If annealing, superposition, and qubits all sound like new Netflix films to you, then I too was in the same position only a week ago, until I fell down the rabbit hole of quantum computing.

The diagram above, taken from Tyler Takeshita’s video (included in the resources), shows that using the AWS system has some real advantages like the rest of their Cloud hosting offering. While the actual quantum computers sit outside the AWS cloud, how you interact with them is very similar to how you’d do anything else on AWS.

You can set specific permissions within the AWS console to allow access to the quantum computing solution. This works similarly to MS AD, where you can create users, groups and roles with associated permissions.

Amazon Braket gives you access to quantum computers and allows users to test their code on quantum emulators, which can be managed using the intrinsic IAM system.

You only pay for what you use, and to save time and cost, you can use the quantum emulators to make sure your programming hasn’t got any technical debt and so you don’t waste money on quantum computer use.

The costing is based on a set price for a task (essentially the problem definition) at $0.30 and then the “shot”, which is the number of times the task is executed at around $0.00019 per shot (different machines have different prices).

Emulator prices work slightly differently in that you pay how long they’re in use.

After running a task on the quantum machine, the results are accessed in a secure AWS S3 Bucket for easy access within the AWS console.

This service monitors resources and applications you run on AWS in real-time. You can use CloudWatch to collect and track metrics, create alarms that watch metrics and send notifications or automatically make changes to the resources you are monitoring when a threshold is breached.

Concerning the use of quantum computers, it will allow you to look at historical data and give you a holistic perspective on how Amazon Braket is performing for your business.

A serverless monitoring tool that allows you to create event-driven applications at scale using events generated from your applications. This is useful when your system looks de-coupled when launched, but over time has become subject to a fragile API with a single point of failure.

Allows the user to create and manage cryptographic keys.

Now I’m no mathematician, having resat my Physics A-Level, and not getting very far with the .net teams Euler challenge in Rippleffect. Still, I see the broader point in using computing to solve problems.

Datasets which are large or complex are commonplace now with all things digital. These could range from life-saving pharmaceuticals, complex aero engineering, fintech data to military applications.

Historically companies have just used bigger and bigger computers, i.e. adding in more and more processes. But quantum computers work in a slightly different way.

The quantum processors store the data you’d usually associate with conventional chips as small magnetic fields (aka single flux quanta) rather than voltages over transistors.

The chips are made from Niorobium, which exhibits superconductor properties (zero electrical resistance) at low temperatures, but they need to be shielded from magnetic interference.

The operating environment is extreme. Forget the fan running at the back of your laptop when you’ve got MS Teams running. Forget the snazzy water-cooled motherboard your boyfriend treated himself to for gaming.

The quantum processor needs temperatures of around 0.01K which is significantly colder than interstellar space (3k). If you cast your mind back to “A-Level” physics, it is as good as absolute zero (-273 degrees C) as you’re likely to get.

Without this chill and the magnetic shielding, your quantum processor isn’t going to work. This is why quantum computers are so big; they have to include rack space for the conventional computer to feed them information, they need a cooling unit, and the processor needs up to 16 layers of shielding when in use.

I used the “D-Wave — Advantage System 4.1”, which operated with quantum annealing.

It can be loosely defined as “Seeks to use the intrinsic qualities of quantum mechanics to solve optimisation problems & probabilistic sampling”.

Optimisation problems are when you’re looking for the most efficient configuration out of many options. For instance, if you’re looking at building a car on a fixed budget but you have a variety of components, what combination (configuration) of those components will be the best (e.g. performance) for that fixed budget.

The latter, probabilistic sampling, is linked to the former but used to take a model sample from a probabilistic point of view to improve the model over time – one such application may be machine learning.

A more complex method where rather than relying on the natural order of a system reaching a low energy state (and giving you your answer), you attempted to control and manipulate that quantum state over time.

The downside of the gate model approach is having the quantum process fail because it is delicate. Still, the advantage is more significant mathematical problems can be solved by utilising the quantum computer.

Gated machines tend to have significantly lower qubits than their annealing relatives and are even more so in their infancy.

See “What is Quantum Annealing?” in the resources below for more info.

Amazon Braket also offers access to other quantum computers by Rigetti and IonQ.

It’s also worth noting here that Google does have a Quantum computer capability using the Sycamore processor (link in the resources section). I only found this out when I looked up Jeremy Hilton, who pulled one of the resources I’ve cited, to find he moved from D-Wave to Google.

If a picture says a thousand words, a video will say more. Rather than going into too much detail on how this hangs together (I’ll give some brief information), have a look at Tyler’s video as he talks the user through solving the Minimum Vertex Cover Problem. I’ve included a link in the resources about what this is.

Steps to using the quantum computer on Amazon Braket.

1. Set up or log in to your AWS account.

2. Search for S3 and create an S3 bucket with amazon-braket as the prefix (then some other name after that). e.g. amazon-braket-maffingoesquantum.

3. Set up a sub-folder and note the name.

4. Search Amazon Braket.

5. Under “Devices” click on and make a note of the ARN of the machine you want to use (there’s a handy copy button). Check it’s online too.

6. Make sure you’re in the right region as the quantum machine (e.g. US-West) won’t work if you’re in a different region.

7. Go to Notebooks and Create a Notebook (this takes a few minutes) – you can pick a user from the IAM – I left mine as root.

8. Within the Notebook, look at “Braket Examples”. Pick one to try.

9. Create New “Conda_Braket” (the CLI for submitting tasks to the quantum computer).

10. Dump in your code. Remember to watch out for “Or” examples. Some of the worked examples suggest two options; adding both will give you an error. Also, put in the S3 details you noted down earlier. Failure to do this will throw more errors.

11. Execute your code.

12. Head over to Tasks – if it has been executed, the correct details will be here. Make sure you’re in the right region!

13. Check your S3 bucket. The problem results should be located as a JSON file.

14. Congratulations, you’ve just gone quantum. Now tell everyone at work that’s what you did at the weekend when they said: “I went to Cheshire Oaks shopping”.

If you’re an accountant or a climate activist, the cost of a supercomputer is pretty high in terms of energy usage. They have thousands of processor chips that need to be powered and cooled, compared to a quantum computer that has just one.

This means that while there’s a lot of energy to keep the chip at 0.01K, a traditional supercomputer rack might use up to 10x that power without solving the problem 10x faster.

It’s not hard to see the potential use of quantum computers in real-world military situations, either. These include NP-Hardness problems that conventional supercomputers struggle to solve. For instance, consider the following purely hypothetical position:

NATO has a set of 10 mobile Patriot missile units (MIM-104) that it can deploy anywhere in a list of 100 accessible locations next to an aggressive non-NATO country.

NATO wants to cover as much geographic area as possible. However, some locations interfere with each other, which reduces their operational and, therefore, cost-effectiveness. Where should the MIM-104 sites be located?

Objective: Maximise coverage. 

Constraints: i. Only 10 Patriot units are available for operational deployment. ii. Choose locations that do not cause interference.

Given the number of potential combinations, doing this manually will involve a lot of complex guesswork and an attempt and failure each time.

If the situation is escalating quickly and the lead time to deployment is significant, the only realistic solution is looking at automation. Failure to find the most efficient setup could cost many lives in the event of a missile launch. In this case, a quantum computer could be queried by command and the locations fed out to the relevant units. This is especially useful if the number of locations or units changes quickly (which happens in conflict as ground is lost or gained and hardware destroyed). The updated numbers could be rerun and new orders distributed.

Back in civilian life, sadly, given the problems quantum computers are best suited to solve, their sheer size and cost, it’s unlikely you’ll be booting up Counter-Strike Source on one any time soon for a game with the chaps at D-Wave. But as they follow a similar path to conventional computers (smaller and cheaper) and become more accessible through portals like AWS right now – they can be used to help solve environmental, financial and engineering problems quickly and relatively cheaply.

I’ve included some additional reading in the form of Simon Singh’s “Code Book”. It first introduced me to the idea of quantum computers back in 2004 and their potential use to brute force attack standard SSL and remove the underpinning security of that little padlock on your browser.

Financial institutions like PayPal are looking at how they can “quantum-proof” their companies against this type of attack and also use quantum computers to analyse large amounts of data in their fraud protection departments, knows as the “quantum advantage” (see resource PayPal: Harnessing Quantum Computing in FinTech | D-Wave Qubits 2021 below).

Amazon Web Services (AWS)

http://aws.amazon.com/

Tyler Takeshita Amazon Braket Demo Video

D Wave – quantum computer company

https://www.linkedin.com/company/d-wave-systems-inc./

D-Wave Quantum Computer Tour

(three parts and a warning a few years old)

Google Quantum Computer Info

https://quantumai.google/quantum-computing-service

What is Quantum Annealing?

https://youtu.be/zvfkXjzzYOo

https://docs.dwavesys.com/docs/latest/c_gs_2.html

Minimum Vertex Cover Problem

https://en.wikipedia.org/wiki/Vertex_cover

For those who are okay with terrifying mathematics, check out:

Shor’s algorithm

https://en.wikipedia.org/wiki/Shor%27s_algorithm

Grover’s algorithm – used for databases

https://en.wikipedia.org/wiki/Grover%27s_algorithm

Satellite Problem akin to the Patriot Missile Scenario

https://github.com/dwave-examples/satellite-placement

Additional Reading – Simon Singh Code Book

PayPal: Harnessing Quantum Computing in FinTech | D-Wave Qubits 2021

Cover Image by

https://unsplash.com/@igrevsky