EARLY

QUANTUM
COMPUTING
SARAH LI
What’s a Computer?

A computer is an
11011110 information processor
00110001
01101010
Input

Output
Black box 10000111
Obeys certain logical rules 00001010
00110101
What’s a Computer?

Credit: Mike Davey http://aturingmachine.com/

What’s a Computer?

Despite their different appearances,

all of these computers
follow the exact same rules.
What’s a Quantum Computer?
We’ve already seen that we can encode and process
information in a quantum system
11011110 A quantum computer is a
00110001 quantum information processor
01101010
Input

Output
Black box obeys different rules, 10000111
such as unitary evolution, 00001010
but can take advantage of 00110101
superposition and entanglement
 Classical vs. Quantum Computing

• Uses bits as input and output • Uses bits as input and output
• Gives one answer per run • Gives one answer per run
• Can fake randomness • Is fundamentally uncertain
• Can observe the computation • Observing the computation partway disturbs
partway through ruins the process
• Can only measure the bits in one way • Can measure qubits in infinite ways
 A Timeline of Quantum Computing
1996
DiVincenzo
1992 criteria
Deutsch-Josza
algorithm 1994
1981
Q. Computing Shor’s
enters the lexicon factoring
algorithm

1995 2001
Trapped-ion Linear optics
1984 CNOT for QC outlined
2007
Q. Key Distribution
1981 2001 Superconducting
is proposed 1997
Nonlocality of Cluster-state circuit CNOT
NMR-based
entanglement QC outlined QC proposed 2015
verified 2007 Loophole-free
1997 Q-secure entanglement
Topological QC elections verification
proposed in Geneva
Early Quantum Computing

• Oracle problems
• The Deutsch-Josza Problem
• The DiVincenzo Criteria
• Overview of Quantum Computing Implementations

Detailed notes posted in Slack

The Deutsch-Josza
Problem
The made-up problem that started it all
Oracle Problems
𝑥𝑥? 𝒇𝒇(𝒙𝒙)

Phone Books Abstract Oracle

Give them a name, Give an input,
they give you a phone number get an output

But what if we want other kinds of information? Collective property

e.g. How many phone numbers have the 519 area code? of the possible outputs,
Is there an efficient way to get that out of the oracle? not one specific output
 The Deutsch-Josza Problem
You are given a binary function 𝑓𝑓(𝑥𝑥)
There are two possible inputs (0 or 1)
There are two possible outputs (0 or 1)
Your mission: Determine if 𝑓𝑓(𝑥𝑥) is constant or balanced

There are four possible functions:

x f1(x) x f2(x) x f3(x) x f4(x)

0 0 0 1 0 0 0 1
1 0 1 0 1 1 1 1
Constant Balanced Balanced Constant
The Deutsch-Josza Problem

𝑥𝑥?
𝒇𝒇(𝒙𝒙)
 The Deutsch-Josza Problem
What is the minimum number of queries you’d need to ask the
oracle to learn if the function is constant or balanced?

x f1(x) x f2(x) x f3(x) x f4(x)

A. Zero
0 0 0 1 0 0 0 1
B. One 1 0 1 0 1 1 1 1
C. Two Constant Balanced Balanced Constant
D. Three 0?
𝒇𝒇 𝟎𝟎
= 𝟏𝟏
E. Four Still no information
on whether it is
constant or balanced
After two queries, I can tell you
which function it is,
which is more information
than we need!
Question Break
The Quantum
Deutsch-Josza
Solution
Play that one again DJ
 The Quantum Oracle
Let’s say we’re able to ask for the oracle as a quantum gate:
Two-qubit
Control Qubit
binary input

Target Qubit

We have the two-qubit gate 𝑈𝑈𝑓𝑓

which has the function 𝑓𝑓 𝑥𝑥
programmed into it as:
Addition Mod 2
(Bitwise / XOR)
0 1
0 0 1 If 𝑓𝑓 𝑥𝑥 = 1, flip 𝑦𝑦
1 1 0 Otherwise, do nothing
The Quantum Oracle

x f2(x)
0 1
1 0

The control doesn’t change

The target either flips or doesn’t flip
The Quantum Oracle

What if we send in the What if the control is in

computational basis states? a superposition state?

Both 𝑓𝑓(0) and 𝑓𝑓(1) in the

output state!

The two-query method But when we measure,

still works just as well we’ll get one or the other
randomly
The Deutsch-Josza Solution

What if the target is in the |−⟩ superposition state?

Does nothing if 𝑓𝑓 𝑥𝑥 = 0
Flips |−⟩ to −|−⟩ if 𝑓𝑓 𝑥𝑥 = 1
The Deutsch-Josza Solution
What if both the control and target are in superposition?

Global Phase
Relative Phase

Measuring the control qubit in the X basis tells us

whether the function is constant or balanced in one query
The Deutsch-Josza Circuit

Measure the
Initialize Prepare the Implement
control in
both states “+” and “-” the DJ
the +/- basis
to “0” superpositions unitary
Deutsch-Josza with IBM Q
quantum-computing.ibm.com/ DJ Oracle for 𝑓𝑓(𝑥𝑥) = 1
Question Break
 Lessons from Deutsch-Josza
• Preparing the target in superposition wasn’t enough!
• We needed to also measure in a superposition basis

The First Rule of

Quantum Computing Club
It’s not just about
“querying all possibilities
in superposition”
 Lessons from Deutsch-Josza
The “target” measurement qubit wasn’t actually measured
The phase was “kicked back” to the control qubit

Key to many
quantum algorithms
Lessons from Deutsch-Josza
We still don’t know what 𝑓𝑓(𝑥𝑥) is exactly
x f1(x) x f2(x) x f3(x) x f4(x)
0 0 0 1 0 0 0 1
1 0 1 0 1 1 1 1
Constant Balanced Balanced Constant

We still need two queries to know which 𝑓𝑓 𝑥𝑥 we have,

but quantum computers allow us to extract some properties more efficienctly

The Second Rule of

Quantum Computing Club
Quantum computers
don’t speed up everything
 Lessons from Deutsch-Josza
It scales to many qubits

2 constant functions
2𝑛𝑛 2𝑛𝑛 ! We can design a quantum circuit
= 2𝑛𝑛−1 ! 2 balanced functions
2𝑛𝑛−1 which tells us in one query
Promise: It’s either constant or balanced if it’s constant or balanced

We need 2𝑛𝑛−1 + 1 queries Exponential

quantum Details a bit complicated!
to be 100% positive if
speedup! See the notes for more
𝑓𝑓 𝑥𝑥 is constant or balanced
 Classical vs. Quantum

Need one measurement to know the “state” Need three measurements to know the “state”
Classical vs. Quantum

2𝑛𝑛 numbers to describe

…
Exponential separation
As the system grows,
the quantum system is exponentially more complex

…
2𝑛𝑛 × 2𝑛𝑛 numbers to describe
But remember…

Quantum computers do not provide

exponential enhancement for every problem
The Second Rule of
Quantum Computing Club Factoring → Exponential

Search → Quadratic
Quantum computers
Many others → Nothing
don’t speed up everything
Many many others → Unknown
What’s it useful for?
Question Break

Check out other quantum algorithms

https://quantumalgorithmzoo.org/
by Stephen Jordan (Microsoft Quantum)
 The DiVincenzo
Criteria
Five simple rules for building my quantum computer
 The DiVincenzo Criteria

A minimum set of requirements needed to build quantum computers

If your system can do these, then there is hope

D.P. DiVincenzo,
Fortschritte der Physik 48, 771 (2000).
 The DiVincenzo Criteria
1. A scalable system with well-characterized qubits

…
 The DiVincenzo Criteria
1. A scalable system with well-characterized qubits

2. The ability to initialize the qubits to, e.g., |𝟎𝟎𝟎𝟎𝟎𝟎 …⟩

…
 The DiVincenzo Criteria
1. A scalable system with well-characterized qubits

2. The ability to initialize the qubits to, e.g., |000 …⟩

3. Decoherence times longer than gate times

 The DiVincenzo Criteria
1. A scalable system with well-characterized qubits

2. The ability to initialize the qubits to, e.g., |000 …⟩

3. Decoherence times longer than gate times

4. A universal set of quantum gates

 Universal Gate Sets
Controlled Phase

Controlled-Controlled-NOT

Might not be easy, but any gate can be faithfully approximated

 The DiVincenzo Criteria
1. A scalable system with well-characterized qubits

2. The ability to initialize the qubits to, e.g., |000 …⟩

3. Decoherence times longer than gate times

4. A universal set of quantum gates

5. A qubit-specific measurement capability

…
 The DiVincenzo Criteria
1. A scalable system with well-characterized qubits

2. The ability to initialize the qubits to, e.g., |000 …⟩

3. Decoherence times longer than gate times

4. A universal set of quantum gates

5. A qubit-specific measurement capability

i. An ability to connect with flying qubits

 Spin Systems

Use nuclear or electron spins in NMR/ESR systems

Couple through J coupling (spin-spin interaction)

Move towards nanoscale or monolayers for true single-systems

NMR QIP Review: J. Baugh et al, Physics in Canada (2007). arXiv:0710.1447

 Trapped Atoms and Ions

Trapped individual ions (e.g. Yb+) in dynamic electric traps,

or neutral atoms using optical tweezers

Use electronic energy states as qubits, fluorescence readout

Couple through collective motional modes

Review: C. Monroe and J. Kim, Science 339, 1164 (2013).
 Superconducting Circuits

3μm

Flux or charge quanta in “artificial atoms” as qubits

Write using circuit fab techniques (e.g. Al on Si)

Cool in dilution refrigerators, control with microwaves

Review: J. Clarke and F.K. Wilhelm, Nature 453, 1031 (2008).

Review: M.H. Devoret and R.J. Schoelkopf, Science 339, 1169 (2013).
 Photonics

Generate single photons by nonlinear optics or quantum emitters

Directly use light’s degrees of freedom (e.g. polarization)

Couple probabilistically, or directly generate entangled cluster

Review: P. Kok et al, Rev. Mod. Phys 79, 135 (2007).

 DiVincenzo and Photons
1. A scalable system with well-characterized qubits

2. The ability to initialize the qubits to, e.g., |000 …⟩

3. Decoherence times longer than gate times

4. A universal set of quantum gates

5. A qubit-specific measurement capability

φ0

φ1
Pick Your Qubit

Atoms & Ions Photons Spin Superconducting

Circuits
 Early Quantum Computing

There are many possible physical

systems, but they must satisfy
certain criteria

Check out other quantum algorithms

Quantum algorithms can have up https://quantumalgorithmzoo.org/
to exponential speedups, but only by Stephen Jordan (Microsoft Quantum)
with clever design!
Up next…

• Quantum Bullsh*t with Prof. Chris Ferrie at 1:30 Eastern!

• Live on YouTube, check Slack for the link
• QSYS On-Campus will start on Monday July 25
• Check the schedule for which lectures will be live-streamed