-

Quantum Algorithms -

quantum network :
|④④O④④④④.y
←
-

computational steps
synced in time
quantum register :
f
collection of Nctubits

-
quantum logic gat\ read out :
fixed unitary operation
-

projection
-

on selected qubits measurement

for a fixed time
/
-

9ngi¥sm¥goE
V
A- quantum circuit can

non -

unitary transformation classical outcome

¥¥¥¥
atheism's:# infatuate
'

.
values of a simultaneously
→ exponential speed
up with size
instead of linear like classical
Deutsch

[mmmmgy9uaonftamwaaygaabothd .
.at?orithm

asks a partial question about a

global property about
-

the system

"

-
o¥#YnE¥÷÷÷¥¥¥÷
singe:* .

Eg complot
:
.

don't know Kare about how foe) is calculated

.

•
.

time
limit # exec
•
we know : every evaluation of foe) is costly to abs min .
 f CO) fu ) ( unfair coin) f is constant
question Is or
-

: →
-

f- CO) Ffu ) ( fair coin ) → f is balanced

interested in the individual values of f-co) and fu )

we are not
.

whether f is constant or balanced

only the global property :

NM needs two evaluations of f

coµpuTATOl
KA
mass

÷÷÷÷:oi¥÷n::%aihai NO BETTER THAN GUESS ING

with less evaluations ( S2 ) → can only guess whether
f- is constant or balanced

QUANTUM
COMPUTATI O N
Mri with a single evaluation

A DON 't : obtain any information on individual values

:enstain *I④÷÷÷÷
④
arbitrary product state
eaq.io
,

1ny=qµy+p€
me
unitary leaves the first
←
→ut - qubit
tenet necessarily
unchanged

€f
a

few →
y is unchanged product state
i. e in
if foe) general
I is flipped 071
→
y
.
-
-

in , y ⑦ foe , >
I→ O
+ In > ly ⑤ folk
.

every application of Uf requires one evaluation of f-

o
-

Apply Uf twice :
Uf Uf la , y>
=
Uf IN , y ⑦ fol) > =
IN , y ① f- (a) ① fol ) )

Uft Uf
'
=
IN , y Y Uf -
- I =
 can we answer the question by only one application of Us ?
=

At!÷
-

'

position both possible inputs to get outcome

time

t¥
at the same

yin y >
Uff 107
)
=
,

it
,

10014101
=

# Uf ( )
contains both function
( Uf 100 > tuft 107 )
#
=

values in superposition

Only quantum parallel

,§ }
190 ④ HOD t l l ,
Offaly

fu ) )
10 f lol > t
l l ,
#
=

not ,

enough !
Problem : But we can 't access the information

If we measure the first qubit → superposition

no better than collapses

Tiffany
classical computations
,

iAt9fIiethYnm -

construct ly interfering the paths

such that information need

t.Iaanf.gl
we

-
? amplified
\
-

information we do not care about

-
inaccessible

Ii
' '

:÷:i¥a
 -

Attempt
/ 4-
"" "
¥41241104 HD
-

(Uf la 07
,
-

Ufln , ly )
lys I -7
-
-

ANY IN I ⑤ FCK) )
( IN , O ,
-

if fcn) -
- O
.

toy -

I ,y

HN ) I ( 107 II > IN ( in , FCN) >

-

IN HARD
¥2
- -
=
,
-
-

second
sign change depending -
fin , qubit
TD
-

foe) IN
on
leg remains

same
now take the It >
-_

-7=41
fintasoaonewdhoerthftrtis
Yt Uf It > I
Uf IN , =

fefingdiavinddual
iaa9f%b'
tht 'm
"" "

Fz f- is to>

f-
+

NOW if fo ) -
-
fu ) :
Uf It > I -7=47 ) IHI -7 fu ) as global
phase factor
FCO) Ffa ) :
Uf It > I y -
-

⇐ DI y -

l y
-

¥
*
What to do now ? Measure ist qubit in the K -

basis i
v
( = measure Enon 1H ) n
t
,
"

why ? because It> and I -7 are the eigenstates of K

Hts if f constant probability to obtain t.tl I

t> Liff balanced probability to obtain it -

I I

we have access to the information we need

DEFERIMDINHSSITDEAKK Hp