More Modern View of CT Thesis

Take it as an empirical claim about physics—say,

Any bounded physical system can be simulated
by a Turing machine, to any desired precision
 Asymptotics 2n
Key distinction:
polynomial vs. exponential

n2

n
 Asymptotics 2n
Key distinction:
polynomial vs. exponential

“What about n2
n1000? 21000n?
1.00000001n?”

n
Problem: Modeled as a set of binary strings, L{0,1}*.
Given an input x{0,1}*, the task is to decide if xL
The class of problems for which there’s an

P
algorithm, for a deterministic digital
computer, that always correctly decides if
xL, after a number of steps upper-bounded
by some polynomial in |x| (the length of x)

Examples: Primality Connectivity

Testing

Linear Programming
NP (Nondeterministic
Polynomial-Time)
Informally, the class of problems for which there’s a
polynomial-time algorithm to recognize valid solutions
(but the solutions might be exponentially hard to find)

Examples: Factoring

Traveling
Bin Packing Theorem Proving Salesperson
NP (Nondeterministic
Polynomial-Time)
Informally, the class of problems for which there’s a
polynomial-time algorithm to recognize valid solutions
(but the solutions might be exponentially hard to find)

Examples: Factoring

NP-complete
Traveling
Bin Packing Theorem Proving Salesperson
 P=NP? The (literally)
$1,000,000 question
If there actually were a machine with
[running time] ~Kn (or even only with ~Kn2),
this would have consequences of the
greatest magnitude.
—Gödel to von Neumann, 1956

For masochists:

(122 pages)
 Suppose we believe PNP. Is there any physical
means—going beyond Turing machines—to solve
NP-complete problems in polynomial time?

The Extended Church-

Turing Thesis (ECT)
Feasibly computable in the physical world =
In P (feasibly computable by Turing machine)

What sort of thing would overthrow the ECT, while

still upholding the “original” Church-Turing Thesis?
 Feynman 1981: Why not turn this
lemon into lemonade, by making
the computer itself quantum?
But supposing we built one, what would such a
“quantum computer” be good for?

Feynman’s answer: Simulating

quantum mechanics itself!

Shor 1994: Also factoring,

discrete log  breaking today’s
public-key cryptography (!!)
 Conjectured Picture

PSPACE
NP-complete
Bounded-Error
Quantum Post-quantum
Polynomial Time
NP crypto?

Factoring

BQP P
 n

y y
Z z

CCHOT
Toffoli

then NOT gate to 2

if x AND
y
are 1 apply
n
y roy
O O O
O 1 I T
I 0 1
a a
I 1 0
b b

c aab
 AND x
y
AND

ND my
AND
I
 AND x
y
AND

K ND
AND ay
y

AND
FNDlay
 Ia b f a
Bf 1a b
f 0 I 0 I
FAN OUT
i
 Bf 1a b I a lb of a

f 0,1 0,1

Phase kickback
last 1 la
Bf
 101 state at to

H H 101
117 107 117 at t
10

to t
lo 117 11 105 113 14 to

4 10 0701 110710
Bf 110
100 111 1101111
111
C1 10 1 3 C F 11 1 7 142
In

H 4 11 110 4 11 11 7
H a

11 10 17 41117
H Yes

1 11 t Ii 140

1 flo 1111
4142

f is const flo off o

lot Id la
H
balanced flo fli I
f is
 1417 100 1017 1107 1117 107 117

Determine 1427 via phase kick back

ti to

Be 4,7 11
100
007 1 101 11 110 if 1 1
 to 100 C1 101 1 401110 1117
c1 1

f 1007 1017 1107 1117 U

f fio 1007 101 1107

H
1117
t ta
f fi 1007 1017 1107 1117

100 101 110 111

 H 07 10 117 HID 10 117

Hla 107 1 119117

b
H 11 b
a
1
HOH la na L1 ly 11 1yd
f

H a 1 11 117
p
n E 0,13
 N 2

Classical
of queries for
zero error 2 I
 17
ND 4,1m
fit as 1
142
Eg
1437 H 1 142

k
4,41 t te te

41 4747 1 7
9 to
11 lys
 fatty
1
ly
tplitudeforafixed.ly
o all a
string bit
y 1
in front of 107
amplitude

f
11 ti t to
probability I
if f is constant

1 0 if f is balanced