=X
i
λi|φiihφi|,X
i
λi= 0
λj6= 0
PM:S → M
hρ, σi=tr(ρ·σ)
PM(ρ) =
k0
X
i=1 ρ, E0
iE0
i
E0={E0
1, .., E0
k0} MEdimR(ME) = k0
k
ρ1, ρ2∈ S
η(ρ1) = η(ρ2)PM(ρ1) = PM(ρ2)
PM(ρ1) = PM(ρ2)
ρ1=ρM+ρ,1
ρ2=ρM+ρ,2
ρM∈ M ρ,1, ρ,2∈ M
η(ρ1) = η(ρ2) = η(ρM)
η(ρ1) = η(ρ2)
hρ1, Eii=hρ2, Eii ∀i
E0={E0
1, .., E0
k0} ME
PM(ρ1) = PM(ρ2)
k0
X
i=1 ρ1, E0
iE0
i=
k0
X
i=1 ρ2, E0
iE0
i
X
iρ1, E0
iρ2, E0
iE0
i= 0
ρ1, E0
iρ2, E0
i= 0 i
ρ1, E0
i=ρ2, E0
ii
E0MEρ1, ρ2
M M
ρ1=ρM,1+ρ,1
ρ2=ρM,2+ρ,2
η(ρ1) = η(ρ2)⇔ hρM,1, Eii=hρM,2, Eii ∀i
ρ1, E0
i=ρ2, E0
iiρM,1, E0
i=ρM,2, E0
iiρM,1=ρM,2
hρM,1, Eii=hρM,2, Eii ∀iρM,1=ρM,2
hρM,1, Eii=hρM,2, Eii ∀i
ρM,1=
k0
X
l=1
c1lE0
l, c1lR
ρM,2=
k0
X
l=1
c2lE0
l, c2lR
ρM,16=ρM,2n1ρM,1ρM,2
n
dl:= c1lc2l6= 0
1ln c1l=c2lcln+ 1 lk0
ρM,1=
n
X
l=1
c1lE0
l+
k0
X
l=n+1
clE0
l
ρM,2=
n
X
l=1
c2lE0
l+
k0
X
l=n+1
clE0
l
ρM,1ρM,2=
n
X
l=1
(c1lc2l)E0
l
n
X
l=1
dlE0
l, d 6= 0
hρM,1, Eii=hρM,2, Eii ∀i⇔ hρM,1ρM,2, Eii= 0 i
*n
X
l=1
dlE0
l, Ei+= 0 i06=
n
X
l=1
dlE0
l∈ M
ρM,16=ρM,2
~
f
~
P=η(ρ)~
f
ρ
η
ρSST := η1(~
f)
η
3.1
ρSST =η1
M(~
f) + ρ=: ρ
SST +ρ
ρMρ
SST
S
ME=SηES
ρSST =η1(~
f) = ρ
SST
η
Sd2
~
f=η(ρSST )
X
i
fi=tr(ρSST )=1
ρSST
ρtomo
S~
fRk~
f η
ρ1, ρ2∈ S ~
f
η(ρ1) = η(ρ2) = ~
f
ρ1ρ2M
PM(ρ1) = PM(ρ2) = ρM
ρ1=ρM+ρ1, ρ2=ρM+ρ2
λρ1+ (1 λ)ρ2=ρM+λρ1+ (1 λ)ρ2
PM(λρ1+ (1 λ)ρ2) = ρM
η(λρ1+ (1 λ)ρ2) = ~
f
λρ1+ (1 λ)ρ2η1~
f
~
P
D(~
P) := nρ∈ D | η(ρ) = ~
P)o=η1~
P∩ D
nρ=ρM+ρ|ρM=η1
M(~
P), ρ∈ M:ρ0o
D(~
P)Dη1~
P
ηMρ∈ D(~
P)
PM(ρ) = η1
M(~
P)
ρM∈ M tr(ρM)=1
D(ρM) := nρ=ρM+ρ|ρ∈ M, ρ ∈ Do
~
f∈ W
D(~
f) =
WD
ρ∈ D
WD:= n~
P∈ W | ∃ρ∈ D :η(ρ) = ~
Po⊆ W
WDW
~
P1,~
P2∈ WD
ρ1∈ D :~
P1=η(ρ1)
ρ2∈ D :~
P2=η(ρ2)
~
Pλ:= λ~
P1+ (1 λ)~
P2∈ WD0λ1
~
Pλ=λ~
P1+ (1 λ)~
P2∈ W
Wρλ∈ D
η(ρλ) = ~
Pλ=λ~
P1+ (1 λ)~
P2
ρλ:= λρ1+ (1 λ)ρ2
Dρλ∈ D
η(ρλ) = η(λρ1+ (1 λ)ρ2) = λη(ρ1) + (1 λ)η(ρ2) = λ~
P1+ (1 λ)~
P2=~
Pλ
S
MD⊆ M
MD:= nρM∈ M | tr(ρM)=1,ρ∈ M:ρM+ρ0o
MD
MDPM(D)
S
W Mw:= η1(W)
MwMN
MD
ρ1
M, ρ2
M∈ Mw~
P1=η(ρ1
M)∈ W ~
P2=
η(ρ2
M)∈ W 0λ1
η(λρ1
M+ (1 λ)ρ2
M) = λ~
P1+ (1 λ)~
P2∈ W
MD⊆ MwWD⊆ W
WDMD
MDWD
η:MD→ WD
ρM7→ η(ρM)
η η MDWD
η(MD)⊆ WDρ∈ MD
ρ∈ Mρ+ρ∈ D WD3η(ρ+ρ) = η(ρ)
η~
P∈ WD
ρ∈ D η(ρ) = ~
P ρM=PM(ρ)ρM∈ MD
η(ρM) = ~
P
η η M MD⊆ M
ηMD
η
η
ρM∈ M MD
ρ∈ MρM+ρ0
ρM+ρ
ρM∈ MDkρMk2=tr(ρ2
M)1
ρM∈ MDρ∈ Mρ:= ρM+ρ∈ D
1≥ kρk2=tr(ρ2) = tr((ρM+ρ)2)
=tr(ρ2
M+ρMρ+ρρM+ρ2
)
=tr(ρ2
M)+2tr(ρMρ)
| {z }
= 0
+tr(ρ2
)
| {z }
0
=tr(ρ2
M) + tr(ρ2
)
11tr(ρ2
)tr(ρ2
M)
ρSST
ρ0
ρ
ρAdhoc =ρ0
tr(ρ0)
ρSST
ρSST
~
P=η(ρ)~
f
m p(m) = tr(Emρ)
N
ρ N
~
f
E ρ
~
f
L(~
f , ρ, E) = N!
(Nf1)!(N f2)!..(N fk)!
k
Y
j=1
pfjN
j
=N!
(Nf1)!(N f2)!..(N fk)!
k
Y
j=1
tr(Ejρ)fjN
ρML ∈ D
L(~
f , ρML, E ) = max
ρ∈D L(~
f , ρ, E)
~
f E
L(~
f , ρ, E)
ρ
N L N
L
L(~
f , ρ, E) :=
k
Y
j=1
tr(Ejρ)fj
max
ρ∈D L(~
f , ρ, E)b= max
ρ∈D L(~
f , ρ, E) = max
ρ∈D
k
Y
j=1
tr(Ejρ)fj
L
ρ ρ ∈ D
~x = (x1, .., xk), ~y = (y1, .., yk)Rk
xi, yi0iPk
i=1 yi= 1
k
Y
i=1
xyi
i
k
X
i=1
yixi
k
Y
i=1
xyi
i=
k
X
i=1
yixixix0i
k
Y
i=1
xyi
i
k
X
i=1
yixi
ln Y
i
xyi
i!ln X
i
yixi!
X
i
yiln(xi)ln X
i
yixi!
xix0i
L~
f(~
P) :=
k
Y
i=1
pfi
i
W
WD⊆ W
L
g(~
P) := ln k
Y
i=1
pfi
i!=
k
X
i=1
ln pfi
i=
k
X
i=1
piln(fi)
L
ln(L)~x 6=~y ∈ W
0<λ<1
g(λ~x + (1 λ)~y)> λg(~x) + (1 λ)g(~y)
X
i
piln(λxi+ (1 λ)yi)> λ X
i
piln(xi) + (1 λ)X
i
piln(yi))
X
i
piln(λxi+ (1 λ)yi)>X
i
pi(λln(xi) + (1 λ) ln(yi))
xi6=yi
ln(λxi+ (1 λ)yi)> λ ln(xi) + (1 λ) ln(yi)
~
P∈ W pi0pl>0
X
i
piln(λxi+ (1 λ)yi)>X
i
pi(λln(xi) + (1 λ) ln(yi))
~x 6=~y ∈ W 0<λ<1
ln(L~
f(~
P)) = ln k
Y
i=1
pfi
i!=
k
X
i=1
piln(fi)
D(~q, ~p) := X
i
qiln qi
pi=X
i
qiln qiX
i
qiln pi
ln(L~
f(~
P)) = D(~
f, ~
P) + X
i
filn fi
~
P~
fWD
max
~
P∈WDL~
f(~
P)b= max
~
P∈WD
ln(L~
f(~
P)) = max
~
P∈WDD(~
f, ~
P) + X
i
filn fib= min
~
P∈WD
D(~
f, ~
P)
W
D(~q, ~p)0~q, ~p ∈ W
D(~q, ~p) = 0 ~p =~q
L~
f(~
P) :=
k
Y
i=1
pfi
i
WDρ
η(ρ) = max
~
P∈WDL~
f(~
P) := ~
PML
L~
fWD
ρ1, ρ2∈ D L(~
f , ρ)D
PM(ρ1) = PM(ρ2)
η(ρ1) = η(ρ2) = ~
PML
PM(ρ1) = PM(ρ2)
~
f∈ W
E= (E1, .., Ek)
L(~
f , ρ, E) =
k
Y
i=1
tr(Eiρ)fi
D
Γ = nρ=ρM+ρ|ρ∈ M:ρ0o=D(ρM)⊆ D
ρM∈ MDL(~
f , ρM, E) = maxρ∈MDL(~
f , ρ, E)
Γ
L(~
f , ρ)D~
PML L~
f
WDWDMD
MD
ρM∈ MDη(ρM) = ~
PML Γ = D(~
PML)
ρ0
Γ = nρ=PM(ρ0) + ρ|ρ∈ M:ρ0o
PM(ρ0)
dimR(M) = d2
M={0}
Γ = nρ=ρM+ρ|ρ∈ M={0}:ρ0o={ρM}
ρ
SST =η1
M(~
f)) MD
ρ
SST MD
ρ
SST
Mw
max
ρ∈MwL(~
f , ρ) = L(~
f , ρ
SST ) = L(~
f , η1
M(~
f))
max
ρ∈MwL(~
f , ρ)b= max
ρ∈MwX
i
tr(Eiρ) ln(fi)b= min
ρ∈Mw
D(~
f , η(ρ))
D(~
f , η(ρ)) = 0 ~
f=η(ρ)ρ=η1
M(~
f) = ρ
SST
ρ
SST ∈ MD
ρ
SST =η1
M(~
f)∈ MD⇒ L(~
f , ρ
SST ) = max
ρ∈MDL(~
f , ρ)
MD⊆ Mw
ρ
SST =η1
M(~
f)MD
ρML MDMD
MDWDL
MDρML MD
MwρML L Mw
ρML
ρ
SST /∈ MD
MD
R(ρ)ρ=ρ
R(ρ) := Pk
i=1
fi
tr(Eiρ)Ei∈ S
R(ρ)ρ
R(ρ)ρ=ρ(R(ρ)ρ) = ρ
ρR(ρ) = ρR(ρ)ρR(ρ) = R(ρ)ρ=ρ
R(ρ)ρR(ρ) = ρ
ρi+1 = Φ(ρi) = (1 α)ρi+αR(ρi)ρiR(ρi)
tr(R(ρi)ρiR(ρi))
0< α 1
ρi∈ D ⇒ Φ(ρi)∈ D
R(ρi)ρi=ρiα
S
MD
MDD
|ΨihΨ|ρ
p0
ρtomo
D
x2+y2+z21
ρ=1
21 + z x iy
x+iy 1z
σxσyN= 2k k N
k k
ρ
L=(2k)!
k!k!2
pk
x(1 px)kpk
y(1 py)k
px:= tr(ρ|1xih1x|)py:= tr(ρ|1yi h1y|)
1px=
py= 1/2ρML =1/2
Lmax =(2k)!
k!k!21
24k
(x, y) = (0,0)
¯
L
¯
L:= L
Lmax
= 24kpk
x(1 px)kpk
y(1 py)k
pxpyx, y, z
¯
L=L
Lmax
= 4k(x+ 1)k1x+ 1
2k
(y+ 1)k1y+ 1
2k
=4(x+ 1) 1x+ 1
2(y+ 1) 1y+ 1
2N/2
xy
N
¯
L
N
x y b1/2
x=rcos(φ), y =rsin(φ),0r1
¯
L(r, φ) = (1 r2r4cos2(φ) sin2(φ))N/2(1 r2)N/2
¯
L(1 r2)N/21/2
(1 r2)N/21
2rp122/N =: b1/2(N)
2
N
limN→∞ b1/2(N) = 211=0
N
ρML
ρBM E := ZD
Π( ~
f , ρ)ρ dρ
Π( ~
f , ρ)∝ L(ρ)Π Π( ~
f , ρ)0
RDΠ( ~
f , ρ)= 1
D
ρBM E ∈ D
ρBM E
Γ
Γ
Γ
M
I(λ, ρ) := λS(ρ) + 1
Nlog L(ρ)
λ > 0S(ρ)
λ
S(ρ) Γ
ρMLME
M
LH(~
f , ρ) := det(ρ)βL(~
f , ρ)
β0
β0
ρ
LH
det(ρ)βρHM L
β > 0ρM L
ρ1
LHρ
β
N r
mdr log2d
d= 2N
w=
N
O
i=1
wi
wi∈ {1, σx, σy, σz}
d2
dr d2dr log2d
ρ
ρSST
ρSST
ρAdhoc
k
kd2
ρML
ρ
ML ∈ MD
L(~
f , ρ
ML) = max
ρ∈MDL(~
f , ρ)
ρ
ρi+1 = (1 α)ρi+αR(ρi)ρiR(ρi)
tr(R(ρi)ρiR(ρi)) 0< α 1
ρi+1 = (1 α)ρi+α PMR(ρi)ρiR(ρi)
tr(R(ρi)ρiR(ρi)) 0< α 1
ρ0=1
d
ρi∈ MD
i
i= 0 ρ0∈ M
ρi∈ MDρi+1
MD
ρ
i∈ Mσi:= ρi+ρ
i0
α= 1 MD
ρi+1 =PMR(ρi)ρiR(ρi)
tr(R(ρi)ρiR(ρi))
4R(σi)σiR(σi)σi
R(σi)σiR(σi) = R(ρi+ρ
i)(ρi+ρ
i)R(ρi+ρ
i)0
R
R(ρi+ρ
i) =
k
X
j=1
fj
tr((ρi+ρ
i)Ej)Ej=
k
X
j=1
fj
tr(ρiEj) + tr(ρ
iEj)Ej
=
k
X
j=1
fj
hρi, Eji+Dρ
i, EjE
| {z }
0
Ej=
k
X
j=1
fj
tr(ρiEj)Ej=R(ρi)
˜ρ:= R(ρi)ρiR(ρi) ˆρ:= R(ρi)ρ
iR(ρi)
R(σi)σiR(σi) = R(ρi)(ρi+ρ
i)R(ρi)
=R(ρi)ρiR(ρi) + R(ρi)ρ
iR(ρi)
= ˜ρ+ ˆρ˜ρM+ ˜ρ+ ˆρM+ ˆρ0
˜ρM:= PM(˜ρ) ˆρM:= PM(ˆρ)
PM(R(σi)σiR(σi)) = PM(˜ρM+ ˜ρ+ ˆρM+ ˆρ) = ˜ρM+ ˆρM∈ M
PMR(σi)σiR(σi)
tr(R(σi)σiR(σi)) =PM(R(σi)σiR(σi))
tr(R(σi)σiR(σi)) =˜ρM+ ˆρM
tr(˜ρM+ ˜ρ+ ˆρM+ ˆρ)
=˜ρM+ ˆρM
tr(˜ρM+ ˆρM)∈ MD
˜ρ+ ˆρ
tr(˜ρM+ ˆρM)∈ M
˜ρM+ ˆρM
tr(˜ρM+ ˆρM)+˜ρ+ ˆρ
tr(˜ρM+ ˆρM)=˜ρM+ ˜ρ+ ˆρM+ ˆρ
tr(˜ρM+ ˆρM)0
tr ˜ρM+ ˜ρ+ ˆρM+ ˆρ
tr(˜ρM+ ˆρM)=tr( ˜ρM+ ˜ρ+ ˆρM+ ˆρ)
tr(˜ρM+ ˆρM)=tr( ˜ρM+ ˆρM)
tr(˜ρM+ ˆρM)= 1
ρi+1 MD
PMR(ρi)ρiR(ρi)
tr(R(ρi)ρiR(ρi)) =˜ρM
tr(˜ρM)∈ MD
˜ρM+ ˆρM
tr(˜ρM+ ˆρM)∈ MD?
˜ρM
tr(˜ρM)∈ MD
16
ρ
E
ηE(ρ)
16 3 ·108
~v =~a +i~
b ~a ~
b
S
ρ= (1 m)|Ψi hΨ|+m1
d
0m1
2 16
σiσji, j = 0,1,2,3x, y, z
σ0σ0=1
1
k k 1 (i, j)6= (0,0)
σiσj
k
k1
ηE(ρ)
[0,1] k
N[0,1]
α= 1
d(ρi+1, ρi) = ptr((ρi+1 ρi)2)
S
d(ρi+1, ρi)2=tr((ρi+1 ρi)2)!
2=: 0
0= 1012
0
Nmax = 5 ·104
1%
R(ρi)ρiR(ρi)
M
Mk
ρ
ML ρML
< d(ρtomo, ρ
ML)2>
N
k
2000
k= 4,8,12
~
f /∈ WDm
k= 4 N= 16 0,6%
k= 8 N= 64 0,7%
k= 12 N= 144 0,3%
MD
ρM∈ MD
P
ρ=AA
A
log L(ρ=AA)Str(AA)=1
k
X
i=1
filog(tr(EiAA)) + λ(tr(AA)1)!!
= 0
MDρ=AA
ρ=PM(AA)
M M
PM(AA)
| {z }
∈M
+ (AA PM(AA))
| {z }
∈M
=AA 0
MD
log L(ρ=PM(AA))
Str(PM(AA))) = 1
tr(PM(AA)) = tr(AA)=1
AA M
k
X
i=1
filog(tr(EiPM(AA))) + λ(tr(AA)1)!!
= 0
SB={B1, .., Bd2}
SBB0={B1, .., Bk} M
A ciR
A=
d2
X
i=1
ciBiA2=
d2
X
i,l=1
ciclBiBl∈ S
A2SB
AA =
d2
X
m=1 hBm, AAiBm=
d2
X
m=1 *Bm,
d2
X
i,l=1
ciclBiBl+Bm=:
d2
X
m=1
qmBm
AM
PM(AA) =
k
X
l=1 *Bl,
d2
X
m=1
qmBm+Bl=
k
X
l=1
d2
X
m=1
qmhBl, BmiBl
hBl, Bmi=(δlm 1l, m k
0 1 lk k + 1 md2
PM(AA)
PM(AA) =
k
X
m=1
qmBm
qm:=
d2
X
i,l=1
ciclhBm, BiBli
~q = (q1, .., qk)
~c F:= ~c k
X
i=1
filog k
X
m=1
qmtr(EiBm)!+λ k
X
m=1
qmtr(Bm)1!! !
= 0
~c = (c1, .., cd2)A
F
∂F
∂cα
=
k
X
j=1
∂F
∂qj
∂qj
∂cα
∂F
∂qj
=
∂qj
k
X
i=1
filog k
X
m=1
qmtr(EiBm)!+λ
∂qj k
X
m=1
qmtr(Bm)1!
=
k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)+λtr(Bj)
~c
∂qj
cα
=
cα
d2
X
i,l=1
ciclhBj, BiBli
hBj, BiBli hjiliqj
qj=
d2
X
i=1
ci
d2
X
l=1
clhjili=X
i6=α
ciX
l
clhjili+cαX
lhjαli
=X
i6=α
ci
X
l6=α
clhjili+cαhjiαi
+cα
X
l6=α
clhjαli+cαhjααi
∂qj
∂cα
=
∂cα
X
i6=α
ci
X
l6=α
clhjili+cαhjiαi
+cα
X
l6=α
clhjαli+cαhjααi
=X
i6=α
cihjiαi+X
l6=α
clhjαli+ 2cαhjααi
=X
i6=α
cihjiαi+cαhjααi+X
l6=α
clhjαli+cαhjααi
=X
i
cihjiαi+X
l
clhjαli=X
l
cl(hjlαi+hjαli)
=
d2
X
l=1
clhBj, BlBα+BαBli
∂F
∂cα
=
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)+λtr(Bj)!d2
X
l=1
clhBj, BlBα+BαBli!
= 0
~c
∂qj
∂cα~q
qj=X
i,l
ciclhjili=X
i,l
ciclhjili+X
i,l
ciclhjlii − X
i,l
ciclhjlii
| {z }
=qj
2qj=X
i,l
cicl(hjili+hjlii) = X
l
cl X
i
ci(hjili+hjlii)!=
d2
X
l=1
cl
∂qj
∂cl
6.4~c
d2
X
α=1
cα
∂F
∂cα
=
d2
X
α=1
cα
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)+λtr(Bj)!qj
∂cα
=
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)+λtr(Bj)!d2
X
α=1
cα
∂qj
∂cα
=
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)+λtr(Bj)!2qj
= 2 X
i
fiPjqjtr(EiBj)
Pmqmtr(EiBm
+ 2λX
j
qjtr(Bj)
= 2 X
i
fi+ 2λqjtr(Bj) = 2 + 2λX
j
qjtr(Bj)
d2
X
α=1
cα
∂F
∂cα
= 2 + 2λX
j
qjtr(Bj)
ρ=PM(AA)
PmqmBm∂F
∂cα=
0α
0 =
d2
X
α=1
cα
∂F
∂cα
= 2 + 2λX
j
qjtr(Bj)
λ
tr(ρ) = Pjqjtr(Bj)=1
0 = 2 + 2λX
j
qjtr(Bj)⇔ −1 = λX
j
qjtr(Bj)
⇔ −1
λ=X
j
qjtr(Bj)!
= 1 λ=1
∂F
∂cα
=
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)tr(Bj)!qj
∂cα
!
= 0
~c
F
F=~cF·~c =
d2
X
α=1
∂F
∂cα
cα=
d2
X
α=1
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)tr(Bj)!qj
∂cα
cα
=
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)tr(Bj)!d2
X
α=1
∂qj
∂cα
cα
k
X
j=1 k
X
i=1
fi
tr(EiBj)
Pk
m=1 qmtr(EiBm)tr(Bj)!qj(∆~c)
=
k
X
j=1
∂F
∂qj
qj(∆~c) = ~qF·~q(∆~c)
~cF= 0
F=~cF·~c !
= 0 ~c
⇔ ∇~qF·~q(∆~c)!
= 0 ~c
~q ~c
~cF·~c !
= 0 ~c ⇔ ∇~cF= 0
~qF·~q(∆~c)!
= 0 ~c ⇔ ∇~q F= 0
MD(q1, .., qk)
F= 0
S
MD
ρtomo
L(ρtomo)maxρ∈D L(ρ)
N→ ∞
L/Lmax
lim
N→∞
L
Lmax
(ρ) = (1ρ=ρ0
0
W(~q)
L L0=W(~q)βL
α= 1
α
k= 8 α= 0.5
α= 1
k= 8 N= 64 α= 0.5
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
k= 8
%
%
%
%
%
%
ρtomo
ρ∈ D
tr 1
d2!=1
dtr(ρ2) = kρk21
dρ, 1
d2
=
ρ1
d
2
=tr ρ1
d2!=tr ρ22
d1ρ+12
d2
=tr(ρ2)2
dtr(ρ) + d
d2=tr(ρ2)2
d+1
d=tr(ρ2)1
d
ρ∈ D
0dρ, 1
d2
11
d=: R(d)2
D
1
dR(d)
N
KrrN
ρ ρ 0
Kr: = ρ∈ N | d(ρ, 1
d)r=ρ∈ N | tr(ρ2)1
dr2
=ρ∈ N | tr(ρ2)r2+1
d=: ρ∈ N | tr(ρ2)˜r2
tr(ρ) = 1 tr(ρ2)˜r2ρ0
λ1, .., λdρ tr(ρ) = Piλi= 1 tr(ρ2) = Piλ2
i
˜r2λi0i
λj<0
λj~
λ0= (λ1, .., λj1, λj+1, .., λd)
X
i
λi=λj+X
i6=j
λi= 1
Pi6=jλi>1
k~
λ0k2
1=X
i6=j|λi| ≥ X
i6=j
λi>1
k~vkp:= (Pi|vi|p)
1
p
tr(ρ2) = X
i
λ2
i=λ2
j+X
i6=j
λ2
i
=λ2
j
|{z}
>0
+k~
λ0k2
2˜r2
k~
λ0k2˜r
α
k~
λ0k1αk~
λ0k2
d1
1<k~
λ0k1d1k~
λ0k2
1
d1<k~
λ0k1
d1≤ k~
λ0k2
˜r:= 1
d1r2= ˜r21
d=
1
d11
d=1
d(d1)
Kr=(ρ∈ N | d(ρ, 1
d)r=1
pd(d1))
d1/2d
N
N
ρtomo
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Accurately inferring the state of a quantum device from the results of measurements is a crucial task in building quantum information processing hardware. The predominant state estimation procedure, maximum likelihood estimation (MLE), generally reports an estimate with zero eigenvalues. These cannot be justified. Furthermore, the MLE estimate is incompatible with error bars, so conclusions drawn from it are suspect. I propose an alternative procedure, Bayesian mean estimation (BME). BME never yields zero eigenvalues, its eigenvalues provide a bound on their own uncertainties, and it is the most accurate procedure possible. I show how to implement BME numerically, and how to obtain natural error bars that are compatible with the estimate. Finally, I briefly discuss the differences between Bayesian and frequentist estimation techniques. Comment: RevTeX; 14 pages, 2 embedded figures. Comments enthusiastically welcomed!
Article
We examine a variety of strategies for numerical quantum-state estimation from data of the sort commonly measured in experiments involving quantum-state tomography. We find that, in some important circumstances, an elaborate and time-consuming numerical optimization to obtain the optimum density matrix corresponding to a given data set is not necessary and that cruder, faster numerical techniques may well be sufficient; in other words, “the best” is the enemy of “good enough.”
Article
If a photon of definite polarization encounters an excited atom, there is typically some nonvanishing probability that the atom will emit a second photon by stimulated emission. Such a photon is guaranteed to have the same polarization as the original photon. But is it possible by this or any other process to amplify a quantum state, that is, to produce several copies of a quantum system (the polarized photon in the present case) each having the same state as the original? If it were, the amplifying process could be used to ascertain the exact state of a quantum system: in the case of a photon, one could determine its polarization by first producing a beam of identically polarized copies and then measuring the Stokes parameters1. We show here that the linearity of quantum mechanics forbids such replication and that this conclusion holds for all quantum systems.
Article
Treatment of the predictive aspect of statistical mechanics as a form of statistical inference is extended to the density-matrix formalism and applied to a discussion of the relation between irreversibility and information loss. A principle of "statistical complementarity" is pointed out, according to which the empirically verifiable probabilities of statistical mechanics necessarily correspond to incomplete predictions. A preliminary discussion is given of the second law of thermodynamics and of a certain class of irreversible processes, in an approximation equivalent to that of the semiclassical theory of radiation. It is shown that a density matrix does not in general contain all the information about a system that is relevant for predicting its behavior. In the case of a system perturbed by random fluctuating fields, the density matrix cannot satisfy any differential equation because rho&dot;(t) does not depend only on rho(t), but also on past conditions The rigorous theory involves stochastic equations in the type rho(t)=G(t, 0)rho(0), where the operator G is a functional of conditions during the entire interval (0-->t). Therefore a general theory of irreversible processes cannot be based on differential rate equations corresponding to time-proportional transition probabilities. However, such equations often represent useful approximations.
Article
It is shown that the probability distribution for the rotated quadrature phase [{a}^{\mathrm{\ifmmode^\circ\else\textdegree\fi{}}}exp(i\ensuremath{\theta})+a exp(-i\ensuremath{\theta})]/2 can be expressed in terms of quasiprobability distributions such as P, Q, and Wigner functions and that also the reverse is true, i.e., if the probability distribution for the rotated quadrature phase is known for every \ensuremath{\theta} in the interval 0\ensuremath{\le}\ensuremath{\theta}<\ensuremath{\pi}, then the quasiprobability distributions can be obtained.
  • T Monz
  • P Schindler
  • J T Barreiro
  • M Chwalla
  • D Nigg
  • W A Coish
  • M Harlander
  • W Hänsel
  • M Hennrich
  • R Blatt
T. Monz, P. Schindler, J. T. Barreiro, M. Chwalla, D. Nigg, W. A. Coish, M. Harlander, W. Hänsel, M. Hennrich und R. Blatt: 14-Qubit Entanglement: Creation and Coherence, PRL 106, 130506, 2011.
  • H Häner
  • W Hänsel
  • C F Roos
  • J Benhelm
  • D Chek-Al-Kar
  • M Chwalla
  • T Körber
  • U D Rapol
  • M Riebe
  • P O Schmidt
  • C Becher
  • O Gühne
  • W Dür
  • R Blatt
H. Häner, W. Hänsel, C. F. Roos, J. Benhelm, D. Chek-al-kar, M. Chwalla, T. Körber, U. D. Rapol, M. Riebe, P. O. Schmidt, C. Becher, O. Gühne, W. Dür und R. Blatt: Scalable multiparticle entanglement of trapped ions, Nature 438, 643, 2005.
  • D Gross
  • Y.-K Liu
  • S T Flammia
  • S Becker Und
  • J Eisert
D. Gross, Y.-K. Liu, S. T. Flammia, S. Becker und J. Eisert: Quantum state tomography via compressed sensing, arXiv: 0909.3304v4, 2010.
  • Y S Teo
  • H Zhu
  • B.-G Englert
  • J Rehacek
  • Z Hradil
Y. S. Teo, H. Zhu, B.-G. Englert, J. Rehacek und Z. Hradil: Knowledge and ignorance in incomplete quantum state tomography, arXiv: 1102.2662v2, 09.07.2011.
Maximum-likelihood methods in quantum mechanics
  • Z Hradil
  • J Rehacek
  • J Fiurasek
  • M Jezek
Z. Hradil, J. Rehacek, J. Fiurasek und M. Jezek: Maximum-likelihood methods in quantum mechanics, Lect. Notes Phys., 649:59, 2004.