Quantum Computing - business changer?

Quantum - the latest buzzword, at the experimental stage with tools and technical implementations being rough and ready - but what does it mean for my company and how may it affect my company's operations?

A business answer

Historically businesses have always looked for "an answer" being the singular, infact the answers are really sets of possible answers to guide their operation, but culling during qualification to result with what they believe will be the strongest course of action. The move towards paradigm of agility has rewarded organisations that have been able to operate and quickly cycle to evaluate as the marketplace has become faster. Adapt or die.

However what happens over time is that companies "build on their strengths" by reducing the cost, iterating and delivering every smaller movements to adapt - convinced their products are reaching perfection with the customer. Enter a new competitor with a "disruptive" product and what was perfect is no longer perfect to which surprise and the inability to switch away from the group thought that built the initial product success exist. (my view here is that every few years an organisation needs to throw away it's strength argument and perform a full re-evaluation for obvious reasons)

So what if you could continuously consider all possibilities each time without the massive outlay in cost to perform it? Adding any new sources for analysis and modelling how that could impact your business or adapting quickly to the change? What new ways of working could your adopt to support it?

The corporate economic silver bullet? Some sell this as Quantum Computing.

Business analysis computation evolution

I'm going to attempt to explain how this relates to business computing now.

Serial computation. Processing scenarios one at a time. The majority of developed business applications unfortunately fits into this domain. 

Parallel computation. Processing sets of scenarios simultaneously with each element representing a single value for an input. It's essentially performing parallel serial processing. It allows reading the load over multiple systems but limited in the data structure and dependancies. Careful design and algorithms result in massively parallel computation that performs some or all of the work in parallel.  You may have heard of "GPU", the chip that computes the graphics of your screen right now by processing it's pixels in parallel. The result is a large data processor (with some drawbacks). The same chip can be "abused" from it's original design into performing processing of data within it's limitations. Numerical processing in oil, gas or statistics can be done - this is the reason Google use GPU to power their search engines using #TensorFlow.

What is important is that developers are getting experience in creating code where data can flow like wide rivers. It paves the way for Quantum Computing. 

Quantum computation. By referencing an input, the algorithm processes all possible values and therefore scenarios simultaneously for that input.  Another way to think if it - imagine a tree with it's leaves and root tips (inputs, each with it's range of values), branches and trunk sap tubes (quantum circuits).  Without cutting into it - all the possible scenarios exist for all the input values. You see the tree holistically from the outside but not it's internal levels, best leaves/roots that have the largest tubes (best answer) or state of sap - so how do you get the business answer for the best possible scenarios?

This is the complication in Quantum space.. how/what is the answer. Reading the system is akin to killing the tree by cutting it so we can see the state of it's innards to give that answer.  In normal computing you'd have a list of highest probabilities to the best business answer. In quantum you collapse all the qbits from being "all the values" to create the distribution. Depending on the how the programming (quantum circuit and algorithms used) this may need more than one run. (that's the simple explaination)

So the question you ask and how you ask both in business and developer implementation is paramount so the result (best inputs). To get the ROI for business answers out of quantum means developers need to create techniques to maximise the return and hardware designers need to make programmable Quantum systems to support that.

The Awakening - Age of Quantum

Technically Quantum is like the valves in the 1940s at the moment. You could draw a parallel between valves/transistors and the quantum gate that are chained together as quantum circuits to work on the limited number of value limited qubits (registers). To solve a problem needs building of a circuit in hardware. All cryogenically cooled. Compare that to the modern silicon microchip with millions of transistors and bytes of value storage for the processing. So it is possible to deliver quantum computing now problems with the programming done by building quantum circuits - just a little primitive to the modern generic CPU/software/flexible approaches.

So hardware and software vendors are rushing to both (a) prepare the developer community through training, experience and create fast cross-over tools and (b) developing the Quantum technology to be able to cope with real world processing demands.

Microsoft have just released a Quantum simulator for normal computers with just 30bits requiring 32GB of memory on the host machine. Their Quantum program publicly states they're working on tools that developers can use to bridge the gap without needing to worry about the quantum hardware aspects. I'd suspect that MS would then offer this as a Azure cloud service - including managing the Quantum hardware along with any data. The compiler takes the developer code, generating the quantum circuit from the code and then executes it by scheduling in the cloud. 

Other cloud compute vendors will attempt to deliver similar services, with development tools and environments being proprietary. I would certainly expect Wolfram to create Mathematica Quantum and Mathworks to create a MatLab Quantum with a corresponding cloud to host and manage the hardware.

University courses should be providing Quantum Computing, however there's a large "me too" between courses for applied statistics, data scientists, computer science and software engineering. The reality will always come back to one core thing - just like existing software, how do you best define the business problem (question) and then how do you best define the system.  I would expect better quality of development to come out of people that are familiar with statistics, numbers and set theory rather than those that have been taught C,C++,C#,.Net, Java etc are all serial instruction-based programming languages with parallel add-ons. Those such as Matlab, R and others where data is processed in probability or set-based maths will excel in Quantum era with the tools generating the quantum code and executing.

So what does that mean for my company?

In the short term it means that faster data processing - developing internal strategy to mirror the movement from serial computing through parallel and be ready for quantum when it becomes available.

Moving some of the expert business systems analysis (i.e. marketing, business intelligence, customer engagement analysis) from manual processes into computer maths models to harness the shift to faster data processing. The side result is that your businesses cycles costs will reduce, allowing for faster adaption to external change - with the need to perhaps review how executives and approval owners operate within the organisation to maximise speed.

A good example is wholesale processing business intelligence data to infer organisation's capabilities and possible strategic moves, as demonstrated in my rather detailed post: Quantum Jacking: Social & Big Data = security risks

How does your company operate at speed?
Will it be able to processes and adapt faster than your competitors?
Will you be able to perform continuous holistic analysis to prevent disruptive competitors side lining your strengths and products built on that?

Silver bullets?

Quantum Computing isn't a silver bullet. However it affords the capability to create new approaches in how the organisation does business based on the speed and scope of processing that could be possible.

#wolfram #matlab #microsoft #quantumcomputing #cloud #compute #gpu

The Author

Nick spent 4 years doing numerical computation, parallel systems, set theory and Z as a mandatory part of his university degree (no nervous twitches) before a long career in software integrators, product and project organisations - developing products, services and business processes, including strategic analysis. He's currently looking for roles in Security, IoT and Computational cloud products services.