I recently attended the third annual Q2B Conference on December 10-12, 2019 in San Jose, California. The Q2B conference continues to grow and had about 540 attendees representing many different quantum hardware/software companies, end users, universities, government agencies, and venture capital communities. This number is double the attendance of the first Q2B conference in 2017.
This report is not a comprehensive listing of everything that went on at the conference. There were several sessions that occurred in parallel and I needed to choose which ones to attend. In addition, there were several sessions that covered topics that were similar to items covered in other conferences, published technical papers or previous Q2B events. So I will only cover those items that I found new, interesting and relevant. My apologies in advance to those folks who wanted to know about things that I do not cover here. However, QC Ware did record videos of all the presentations and will be posting most of the videos and presentations from the conference within the next month or two.
John Martinis presented the Sycamore chip and the previously published results of the Quantum Supremacy experiment. But the most interesting part of his talk was when he mentioned that they have already fabricated an improved version of the Sycamore chip and were currently testing it. Although he didn’t disclose full details of the chip, he did hint that the number of qubits have been increased from 53 to 57+. Also, this new device should have even better performance than Sycamore with a particular mention of improved readout fidelity.
Anthony Annunziata discussed the IBM Q systems and the IBM Q Network. They current have a total of 15 systems available in the cloud. The most interesting thing for us in this presentation was the announcement that they were opening up access to the OpenPulse API for people who want to research how to control the pulses that perform the actual control of the qubits. This will require newly released version 0.14 of Qiskit but will allow people to control the pulses over the cloud for the first time. More about this can be found in a newly released IBM quantum computing blog entry titled Get to the heart of real quantum hardware.
The most significant thing to us at Q2B was not the presentation itself, but the display in a glass case at their booth of a new cryo-CMOS control chip. The chip would be able to potentially control up to 50,000 qubits with just three wires that come in from outside the fridge. We recently posted a news article describing this chip in more detail and you can find the article here.
Honeywell has been in stealth mode with their ion trap technology development, but they pulled the curtains open slightly with a presentation by Tony Uttley. Although he didn’t mention too many of the technical details of their machines, he did mention three advantages that they have including long coherence time qubits, high resolutions for qubit rotations, and the ability to take a measurement on a single qubit and do conditional executions based upon the result (a quantum IF statement, if you will). This capability is enabled by the long coherence times of the ion traps and their ability to take the measurement on one qubit while keeping all the others in their quantum state. Although we will need to see the details of this when Honeywell has the broad commercial launch of their machines in the Spring of 2020 to fully understand how it works, we are not aware of any of the other quantum platforms currently having this capability. Later on, in a software session, one of their researchers Mike Foss-Feig presented a paper titled “Solving large problems with small quantum computers” that appears to utilize this capability.
Chad Rigetti discussed their 32-qubit Aspen-7 processor and their newly announced relationship with Amazon Web Services. We had previously covered this in a news article published earlier this month which you can see here. However, in his presentation he disclosed for the first time that they had implemented a new and additional family of parameterized two-qubit gates in their architecture called the XY(θ) gate. And when the value of θ is equal to ?, this will implement the iSWAP gate. The importance of having a richer set of gates available is that it allows a reduction in circuit depth which will reduce errors and improve solution quality. In a newly released blog paper, a Rigetti engineer indicated that this could provide an average gate depth reduction of 32% for a possible random circuit. For those of us coming from a classical computer background, this is somewhat like adding a new instruction to a microprocessor’s instruction set that allows one to collapse a multi-instruction sequence down to one. Details of this new family of two qubits gates can be found in a Rigetti blog entry here and a technical paper posted on arXiv here.
Xanadu is working in both the software and hardware areas. In the software area they have been working on several quantum readiness projects using their Strawberry Fields and PennyLane software. PennyLane can work with several different quantum hardware platforms and machine learning libraries. Not just their own. Zach Vernon presented the Xanadu hardware technology based upon Gaussian Boson Sampling (GBS) which utilizes a continuously variable unit called a qumode instead of a qubit. Zach mentioned that Xanadu currently has a 12 qumode system running in their lab with a goal of having a system with 50+ qumodes available by the end of 2020.
Horizon Quantum Computing
Joe Fitzsimons of Horizon Quantum Computing presented a new and unique software approach and working to provide a means of compiling classical code for quantum processors. The company is creating a new programming language called Carbon which is intended to be a subset of the classical Octave/Matlab programming languages for numerical programming. The intent is to provide a capability for classical numerical programmers to convert their programs into something which can run on a quantum computer without requiring the programmer to understand the intricacies of how the quantum computers work. Their software will transform an application program through four levels of software abstraction until it reaches the lowest level which would then be able to work on a gate level machine. Horizon’s software is still in development but should present some very interesting possibilities when it is released.
December 14, 2019