Bit or qubit? The future of high-performance and quantum computing
- Posted on November 23, 2020
- Estimated reading time 7 minutes
Quantum computing (QC) is an emerging technology with a lot of promise. It fundamentally changes the way we perform calculations by relying on probabilities rather than logical paths. But how does this new technology compare to high-performance computing (HPC), an already existing technology that has proven its usefulness in solving complicated problems? Will there be a competition between quantum and high performance within the near future?
A fundamental concept in quantum computing is the idea of qubits. Qubits are the most basic logical unit of quantum computing. The standard computing equivalent is a bit. A notable characteristic of qubits is the idea of superposition.
In quantum physics, this refers to the concept that the state of a physical system can be in one or many states at once. This means that the qubit can exist as 1, 0, or both. However, when this qubit is observed or measured, it will collapse to a single state.
The importance of that characteristic is difficult to understand when looking at one qubit but consider two qubits. Two qubits can represent a greater combination of states simultaneously because each qubit can be either a 1 or a 0, giving the possible combinations all at once: 00, 01, 11, 10. For conventional bits to achieve this combination, they must be manually flipped.
Another key concept of quantum computing is entanglement. Entanglement is when two quantum particles are put into an entangled state where they have an intrinsic relationship that can span large distances even across the universe. For example, we will entangle two qubits and then separate them. If we observe one of the qubits and the result is 1 then we know without a doubt that if we were to observe the second qubit that the result would also be 1. This concept is hard to comprehend because it seems to contradict the normal laws of physics as there is no data transferred between these two particles. They are simply linked in an unknown fashion.
Professionals in computing and physics alike are excited about the potential that quantum computing can hold. While the industry is still years away from being able to properly utilize its power, dreamers and scientists alike are looking towards quantum as the key to unlocking new scientific breakthroughs, performing calculations that traditional computers cannot, and performing certain algorithms at never-before-seen speed.
How can quantum computing outperform high-performance computing?
Grover’s Algorithm is a very famous and common example of how a quantum algorithm can achieve an answer faster than a traditional algorithm. Grover’s Algorithm is essentially an optimized way to zero in on a unique input to a black box or mystery function that produces the desired output. To help understand this algorithm, let’s look at its application in searching unstructured data.
The idea is that each piece of data is assigned an index such as 0000, 0001, 0010, etc. For the sake of this problem, let's also say that we have an oracle function that returns 1 when the input is desired index and 0 for every other input.
Searching unstructured data via traditional methods requires brute force. Because the data is unstructured, the traditional algorithm would need to look at each entry in the database. This would mean the number of steps required to search the entire database would be O(n) where n is the size of the database. So how can we speed this up?
The quantum algorithm takes a different approach. This approach is to first put a series of qubits equal to the number of input bits in the index in superposition. This means that these qubits have an equal opportunity to collapse to every answer when observed. We then pass these qubits through the oracle function from the problem definition. What does this do? It inverts the possibility that the qubits collapse to the correct answer but leaves the rest of the probabilities alone. This isn’t very helpful at the current moment, but we can use Grover’s diffusion operator to make all other probabilities interfere destructively with each other. However, we must perform this operation multiple times to decrease the possibility that the qubits collapse to the wrong answer to essentially 0. The number of steps it takes to achieve this result is √n where n is the size of the database. This makes the algorithm O(√n) and when compared to O(n) of the traditional method and at the large size the speed up significantly.
This example is oversimplified, and Grover’s Algorithm is not a solution that would be put into production, but it shows an example of how quantum computing can be extremely useful.
What is happening right now in the industry?
The computing industry has predicted that quantum computing will be able to solve business problems such as weather predictions, financial markets, security, and hybrid computing model. IT giants such as IBM, Google, and Rigetti are attempting to build quantum computers that will achieve quantum supremacy.
The idea of quantum supremacy is a matter of much debate, but the basic concept is that a quantum computer has achieved quantum supremacy once it can outperform a traditional computer which is simulating that same quantum computer logic. This measurement is important because if we are never able to prove that a quantum computer can outperform a traditional one then the concept of quantum computing becomes obsolete.
Recently, Google has claimed that it has achieved quantum supremacy with its Bristlecone chip. However, IBM has published a rebuttal to Google’s claim stating the traditional computer simulating the quantum circuits could have been more optimized, which would shrink the gap between the speed of the quantum computer and the traditional one. This shows that that quantum computing discoveries and claims are highly criticized and questioned.
Major nations are spending billions of dollars on quantum computing. Google AI scientists claim their quantum computer “Sycamore” can complete a task involving the verification of the randomness of large numbers in just three minutes and twenty seconds. For companies to continue to achieve a high level of supremacy, better algorithms need to be developed. Progressing in quantum computing will enable unique problem-solving capabilities. For companies to reap these benefits, the continuing fast pace of research and development must take place. Thankfully for the sake of progress, it seems that quantum computing research shows no sign of slowing down. Below is an example of advancements made in quantum computing by showing the exponential growth of the number of qubits in computing systems:
HPC’s current market and projection
According to Gartner Forecasts, by 2022 the market size and growth of the cloud services industry will be nearly three times the growth of overall IT services. It was valued at USD 9.35 billion in 2019 and is expected to reach USD 18.74 billion by 2025.
Amazon as the pioneer and the leader in the Cloud Sector has grown 50% in 2018 and constantly adding state of art features to its HPC series for Genomic sequencing, Reservoir Simulation, and speech and facial recognition. Microsoft is also expanding its services to include every type of workload and has a dedicated VM under the 'H-series' line up for HPC computing. Which are valuable for the calculations required in fluid dynamics and weather forecasting. Similarly, Google has invested in state-of-the-art infrastructure at competitive pricing and discounts help you stay within budget.
There has also been a steady increase from the US government in mil-aero areas, including satellite image processing, signal, and intelligence processing. HPCs are playing a pivotal role in training and simulation related to communications, intelligence, and surveillance.
Quantum and HPC: Allies not enemies
At this moment, HPC still offers better performance for our current computation problems. This is apparent from a large amount of investment in HPC technologies from tech giants such as Microsoft, Amazon, and Google. The practical technology for quantum computing is nonexistent but with recent claims of quantum supremacy, there is a chance we are closer than we thought. The future of computing is not a direct competition of these two technologies but rather a marriage both. One technology can pick up where the other cannot. In examples such as massively parallel workloads, HPC will still offer better performance over quantum but in fields such as pharmaceutics and molecular modelling, quantum pulls ahead. Together both technologies can work in tandem to solve humanity's greatest problems.