What Is a Quantum Computer?

Quantum computer is an emerging technology that will have a major impact

Ajay Shrestha

Follow

Published in

Towards Data Science

7 min readSep 8, 2022

--

Figure 1: IBM Quantum System One (CES 2020) / source link (CC BY-ND 2.0)

Quantum computing and AI can be combined to have an even greater impact. The goal of this blog post is to introduce basic concepts of quantum computers and try to demystify them. Quantum computers use properties of matter observed at the micro (sub-atomic) level to perform computations and solve problems. In contrast, current computers (since the 1950s) use properties of matter at a relatively macro (semi-conductor element) level. Progress in quantum computers will have significant implications in the near future. It could render current encryption standard of the Internet useless, reduce drug development time, transform AI and much more. Quantum computers have already demonstrated (e.g., Shor’s algorithm) exponential speedup compared to their non-quantum (aka classical computer) counterparts. This post covers the following at a high level.

Physics behind quantum computing
How/why it will outpace silicon based computing
What problems will it solve better

Classical to Quantum Physics

The transition to the digital world was made possible by harnessing the unique properties of silicon semi-conductors. It conducts electricity only after certain conditions are met and behaves as an insulator otherwise. This behavior is tapped to perform base level operations, e.g., AND/OR logic, basic arithmetic operations, etc., by silicon gates. Computers are built on top of these foundations.

Classical physics describes the properties of matter at the at element level (e.g., silicon, copper) and we control and use these properties for computation, data storage and information processing and transmission.

Similarly, quantum physics/mechanics explains the properties of matter at microscopic sub-atomic level, i.e., electrons and photons. Quantum computers control and leverage the properties of matter at the sub-atomic level to perform computations. However, the difference between classical and quantum physics is not just about scale or size. The rules of classical physics breakdown and don’t work at the quantum level. We need a completely new set of approach and engineering to build quantum computers.

Table 1: High level difference between Classical and Quantum Physics

It has been over a century since the framework for quantum mechanics was first developed, but there is no consensus among scientists on the implications of observations at the quantum level. Copenhagen interpretation, which is attributed to Niels Bohr and Werner Heisenberg is the most commonly accepted version. Even Albert Einstein didn’t fully accept the Copenhagen interpretation. In fact, Einstein’s theory of general relativity doesn’t align with quantum mechanics and he initially rejected quantum mechanics due to its indeterministic nature.

End-of-the-road for Silicon-based Computing

In 1965, Gordon Moore predicted the speed of computers would double roughly every 2 years. It was encoded as Moore’s law and has turned out to be true to this day. During this time, the size of silicon transistors have continuously reduced. The original size was that of a thumbnail, i.e., 1cm in 1950s. Now in 2022, it is 3 nanometer (i.e., ~7 orders of magnitude smaller than 1cm) and approaching the size of a silicon atom (.2 nanometer) . It would not be physically possible to shrink the size of silicon transistor below its atomic size to gain further speedup. Thus, some have called for the end of Moore’s law, which is based on semi-conductor based computing. The next phase of the computing innovation could be led by quantum computers.

Figure 2: Moore’s Law Chart in logscale / Source: ourworldindata.org & Karl Rupp (CC BY 4.0)

Quantum Solution

Innovation at higher-level software and architectures, e.g., in the form of machine learning and distributed/parallel computing have provided means to overcome some of the compute limitations to solve harder problems. But unlike these solutions, quantum computer provides computational gains from ground up. It has the potential to usher the next wave of innovation in computing to solve intractable problems. Computing gain from quantum computing will be orders of magnitude faster than what is predicted by Moore’s law. Since the speedup happens at the base layer, it can be combined with higher-level innovation like machine learning as quantum machine learning (e.g., Google Quantum AI), and commoditized & offered as managed cloud service (e.g., Amazon Braket Quantum Computers).

Quantum Properties

The two properties that are fundamental to quantum computing are:

Superposition
Entanglement

Before we get into them, it is important to grasp the concept of a qubit.

Qubit (vs Bit)

The unit of information in classical computing, storage and communication is a bit, which is represented as binary numbers 0 or 1. They are numerical representations of low and high charges (voltage) at the hardware (silicon transistors) level. The state of the hardware unit is always either 0 (low voltage) or 1 (high voltage). Things are not so discrete at the quantum level. Until a sub-atomic particle is observed (measured), it is in the state of both 0 and 1, i.e., an electron could be spinning up or down. This unit of information or state in quantum systems is referred to as qubit.

a) Superposition

This phenomenon of being both 0 and 1 simultaneously at the same time in quantum systems is called superposition. Superposition represents all possible states of the quantum bits (qubits). The superposition of different states collapses into a particular state during measurement. This was illustrated in the Schrödinger’s cat thought experiment, where a cat inside a sealed box together with a radioactive source is considered to be both alive and dead until one observes inside the box.

b) Entanglement

It is a phenomenon where 2 or more sub-atomic particles have the same state properties (e.g., 0 or 1) despite the distance between them. No one can explain why or how this occurs. If there were some communication between the sub-atomic particles to make this happen, the signal would have to travel faster than the speed of light, which as per Einstein’s theory of relativity is not possible. Einstein referred to entanglement as “spooky action at a distance”.

Quantum Vs Classical Computers

Quantum computers have multiple qubits in entangled states and carry out all possible combinations of qubits at the same time. They are collapsed into a desired state to solve specific problems. Since a qubit can be in 0 or 1 state at the same time, an n-qubit quantum computer can process 2ⁿ states all at the same time. Classical computers require doubling the number of bits (memory) or processing rate to double the speed. In quantum computer, just adding another quantum bit (i.e., qubit) doubles the speed. Thus, it can gain an exponential speedup over classical computers, which processes one state at a time. Theoretically, a 28-qubit quantum computer would be equivalent to over 268 million-bit (2²⁸) silicon based classical computer as shown in table 2.

Quantum supremacy is the stated goal when quantum computers can solve a problem that no classical computer can in reasonable amount of time. In October 2019, Google claimed they achieved quantum supremacy with 54-qubits.

Table 2: Comparing classical computers to quantum computers based on the bit/qubit sizes

Table 2: The 1st and 2nd columns were extracted from sources [1] and [2]. The third column is an extrapolation based on quantum properties. The 4th column is a theoritical comparison by me and not any official value. In fact, my values could be underreported, as Google confirmed their 54-qubit (5th row in table 2) quantum computer that achieved quantum supremacy in 2019 completed a task that would have taken a supercomputer 10,000 years in just 200 seconds.

Adding qubits to the architecture is not an easy task as they are highly susceptible to environmental conditions, which leads to loss of quantum properties. Quantum computers require near absolute zero temperature, i.e., about -460F, to operate.

Quantum Computer Use-cases

Not all problems will benefit from quantum computers just like not all problems are solved better by machine learning vs traditional software development as discussed in my earlier post.

Machine Learning Vs Software Development

The goal of software development (SD) and machine learning (ML) is to deliver features and capabilities to meet…

ajay-shrestha.medium.com

It is important to note that currently quantum computers’ use is limited to specific problems including some NP-hard problems, which entails sifting through large search space. These are some areas where quantum computers are already making a difference:

Drug Discovery
Encryption cracking
Optimization

Due to the superposition property, quantum computers can simulate all combinations of the entire solution search space in one cycle, compared to multiple iterations required by classical computers. Therefore quantum computers are best suited for search and combinatorial optimization problems. The use-cases listed above fall in this category.

Summary

Quantum computers are very promising and have already started to solve real world problems. As long as we can preserve the quantum properties of the sub-atomic particles by controlling the environment they operate in, we will be able to add more qubits to the architecture. Identifying which problems are better suited for quantum computers and building quantum algorithms are active research areas. So far, Grover’s_algorithm and Shor’s_algorithm are the most popular quantum algorithms that have provided polynomial and exponential speedup respectively compared to their non-quantum counterparts.

Both classical and quantum computer will co-exist in the future and will complement each other. Classical computers are better suited for tasks that are sequential (e.g., writing this blog post on my laptop) and transactional (e.g., calculating shopping cart balance), whereas quantum computers will be used to solve harder problems with large solution search space.