can
refer to state
<
in the spatial representation (which serves as our measuring base)
described by a Gaussian function
2
2
( )
exp(
/ 2
)
x
x
V
<
The pointer movement in that case is connected to the weak value of the operator A defined
by:
w
A
A
M \
M \
where
\
is the initial (preparation) state of the measured system, and
M
the final state
into which it is projected.
For a pre-/post-selected ensemble described by the two-state
M \
, the time evolution of
the total system (measured plus measuring system) is expected to be (
1
1
) [25]:
int
exp(
)
(1
)
exp(
)
W
d
W
d
i H dt
i A P
i A P
M
\
M \
H
M \
H
< |
<
<
³
which
results in
exp(
) ( )
(
)
w
w
i A p
x
x
A
H
H
<
<
For example, when weakly measuring the spin-z (described by the Pauli matrix
z
V
) of an
ensemble of spin-1/2 particles prepared in the X
+
direction, with coupling strength
/
N
H O
, the time evolution is determined by
int
1
exp(
)
exp(
/
)
N
z
n
d
n
W
i H dt
i
P
N
O V
¦
³
so for a single measurement, the evolution of the spin states becomes entangled with the
pointer of Eq. 18 (when
1
V
) :
2
2
(
/
) / 2
(
/
) / 2
2
2
1
[
1
1
]
1
2
[(1
)
1
1 ]
x
N
x
N
z
z
x
x
e
e
Nf
x
x
Nf
N
N
O
O
V
V
O
O
V
V
|
|
To complete the weak measurement, the pointer itself must be strongly measured. Then, the
particle's initial state X
+
changes by only a fraction
2
~
/ N
O
. Statistically, this means that only
2
~
O
out of N particles' states have changed. When
O
is small enough, this number of
(23)
(19)
(20)
(21)
(22)
(18)
EPJ Web of Conferences
00028-p.12
“flipped” spins (changed from the initial X
+
to X
-
) is negligibly small compared to the
ensemble's size [25].
Moreover, if the coupling strength is
/ N
H O
rather than
/
N
H O
, the weak measurement
process will most likely end without a single flip.
In case that all weak measurement are performed on a single particle (without constantly pre-
and post-selecting it as in [11]), using a single pointer but only on one particle, it was shown
[15] that the pointers undergoes a biased random walk with a log-normal distribution (see
Fig. 4). When increasing the number of weak measurements, this procedure tends to a strong
measurement (see Fig. 5).
Fig. 4. Distribution of the number of measurements until the collapse given a fixed standard
deviation σ.
ICNFP 2013
00028-p.13
Fig. 5. Success probability for hypothesis testing with low number of weak measurements. The solid
curve describes the optimal success probability for projective measurement.
3.2 Measuring Non-commuting variables of the same particle
The above technique entails an even more intriguing result: When the two (initial and final)
strong measurements are made on non-commuting operators, then, for the intermediate
states, these two operators can coexist with arbitrary precision.
3.3 Exotic mass and momentum
With the uncertainty principle thus subtly outsmarted and ordinary temporal order strained, it
is perhaps not surprising that these between-measurements states revealed by WM display
other physical oddities as well. Brief examination of Eq. 19 reveals that the weak values may
not belong to the operator's spectrum. As a consequence, particles with odd mass or
momentum, at times being even negative, are predicted by TSVF and amenable to isolation
and measurement by appropriate slicing. Such effects are demonstrated elsewhere
[25]
leading, for instance, to violation of Bohr's correspondence principle [13].
3.4 Weak Values of Entangled States
It was shown that weak measurements can be performed on an entangled pair (or triplet,
quartet etc.) without destroying the entanglement. Utilizing this possibility, we showed [12]
that one can record the results of
,
,
x
y
z
V V V
of two entangled spins which will later violate
EPJ Web of Conferences
00028-p.14