Random
walks
Def:
Let
{
X
i
;
i
≥
1
}
be
a
sequence
of
IID
rv’s,
and
let
S
n
=
X
1
+
X
2
+
· · ·
+
X
n
for
n
≥
1
.
The
integertime
stochas
tic
process
{
S
n
;
n
≥
1
}
is
called
a
random
walk,
or,
specif
ically,
the
random
walk
based
on
{
X
i
;
i
≥
1
}
.
Our
focus
will
be
on
thresholdcrossing
problems.
For
example,
if
X
is
binary
with
p
X
(1)
=
1
,
p
X
(
−
1)
=
q
= 1
−
p
,
then
∞
k
p
Pr
{
S
n
≥
k
}
=
if
p
≤
1
/
2
.
1
p
n
=1
−
2
6.262:
Discrete
Stochastic
Processes
4/27/11
L21:
Hypothesis
testing
and
Random
Walks
Outline:
•
Random
walks
•
Detection,
decisions,
&
Hypothesis
testing
•
Threshold
tests
and
the
error
curve
•
Thresholds
for
random
walks
and
Cherno
ff
1
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Detection,
decisions,
&
Hypothesis
testing
The
model
here
contains
a
discrete,
usually
binary,
rv
H
called
the
hypothesis
rv.
The
sample
values
of
H
,
say
0
and
1,
are
called
the
alternative
hypotheses
and
have
marginal
probabilities,
called
a
priori
probabilities
p
0
=
Pr
{
H
= 0
}
and
p
1
= Pr
{
H
= 1
}
.
Among
arbitrarily
many
other
rv’s,
there
is
a
sequence
Y
(
m
)
= (
Y
1
, Y
2
, . . .
, Y
m
)
of
rv’s
called
the
observation.
We
usually
assume
that
Y
1
, Y
2
, . . .
,
are
IID
conditional
on
H
=
0
and
IID
conditional
on
H
= 1
.
Thus,
if
the
Y
n
are
continuous,
m
f
Y
(
m
)
(
y
) =
f
(
y
n

)
.
H

Y

H
n
=1

3
Assume
that,
on
the
basis
of
observing
a
sample
value
y
of
Y
,
we
must
make
a
decision
about
H
,
i.e.,
choose
H
= 0
or
H
= 1
,
i.e.,
detect
whether
or
not
H
is
1.
Decisions
in
probability
theory,
as
in
real
life,
are
not
necessarily
correct,
so
we
need
a
criterion
for
making
a
choice.
We
might
maximize
the
probability
of
choosing
correctly,
for
example,
or,
given
a
cost
for
the
wrong
choice,
might
minimize
the
expected
cost.
Note
that
the
probability
experiment
here
includes
not
only
the
experiment
of
gathering
data
(i.e.,
measuring
the
sample
value
y
of
Y
)
but
also
the
sample
value
of
the
hypothesis.
This is the end of the preview.
Sign up
to
access the rest of the document.
 Fall '11
 staff
 Probability, Probability theory, Stochastic process

Click to edit the document details