Exploring Quantum

Computing with Qiskit

July 15, 2020

Doug McClure

Research Staff Member

Manager of Quantum System Deployment
IBM Quantum
Caffeine: C8H10N4O2
 Moore’s Law
• Transistors get smaller and cheaper
Performance

3 © 2017 IBM Corporation

3
July 2019 IBM Confidential – Internal Use Only
 Moore’s Law
• Transistors get smaller and cheaper
• Underlying model of computation still the same
Performance

=
Galaxy Z Flip
(2020)

ENIAC (1945)

4 © 2017 IBM Corporation

4
July 2019 IBM Confidential – Internal Use Only
 Beyond Moore’s Law
• Quantum computing is an entirely new model of
computation, NOT simply an extension of Moore’s law
Performance

= =
Galaxy Z Flip
(2020)
IBM Quantum
ENIAC (1945) 20-qubit system
(2018)
5 © 2017 IBM Corporation
5
July 2019 IBM Confidential – Internal Use Only
The road Quantum Quantum Quantum
to quantum Science Ready Advantage
advantage Created the Engage the world Beneficial to use a
fundamental and prepare for quantum computer
theoretical and the quantum to solve real-
physical building computing era. world problems.
blocks of quantum
computing.

IBM Quantum Experience

Launched (May 2016)
Qiskit v0.1 released
(Mar 2017)
IBM Quantum Network
Launched (Dec 2017)

1960s 2016 ~2020s 2050+

 Quantum mechanics:
a two-sided coin

Superposition Uncertainty

Entanglement Decoherence
Computing with quantum mechanics: features
Superposition: a system’s state can be any
linear combination of classical states …until
it is measured, at which point it collapses to
one of the classical states
Example: Schrodinger’s Cat “Classical” states

Quantum Normalization
wavefunction

Entanglement: particles in superposition

can develop correlations such that
measuring just one affects them all
Example: EPR Paradox (Einstein: “spooky Linear
action at a distance”) combination
Computing with quantum mechanics: challenges
1
Decoherence: a system is gradually measured
Qubit State

by residual interaction with its environment,

killing quantum behavior
Consequence: quantum effects observed only
in well-isolated systems (so not cats… yet)
0
Time
Uncertainty principle: measuring one
variable (e.g. position) disturbs its
conjugate (e.g. momentum)
Consequence: complete knowledge of an
arbitrary quantum state is impossible.
→ “No-Cloning Theorem”
What should a quantum bit look like?
Classical bits:
Physical systems: capacitor charge,
transistor state, magnetic polarization,
presence or absence of a punched hole, etc.

Logical states: 0 and 1

Quantum bits (“qubits”)

Physical systems: spin of an electron, state of an atom,
superconducting circuits

Logical states: |0>, |1>, superpositions thereof.

Represented on the Bloch sphere
Physical qubit systems
Superconducting circuits
Atomic systems Electron spins

• Straightforward wafer-scale
Image: http://www.quantumoptics.at/
Topological fabrication with established
materials and processes
systems?
Majorana fermions
Image: http://vandersypenlab.tudelft.nl/
• Accurate device design with
Photons standard software
• Scalable architecture with
circuit QED paradigm
Image: http://topocondmat.org/
w2_majorana/braiding.html • Control and readout using
Image: PSIQuantum readily available components
Superconducting quantum processor building blocks
Superconducting Transmon Qubits:
▪ Josephson Junction acts as a non-linear inductor, allowing
isolation of lowest two allowed energy levels

100 nm
X 100 nm

Superconducting
Microwave Resonators:
▪ read-out of qubit states
▪ multi-qubit quantum bus
▪ noise filter

Phys. Rev. A 76, 04319 (2007)

Anatomy of a quantum chip
Qubits:
Single-junction transmon
Frequency ~ 5 GHz
Anharmonicity ~ 0.3 GHz

Resonators:
Co-planar waveguide
Frequency ~ 6 – 7 GHz
Roles:
1. Individual qubit readout
2. Qubit coupling (“bus”)

Ground plane
Periodic holes prevent 1 mm
stray magnetic field from
hurting superconductor
performance Corcoles et al., Nat. Commun. 6, 6979 (2015)
Controlling individual superconducting qubits
• Typically |𝟎ۧ and |𝟏ۧ differ in energy by E01 ~ 20 meV
• We drive this transition with a microwave pulse at frequency E01/h ~ 5 GHz
• While pulse is on, qubit undergoes Rabi oscillations between |𝟎ۧ and |𝟏ۧ
• Applying a pulse for just the right time and amplitude (a “pi pulse”) flips the qubit
“pi pulse” = NOT gate
Z
| 𝟎ۧ
100%

measuring |𝟎ۧ
Probability of
| −ۧ
| ۧ | ۧ
Y
|+ۧ
| 𝟏ۧ 0%
X
Pulse length
 CNOT Gate Operation
Generating entanglement Initial State Final State
Various approaches demonstrated: Control Target Control Target
– Fast frequency tuning with flux bias |𝟎ۧ |𝟎ۧ |𝟎ۧ |𝟎ۧ
– Tunable couplers |𝟏ۧ |𝟎ۧ |𝟏ۧ |𝟏ۧ
– All-microwave control superposition → a |𝟎ۧ + b |𝟏ۧ |𝟎ۧ a |𝟎𝟎ۧ + b |𝟏𝟏ۧ

IBM Quantum systems use CNOT implemented entanglement!

using cross-resonance technique (all-microwave)

– Bus resonator provides static coupling between
neighboring qubits
– Drive control at target’s frequency → target
oscillates at rate dependent on state of control
– Adjust amplitude/time of pulse to get CNOT
– Error rates around 1% in multi-qubit devices
Superconducting qubit control and readout electronics
Qubit control:
• RF signal source
RF Source: I/Q Mixer
• Produces a continuous sine wave at a
requested frequency/power @ f01 I

• Arbitrary waveform generator AWG: Ch1

Q
• Generates and outputs a Ch2
to qubit
programmed pulse envelope
Qubit readout:
• Mixer
RF Source 1: to readout resonator
• Multiplies a sine wave by an envelope @ fr
to produce an RF pulse
• Down-converts readout signal to MHz RF Source 2:
range to facilitate digitization @ fr + D
readout
• Digitizer Digitizer: signal
@D
• Captures down-converted readout
signal for analysis
Superconducting qubit environment and signal flow
• Cryo temperatures required
▪ Qubits sit at base of dilution refrigerator

• Control and readout performed by

sending pulses over coaxial cables

• Input lines use attenuation to

reduce incoming noise

• Output lines use cryogenic

amplification and isolation
 Launched May 4, 2016

Free, cloud-based GUI

and programmatic access
to small quantum devices
and simulators

Detailed user guide with

example algorithms

> 200,000 users

IBM Quantum Experience > 150 billion circuits run

> 200 scientific papers

 Quantum computing through the cloud
1. User submits “circuits” (sets of instructions) via API

2. Control computer directs instruments to send

pulses to quantum chip

3. Readout signals are analyzed to determine

qubit states at end of each circuit

4. Typically repeat many times to average

away fluctuations

5. Results sent to user

Classical computer API server Quantum computer

Control computer
Control instruments
19 © 2017 IBM Corporation
Writing quantum circuits: the “quantum score”

• “Textbook” way of showing quantum circuits

• Conducive to user-friendly drag-and-drop interface
arxiv.org/pdf/1905.02666.pdf
• Useful for beginners studying simple circuits
• Becomes unmanageable for large/complex circuits
Quantum programming desires

• Build and run circuits • Simulate device behavior

• Study and mitigate errors • Solve real-world problems

 The elements of Qiskit

Open Source • Build and run circuits • Simulate device behavior

(Apache 2.0)

Written in
Python 3
Terra Aer

Modular and
extendible • Study and mitigate errors • Solve real-world problems

qiskit.org
Ignis Aqua
 Basic workflow (Qiskit Terra)
▪Define → build → compile → run → retrieve
Define quantum circuits Compile and run Get results
State Counts
00000 513
00011 487

0.5

Probability
0.0
00000 00011
Outcome
23 © 2017 IBM Corporation
Designing algorithms for today’s quantum computers

• Quantum processors are noisy → long circuits won’t work!

▪ Design algorithms to use many small circuits rather than a single big one

• Example: “hybrid” quantum-classical optimization

▪ Quantum processor calculates objective function for classical optimizer
▪ Applicable to many problems including quantum chemistry (below)

Prepare a trial state 𝝍 𝜽 Use classical optimizer to

and compute its energy 𝑬(𝜽) guess a better value of 𝜽
Quantum Chemistry with the
Variational Quantum Eigensolver (VQE)

Prepare a trial state 𝝍 𝜽 Use classical optimizer to

and compute its energy 𝑬(𝜽) guess a better value of 𝜽

Black dots: VQE results

Density plots: numerical

simulations (classical)
Dashed lines: exact
calculations

More recently: improved

accuracy using error
mitigation technique
Tour of IQX Platform

quantum-computing.ibm.com

arxiv.org/pdf/1905.02666.pdf
Dashboard

arxiv.org/pdf/1905.02666.pdf
Backend detail view

arxiv.org/pdf/1905.02666.pdf
Documentation and support

arxiv.org/pdf/1905.02666.pdf
arxiv.org/pdf/1905.02666.pdf
Circuit composer: quantum score GUI

arxiv.org/pdf/1905.02666.pdf
arxiv.org/pdf/1905.02666.pdf
Jupyter notebook environment

easy access to notebooks in qiskit-iqx-tutorials

arxiv.org/pdf/1905.02666.pdf

Start a new notebook from scratch – or import one

Hands-On Exercise: creating
superposition and entanglement

quantum-computing.ibm.com
Step-by-step instructions at
https://github.com/dtmcclure/exploring-qc-with-qiskit

arxiv.org/pdf/1905.02666.pdf
Join the global Qiskit community
▪Diverse developer and user community
▪Slack workspace for questions and discussions
▪Online and in-person events (contests, hackathons, camps)

35 © 2017 IBM Corporation

Learn more!

Explore the IBM Quantum Experience Learn the basics of programming Discover more about
and start using real machines today quantum computers with Qiskit IBM’s quantum
(don’t miss the embedded tutorial at (I particularly recommend the computing initiative
https://quantum- Coding with Qiskit video series
computing.ibm.com/docs/guide) and the Qiskit Textbook)

ibm.co/iqx qiskit.org ibm.com/IBMQ