Quantum Computing

[Jochen Fromm-5]

What do you think of Quantum Computing, will it be successful? IBM just built the IBM Q System One...

https://youtu.be/LAA0-vjTaNY

...while others make a strong case against it..

https://spectrum.ieee.org/computing/hardware/the-case-against-quantum-computing

..or even call it bullshit

https://scottlocklin.wordpress.com/2019/01/15/quantum-computing-as-a-field-is-obvious-bullshit/

What's your opinion? 

-Jochen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[Marcus G. Daniels]

Hi Jochen,

 

There are currently two main approaches to quantum computing.   The first is called adiabatic quantum computing (AQC).   Complexity enthusiasts that have followed the spin glass literature will be familiar with Ising spin systems.    AQC exploits the tendency of physical systems to go to low energy.   It turns out that many kinds of operations can be implemented even with two body interactions.   For example logic programs map to Ising systems.  In fact a former colleague of mine, Scott Pakin has implemented a Prolog interpreter for a quantum annealer.   The reason to use a quantum annealer over a classical thermal annealer is essentially speed.    An anneal can be repeated every tens of microseconds.   For sampling applications, like finding all the ways to solve an NP-hard constrained optimization problem with discrete variables, this is potentially a big win.   Classical mixed-integer programming approaches can be fast to find local/global optima, but may give a fragile picture of the fitness landscape.  It is possible to use annealers to mimic materials, and study phenomenon like quantum phase transitions.   Quantum tunneling and entanglement have been demonstrated using commercially-available quantum annealers.  

 

The other approach to quantum computing is the gate model.   Here the idea is to compose together unitary operators.   This model is less surprising from a from a (functional) programming perspective:  There are gates that feed into other gates kind of like classical circuits.  This is a what IBM (and others) are trying to do, and the difficulty of the task is reflected by the fact they only have a handful of qubits.   Estimates vary, but to implement the error correction that would give reliability on-par with classical digital computers could take 1000-fold or even more redundancy.   That’s to get _one_ good qubit at > 99.9% reliability.

 

AQC doesn’t require the long coherence times (especially resistance to dephasing) that the gate model requires.    Recently there’s been a middle ground declared called NISQ which is trying to find algorithms (like annealing) that work on imperfect gate-model qubits. 

 

There are two popular foundational technologies for qubits, superconductors and ion traps.   The tradeoff is essentially between latency and stability.  Ion traps can maintain coherence a long time, but are relatively expensive to configure.   The system you mention from IBM is a the former.  Superconductors typically operate near absolute zero (tens of mK) with many layers of protection from electromagnetic radiation and the Ion traps use elaborate laser control systems. 

 

Quantum computing is not BS, but it is very hard to engineer these systems and there is a long road ahead to bring this technology to practitioners. The CMOS-based computing systems we all use are a miraculous accomplishments of humans, and are easy to take for granted.   One of the national labs here in New Mexico actually owns an AQC system.

 

Marcus

 

From: Friam <[hidden email]> on behalf of Jochen Fromm <[hidden email]>
Reply-To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Date: Saturday, January 26, 2019 at 3:23 PM
To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Subject: [FRIAM] Quantum Computing

 

What do you think of Quantum Computing, will it be successful? IBM just built the IBM Q System One...

https://youtu.be/LAA0-vjTaNY

 

...while others make a strong case against it..

https://spectrum.ieee.org/computing/hardware/the-case-against-quantum-computing

 

..or even call it bullshit

https://scottlocklin.wordpress.com/2019/01/15/quantum-computing-as-a-field-is-obvious-bullshit/

 

What's your opinion? 

 

-Jochen

 

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[Jochen Fromm-5]

Yes, the gate model is described in books like "Quantum Computation and Quantum Information" from Michael Nielsen. The theory from David Deutsch et al seems to be established. We have Shor's algorithm, Grover's algorithm and many others but there still seems to be a significant gap between theoretical insights and experimental results. 

How do you program a AQC quantum computer? Somehow it must be setup to execute a certain type of calculation? 

And what do you think about photonic quantum computers? The Canadian company Xanadu from Toronto tries to go in this direction. 

https://www.xanadu.ai/

-Jochen

-------- Original message --------

From: Marcus Daniels <[hidden email]>

Date: 1/27/19 01:49 (GMT+01:00)

To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>

Subject: Re: [FRIAM] Quantum Computing

Hi Jochen,

There are currently two main approaches to quantum computing.   The first is called adiabatic quantum computing (AQC).   Complexity enthusiasts that have followed the spin glass literature will be familiar with Ising spin systems.    AQC exploits the tendency of physical systems to go to low energy.   It turns out that many kinds of operations can be implemented even with two body interactions.   For example logic programs map to Ising systems.  In fact a former colleague of mine, Scott Pakin has implemented a Prolog interpreter for a quantum annealer.   The reason to use a quantum annealer over a classical thermal annealer is essentially speed.    An anneal can be repeated every tens of microseconds.   For sampling applications, like finding all the ways to solve an NP-hard constrained optimization problem with discrete variables, this is potentially a big win.   Classical mixed-integer programming approaches can be fast to find local/global optima, but may give a fragile picture of the fitness landscape.  It is possible to use annealers to mimic materials, and study phenomenon like quantum phase transitions.   Quantum tunneling and entanglement have been demonstrated using commercially-available quantum annealers.  

 

The other approach to quantum computing is the gate model.   Here the idea is to compose together unitary operators.   This model is less surprising from a from a (functional) programming perspective:  There are gates that feed into other gates kind of like classical circuits.  This is a what IBM (and others) are trying to do, and the difficulty of the task is reflected by the fact they only have a handful of qubits.   Estimates vary, but to implement the error correction that would give reliability on-par with classical digital computers could take 1000-fold or even more redundancy.   That’s to get _one_ good qubit at > 99.9% reliability.

 

AQC doesn’t require the long coherence times (especially resistance to dephasing) that the gate model requires.    Recently there’s been a middle ground declared called NISQ which is trying to find algorithms (like annealing) that work on imperfect gate-model qubits. 

 

There are two popular foundational technologies for qubits, superconductors and ion traps.   The tradeoff is essentially between latency and stability.  Ion traps can maintain coherence a long time, but are relatively expensive to configure.   The system you mention from IBM is a the former.  Superconductors typically operate near absolute zero (tens of mK) with many layers of protection from electromagnetic radiation and the Ion traps use elaborate laser control systems. 

 

Quantum computing is not BS, but it is very hard to engineer these systems and there is a long road ahead to bring this technology to practitioners. The CMOS-based computing systems we all use are a miraculous accomplishments of humans, and are easy to take for granted.   One of the national labs here in New Mexico actually owns an AQC system.

 

Marcus

 

From: Friam <[hidden email]> on behalf of Jochen Fromm <[hidden email]>
Reply-To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Date: Saturday, January 26, 2019 at 3:23 PM
To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Subject: [FRIAM] Quantum Computing

 

What do you think of Quantum Computing, will it be successful? IBM just built the IBM Q System One...

https://youtu.be/LAA0-vjTaNY

 

...while others make a strong case against it..

https://spectrum.ieee.org/computing/hardware/the-case-against-quantum-computing

 

..or even call it bullshit

https://scottlocklin.wordpress.com/2019/01/15/quantum-computing-as-a-field-is-obvious-bullshit/

 

What's your opinion? 

 

-Jochen

 

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[Marcus G. Daniels]

Jochen writes:

 

“How do you program a AQC quantum computer? Somehow it must be setup to execute a certain type of calculation?”

 

An AQC program can be thought of as a graph where the nodes have a value that represents a linear bias up or down for qubit spins in the problem.   Values on edges in the graph represent the tendency of the spins to attract or repel one another.   (The graph is sent to the annealer as a matrix.)   The output is a vector of spins that have a Boltzmann-like distribution given the relative magnitude of the coefficients in the graph to the finite temperature of the machine.    You can find examples on the web of factoring / inverting calculations, social network algorithms, vehicle routing, and a range of other applications.    There are theory papers (Aharonov 2004) that demonstrate that AQC is equivalent to the gate model.  

 

“ And what do you think about photonic quantum computers? The Canadian company Xanadu from Toronto tries to go in this direction. 

https://www.xanadu.ai/  “

 

Another well-known one is IonQ.   One of their founders gave a public lecture in Santa Fe a few months ago.  These are intriguing systems, but they aren’t big enough yet to do meaningful calculations.  Honeywell is getting into that area too.  And there are some smaller start-ups like Atom Computer.

 

Marcus

 

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[Marcus G. Daniels]

The main practical difference is that their approach uses continuous variables, not discrete qubits.   I can’t figure out exactly what they have managed to fabricate.  There’s a group in Australia that does similar research,  see _Integrated photonic platform for quantum information with continuous variables_ vs. _Continuous-variable gate decomposition for the Bose-Hubbard model_.  

 

From: Friam <[hidden email]> on behalf of Jochen Fromm <[hidden email]>
Reply-To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Date: Sunday, January 27, 2019 at 4:40 AM
To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Subject: Re: [FRIAM] Quantum Computing

 

Yes, the gate model is described in books like "Quantum Computation and Quantum Information" from Michael Nielsen. The theory from David Deutsch et al seems to be established. We have Shor's algorithm, Grover's algorithm and many others but there still seems to be a significant gap between theoretical insights and experimental results. 

 

How do you program a AQC quantum computer? Somehow it must be setup to execute a certain type of calculation? 

 

And what do you think about photonic quantum computers? The Canadian company Xanadu from Toronto tries to go in this direction. 

https://www.xanadu.ai/

 

-Jochen

 

 

-------- Original message --------

From: Marcus Daniels <[hidden email]>

Date: 1/27/19 01:49 (GMT+01:00)

To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>

Subject: Re: [FRIAM] Quantum Computing

 

Hi Jochen,

There are currently two main approaches to quantum computing.   The first is called adiabatic quantum computing (AQC).   Complexity enthusiasts that have followed the spin glass literature will be familiar with Ising spin systems.    AQC exploits the tendency of physical systems to go to low energy.   It turns out that many kinds of operations can be implemented even with two body interactions.   For example logic programs map to Ising systems.  In fact a former colleague of mine, Scott Pakin has implemented a Prolog interpreter for a quantum annealer.   The reason to use a quantum annealer over a classical thermal annealer is essentially speed.    An anneal can be repeated every tens of microseconds.   For sampling applications, like finding all the ways to solve an NP-hard constrained optimization problem with discrete variables, this is potentially a big win.   Classical mixed-integer programming approaches can be fast to find local/global optima, but may give a fragile picture of the fitness landscape.  It is possible to use annealers to mimic materials, and study phenomenon like quantum phase transitions.   Quantum tunneling and entanglement have been demonstrated using commercially-available quantum annealers.  

 

The other approach to quantum computing is the gate model.   Here the idea is to compose together unitary operators.   This model is less surprising from a from a (functional) programming perspective:  There are gates that feed into other gates kind of like classical circuits.  This is a what IBM (and others) are trying to do, and the difficulty of the task is reflected by the fact they only have a handful of qubits.   Estimates vary, but to implement the error correction that would give reliability on-par with classical digital computers could take 1000-fold or even more redundancy.   That’s to get _one_ good qubit at > 99.9% reliability.

 

AQC doesn’t require the long coherence times (especially resistance to dephasing) that the gate model requires.    Recently there’s been a middle ground declared called NISQ which is trying to find algorithms (like annealing) that work on imperfect gate-model qubits. 

 

There are two popular foundational technologies for qubits, superconductors and ion traps.   The tradeoff is essentially between latency and stability.  Ion traps can maintain coherence a long time, but are relatively expensive to configure.   The system you mention from IBM is a the former.  Superconductors typically operate near absolute zero (tens of mK) with many layers of protection from electromagnetic radiation and the Ion traps use elaborate laser control systems. 

 

Quantum computing is not BS, but it is very hard to engineer these systems and there is a long road ahead to bring this technology to practitioners. The CMOS-based computing systems we all use are a miraculous accomplishments of humans, and are easy to take for granted.   One of the national labs here in New Mexico actually owns an AQC system.

 

Marcus

 

From: Friam <[hidden email]> on behalf of Jochen Fromm <[hidden email]>
Reply-To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Date: Saturday, January 26, 2019 at 3:23 PM
To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Subject: [FRIAM] Quantum Computing

 

What do you think of Quantum Computing, will it be successful? IBM just built the IBM Q System One...

https://youtu.be/LAA0-vjTaNY

 

...while others make a strong case against it..

https://spectrum.ieee.org/computing/hardware/the-case-against-quantum-computing

 

..or even call it bullshit

https://scottlocklin.wordpress.com/2019/01/15/quantum-computing-as-a-field-is-obvious-bullshit/

 

What's your opinion? 

 

-Jochen

 

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[Steve Smith]

FWIW, my nephew is in the middle of his PhD in Materials Science in the UofA working on their Phononic Quantum Computing program.    I can't suss out how fast it is moving... he only sees the project through the tiny sliver of his own contribution, but it *seems* to promise much larger PhiBit (Phononic QBit) counts at room temperature, etc.   Some of the underlying engineering is quite fascinating (e.g. modulating the speed of sound in glass rods using coherent light to simulate an exponential "horn").

https://mse.engineering.arizona.edu/news-events/mse-researcher-uses-phononics-build-quantum-computers

I've been trying to wrap my head around the implications of AQC for the kinds of problems I've been using 3D graph layout to try to develop intuition on.   The goal is *rarely* to truly minimize the energy of the system (modeled as a graph with *vectors* of edge-weights rather than simple scalars) to explore the trade-space of the systems being modeled.   In the problem-domain of interest, the challenge is not to find "the answer" but rather to explore the implied landscape of high-dimensional problems.

- Steve

On 1/27/19 9:00 AM, Marcus Daniels wrote:

Jochen writes:

 

“How do you program a AQC quantum computer? Somehow it must be setup to execute a certain type of calculation?”

 

An AQC program can be thought of as a graph where the nodes have a value that represents a linear bias up or down for qubit spins in the problem.   Values on edges in the graph represent the tendency of the spins to attract or repel one another.   (The graph is sent to the annealer as a matrix.)   The output is a vector of spins that have a Boltzmann-like distribution given the relative magnitude of the coefficients in the graph to the finite temperature of the machine.    You can find examples on the web of factoring / inverting calculations, social network algorithms, vehicle routing, and a range of other applications.    There are theory papers (Aharonov 2004) that demonstrate that AQC is equivalent to the gate model.  

 

“ And what do you think about photonic quantum computers? The Canadian company Xanadu from Toronto tries to go in this direction. 

https://www.xanadu.ai/  “

 

Another well-known one is IonQ.   One of their founders gave a public lecture in Santa Fe a few months ago.  These are intriguing systems, but they aren’t big enough yet to do meaningful calculations.  Honeywell is getting into that area too.  And there are some smaller start-ups like Atom Computer.

 

Marcus

 

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[Marcus G. Daniels]

AQC, and quantum computers in general, are analog computers.    They are good for sampling, but to really find the lowest energies on a complex problem, some classical assist is helpful.

 

From: Friam <[hidden email]> on behalf of Steven A Smith <[hidden email]>
Reply-To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Date: Sunday, January 27, 2019 at 10:32 AM
To: "[hidden email]" <[hidden email]>
Subject: Re: [FRIAM] Quantum Computing

 

FWIW, my nephew is in the middle of his PhD in Materials Science in the UofA working on their Phononic Quantum Computing program.    I can't suss out how fast it is moving... he only sees the project through the tiny sliver of his own contribution, but it *seems* to promise much larger PhiBit (Phononic QBit) counts at room temperature, etc.   Some of the underlying engineering is quite fascinating (e.g. modulating the speed of sound in glass rods using coherent light to simulate an exponential "horn").

https://mse.engineering.arizona.edu/news-events/mse-researcher-uses-phononics-build-quantum-computers

I've been trying to wrap my head around the implications of AQC for the kinds of problems I've been using 3D graph layout to try to develop intuition on.   The goal is *rarely* to truly minimize the energy of the system (modeled as a graph with *vectors* of edge-weights rather than simple scalars) to explore the trade-space of the systems being modeled.   In the problem-domain of interest, the challenge is not to find "the answer" but rather to explore the implied landscape of high-dimensional problems.

- Steve

On 1/27/19 9:00 AM, Marcus Daniels wrote:

Jochen writes:

 

“How do you program a AQC quantum computer? Somehow it must be setup to execute a certain type of calculation?”

 

An AQC program can be thought of as a graph where the nodes have a value that represents a linear bias up or down for qubit spins in the problem.   Values on edges in the graph represent the tendency of the spins to attract or repel one another.   (The graph is sent to the annealer as a matrix.)   The output is a vector of spins that have a Boltzmann-like distribution given the relative magnitude of the coefficients in the graph to the finite temperature of the machine.    You can find examples on the web of factoring / inverting calculations, social network algorithms, vehicle routing, and a range of other applications.    There are theory papers (Aharonov 2004) that demonstrate that AQC is equivalent to the gate model.  

 

“ And what do you think about photonic quantum computers? The Canadian company Xanadu from Toronto tries to go in this direction. 

https://www.xanadu.ai/  “

 

Another well-known one is IonQ.   One of their founders gave a public lecture in Santa Fe a few months ago.  These are intriguing systems, but they aren’t big enough yet to do meaningful calculations.  Honeywell is getting into that area too.  And there are some smaller start-ups like Atom Computer.

 

Marcus

 

============================================================

FRIAM Applied Complexity Group listserv

Meets Fridays 9a-11:30 at cafe at St. John's College

to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com

archives back to 2003: http://friam.471366.n2.nabble.com/

FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove