In recent posts I shared with you how a quantum computer works using the nature qubits vs. the electronic switches, the transistors set for 0, and 1 in our everyday electronics. We know the ability of the qubits to maintain a superposition (both 0 and 1 and anything in between!) is the key to an exponential computation capability, far beyond todays super computers. We now know, even a 72 qubit quantum machine, if running at high fidelity and long coherence time can outperform our best super computers.
In this post, I explore how to assess a quantum computer based on methods of production. It is worth mention that all new trends of quantum computing that is explored today require a set of classical computers for control of the system.
For example, an Ion trapping system, or a superconducting system, even a quantum annealing QPU requires classical computers for controlling tasks such as initializing qubits at the ground state, read and process the output, finally to present solution via classical computer outputs.
Recently when learning about the Shor's algorithm, I found out that many tasks performed by classical computers and then the most challenging part - the process of implementing "period-finding algorithm," which is really equivalent to "prime number factoring" - sent out as a subroutine to the quantum computer.
So, what that said, how do we assess a quantum computer system. A system means, users can use the system without knowing exactly what is happening behind the scene. It means using the Q computer same as how we use our computers and smartphones. That is a system.
As I learned from reading the Quantum Computing: An Applied Approach, The book provides a checklist worth analyzing. Note that you download the entire book using that link from the publisher's website.
Universality
Fidelity
Scalability
Qubits
Circuit depth
Logical connectivity
Cloud access
Universality means if the machine, the system presented is whether a Turning-complete, or a universal. The device may be a non-universal annealer. See my posts on Di Vincenzo quantum computation specifications.
what it means a two-level system made of ensemble of atoms, ions, photons, etc., but whether the atoms (qubits) are not individually addressable. If that's the case, the system fails the Di Vincenzo test.
Fidelity means the ability for qubit to remain in coherence through a computation calculated as 1 minus the error rate (1 - ER). The critical part is the 1 or 2 qubit operation fidelity. You see, fidelity is harder to maintain when spanning 2 qubits with a CNOT gate. The example is when you apply single qubit operators, for example, as X or Y. You need long coherence and high fidelity for computation.
Scalability means if the system is capable of scaling to 10 to the power of 6 qubits or even go beyond. The current era called NISQ (noisy intermediate scale quantum) and current quantum computer designers working hard to scale up. Even 72 qubit system can do very useful things for humanity if running at high fidelity and long coherence. The Fault Tolerant era is about 5-10 years away, but breakthroughs can occur any time. I have my ears to the ground.
Qubits means the number of qubits. In this topic, there are many challenges such as avoid Cross-Talk of adjacent qubits. There are many different architectures for qubits.
Circuit depth refers to the number of operations we can run before the coherence breaks down. Imagine a 1 million qubit quantum computer but user can not process more than a few operations before coherence collapse. This system has low value.
Logical connectivity refers to this question: Can we run a two qubit gate on any pair of qubits from the chain, or only for certain pairs? The reason is that "limited logical connectivity" although can be mitigated by inserting SWAP operations (inside the algorithm to simulate greater connectivity) translates to more noise. More noise produces a noisy output.
Cloud access is the last criteria from the check list. CLOUD access translate to wide-spread use and developing a community.
Comments