IBM 433 QPU low profile! Explaining the past life of superconducting quantum
Just now, the much-anticipated 433-quantum-bit Osprey chip was launched on IBM Quantum Cloud in a low-profile manner.
Back on November 9, 2022 at IBM's annual Quantum Summit, IBM unveiled the Osprey chip, which boasts about three times more than Eagle (127 quantum bits) at 433 quantum bits. At this point, the 433-quantum-bit Osprey became the largest quantum processor available to IBM for customer use.
https://quantum-computing.ibm.com/services/resources?tab=systems
At present, IBM's official website has not yet released the relevant update notice. Still, IBM says, "Our shared mission is to integrate Qiskit Runtime into our partners' applications and services to extract value from quantum circuits."
Superconducting Quantum Bits is currently the leading technology in the commercial field of quantum computing and is being used by IBM, Google, Rigetti, Amazon, Alibaba and many startups such as IQM (Finland), OQC (UK), Anyon Systems (Canada), Alice&Bob (France), Nord Quan- tique (Canada), Origin, Quantum Spin and others chosen by.
A historical timeline of superconducting quantum bits. Yale University's scientific contributions seem to dominate here, hence the nickname "Yale gang".
The history of superconducting quantum bits begins in the mid-1980s and can also be traced back to the BCS theory of 1957, which (partially) explained the behavior of pairs of electrons with opposite spins (i.e., Cooper pairs) at low temperatures: they produced superconducting effects. Then, in 1962, Brian Josephson discovered the Josephson effect, and its experimental proof was completed by John M. Rowell at Bell Labs in 1963.
In 1980, Antony Leggett developed a collective-degree-of-freedom model for superconducting circuits: it is somewhat like a Bose-Einstein condensate of cold neutral atoms, with the Cooper pairs of electrons in the superconducting material behaving like a single quantum object with its own quantum waves. antony Leggett stated that the quantum behavior of Josephson junctions could be inferred from their classical behavior.
In 1985, John Clarke, Michel Devoret (his postdoc) and John Martinis (his PhD student) demonstrated macroscopic quantum tunneling of a current-biased Josephson junction in the zero-voltage state. Soon after, they demonstrated the quantum level of the phase - the first man-made electric atom.
At that time, the JJ (Josephson junction) was realized with Nb-NbOx-PbIn (niobium, lead, indium) and cooled with a He-4-based cryostat.
Early Josephson junction arrays
In 1998, Vincent Bouchiat, then a PhD student in the quantum science group of Michel Devoret, Daniel Esteve and Cristian Urbina at CEA-Saclay, France, realized the first Cooper pair box (CPB) and described its ground state. Briefly, a Cooper pair box is a JJ connected to a voltage box by a capacitor on one side: a capacitor on one side and a Josephson junction on the other.
In 1999, Yasunobu Nakamura, together with Yuri Pashkin and Jaw-Shen Tsai, demonstrated for the first time the phenomenon of quantum coherent superposition with the first excited state at the NEC laboratory in Tsukuba. This was the first "charge quantum bit" with a very short coherence time of 2 nanoseconds. In 2001, they extended it with the first Rabi oscillation measurement related to the transition between two Josephson levels in a Cooper-pair box. 2002, the CEA-Saclay Quantronics team demonstrated the first functional quantum bit version of the Cooper-pair box, the quantronium; later, it took the CEA team about 12 years to reach four quantum bits.
A modern version of the CPB circuit, the vaunted transmon, was developed at Yale University in 2006. The Yale research team, led by Rob Schoelkopf, Michel Devoret and Steve Girvin, welcomed a number of talented theorists and experimentalists who made key contributions to the progress of transmon style quantum bits.
Jerry Chow was also a major contributor between 2005 and 2010, and has since been with IBM, now leading their quantum hardware system development in Jay Gambetta's group. In 2009, Devoret, Schoelkopf, Leonardo Di Carlo (now at Delft University of Technology), Jerry Chow and others created the first programmable two-quantum-bit processor on which a small Grover search was implemented. in 2012, A. Dewes et al. at CEA Saclay demonstrated the first functional two-qubit processor fully equipped with a set of universal gates and a single single quantum bit readout using Grover's search algorithm. blake Robert Johnson proposed the use of Purcell filters to protect quantum bits from spontaneous emission in 2011.
Other noteworthy contributions are Hans Mooij (Delft University of Technology): He created a flux quantum bit with three Josephson junctions in 1999 and performed experiments in 2000. andrew Hook (Princeton University) contributed to the development of transmon bits. in 2010, Andrew Cleland, John Martinis and their PhD Arron O'Connell were able to entangle three flux superconducting quantum bits together, which led to the Xmon tunable quantum bit in 2013, later used by Martinis at Google after 2014; Andrew Cleland now runs his own lab at the University of Chicago.
In 2017, Peter Leek, then at Oxford University, created the coaxial superconducting quantum bit, in which the quantum bit and resonator are located on opposite sides of a single chip, with control and readout lines provided by coaxial lines perpendicular to the plane of the chip; this led to OQC in the same year. in 2022, Mikko Möttönen of IQM created the Unimon superconducting quantum bit , with a simpler setup, stronger nonlinearity and better delay. However, neither of these quantum bits has yet demonstrated advantages over the current state of the art.
Then, in 2013, Mazyar Mirrahimi and Michel Devoret of Yale University created the "cat quantum bit", and many more advances have been made since then.
Different types of superconducting quantum bits encode quantum information in two different states in different ways.
Superconducting quantum bits use an anharmonic oscillator to distinguish between two energy levels corresponding to the ground state and the excited state of the quantum bit.
Superconducting quantum bits are the best scalable structure in the current gate-based model, with 433 quantum bits in IBM, 176 in China, and 361 in Europe as of May 2023, according to official reports - although so far, these quantum bits are still not of sufficient quality to make them be useful in practice.
Specifically, the Josephson junction used in superconducting quantum bits is a thin, nanoscale insulating barrier between two superconducting metals that forms a tunneling junction. It creates a quantum electrical component with a single degree of freedom - the superconducting phase difference between its electrodes, linked to the number of Cooper pairs passing through the junction; the supercurrent through the "junction" (the direct current Josephson effect) is driven by the phase difference. From an electrical point of view, the Josephson junction behaves as a non-dissipative and non-linear inductor whose value depends on the phase and thus on the current. The special feature of superconducting quantum bits is that they are the only prevailing macroscopic quantum bits, i.e., they are independent of the control of individual particles (e.g., individual atoms, electrons or photons, as in most other quantum bit technologies).
At superconducting temperatures well below the superconducting critical temperature, Josephson junctions embedded in the circuit behave like an artificial atom with a quantum level of gate and/or flux control and an electron Cooper pair of about 10^11 electrons (100 billion). They form an artificial atom with precisely controllable energy levels with specific parameters including a Josephson barrier, some series and/or parallel capacitance and inductance, and some readout circuits using nearby resonators.
Superconducting quantum bits use non-dissipative elements: capacitors, inductors and Josephson junctions as nonlinear non-dissipative inductors. Capacitors store energy in the electric field, while inductors store energy in the magnetic grid. However, at any non-zero frequency, superconductors still dissipate some energy through two channels: transport by Cooper pairs and normal charge carriers (quasiparticles), which decreases exponentially in proportion to the quasiparticle density and at low temperatures.
A typical configuration of a superconducting quantum bit lab. Laboratories working on actual QPU now use more dedicated and integrated quantum bit control technologies from industry suppliers (Quantum Machines, Zurich Instruments, Keysight, Qblox, etc.).
The current generation of superconducting gated quantum computers belongs to the pre-NISQ and NISQ classes - NISQ refers to noisy mesoscale quantum computers. It describes quantum computers with more than 50 physical quantum bits; they can bring several advantages over the best classical computers: computational speed, quality of results and some energy advantages. Pre-NISQ systems have less than 50 quantum bits and are usually below the quantum advantage threshold.
NISQ and FTQC may overlap with the competition between NISQ and quantum error mitigation, expanding the capacity of noisy quantum bits, and the modified quantum bits will make FTQC and larger depth algorithms possible.
So far, however, no computer is in the available NISQ region. Some vendors, such as IBM, are planning to release new QPUs (quantum processing units) with fidelity in the range of 99.9%, sufficient to run NISQ quantum algorithms with some quantum advantage. However, in terms of algorithmic depth capabilities, we may have an overlap between the NISQ and FTQC eras, where there is competition between very high latency quantum bits, quantum error correction, and a large number of high latency quantum bits, quantum error correction.
However, the current period is fruitful in terms of using existing NISQ hardware. It aims at learning to program these systems, designing new algorithms and creating various error mitigation techniques. It also provides a feedback loop for quantum computer developers, especially around the various software cloud tools that manage these QPUs.
Outside of or parallel to the NISQ system development is the Fault Tolerant Quantum Computer (FTQC). These QPUs will rely on sets of logical quantum bits and will have lower error rates due to the use of quantum error correction codes and redundancy. Logical quantum bits will have a lower error rate depending on the error correction code, the density of physical quantum bits, their connectivity and their number. It is expected that superconducting-based logical quantum bits will require from a thousand to several million physical quantum bits, depending, of course, on the target application.
Like all quantum bit types, superconducting quantum bits have their challenges in creating useful quantum computers, both in the NISQ and FTQC domains. In November 2022, IBM achieved best-in-class latency with its "Egret" 33 quantum bit processor, demonstrating a two-qubit gate latency of 99.7%. Creating a fault-tolerant quantum computer would require at least 100,000 physical quantum bits with a 99.9% latency: this would allow 100 logical quantum bits and an equivalent computational depth.
These requirements pose a huge challenge:
- Is it possible to control the crosstalk of quantum bits at this scale?
- Is it possible to maximally entangle such a large multi-body quantum system in a controlled way?
- How can quantum error correction codes be designed with minimal physical quantum bit overhead requirements?
- Is it possible to create enough low-power control electronics, cables, multiplexing, and cryogenics to achieve such a scale?
- Will the associated energy consumption be controlled, and can this be addressed in the systematic manner suggested by the Quantum Energy Initiative?
- Is it possible to interconnect several quantum processors using microwaves or entropy? - Is it possible to use microwave or entangled photon resources?
- How can the software tools used to design these quantum bit chipsets be improved?
All these scientific and technical challenges are enormous.
The double quantum bit gate density of superconducting quantum bit computers currently offered by commercial vendors. The blue area corresponds to the area where QPUs can bring computational advantages in the NISQ or FTQC systems. FTQC systems require at least 99.9% latency and can scale to millions of quantum bits, while NISQ systems are based on a few hundred or a few thousand quantum bits.
Now, vendors like IBM, Rigetti and Google are trying to create NISQ systems with hundreds of quantum bits, which could bring some quantum computing advantages: thanks to quantum error mitigation techniques that work with shallow depth algorithms, and especially variational algorithms that work in mixed mode with supercomputers. Creating usable NISQ systems still requires higher gate densities than today, for example, above 99.99% for double quantum bits.
Despite the challenges ahead, it is certain that superconducting quantum technology will bring more surprises to the market in the coming years. We'll see what happens in the years to come.
Reference links:
[1]https://quantum-computing.ibm.com/services/resources?tab=systems
[2]https://research.ibm.com/blog/qiskit-runtime-capabilities-integration
[3]https://link.springer.com/article/10.1140/epja/s10050-023-01006-7
