Science

IBM’s plans for the future of quantum computing

Today’s classic supercomputers can do a lot. However, those calculations are limited to the binary state of 0 or 1, so you can struggle with some very complex problems such as: Natural science simulation.. This is where quantum computers can be advantageous because they can represent information as 0, 1, or even both at the same time.

IBM debuted last year 127 qubit computing chip The structure, also called the IBM Quantum System Two, is intended to house components such as chandelier cryostats, wiring, and electronics for these larger chips. These developments have defeated IBM ahead of other major technology companies like Google and Microsoft in the competition to build the most powerful quantum computers. today, society Has a three-year plan to exceed 4,000 qubits by 2025 using a processor called the Kookaburra. Here are some plans to get there:

To scale up the processing power of qubits, IBM embodies the development of both hardware and software components of quantum chips. First introduced is a new processor called Heron, which boasts 133 qubits. In addition to having more qubits, the Heron chip has a different design than its predecessor, the Eagle. “In fact, we can get most of the two qubit gates that are working. We use a new architecture called tunable couplers,” said Jerry Chow, director of quantum hardware system development at IBM Quantum. Mr. says.

“In addition to this plan for this new processor for Heron, we want to be able to use multiple Herons, all addressable, all through one control architecture,” he adds. “When building these chips and processors, we want to be able to link traditional communications between these chips and processors.”

Better gate level control

Before we can understand what a qubit is, we also need to understand what a bit is and what a gate is. On traditional computers, information is encoded as binary bits (0 or 1). A transistor is a switch that controls the flow of electrons. Transistors are connected to several electrodes, including gate electrodes. Changing the charge on the gate electrode controls whether the transistor is on in state 1 or off in state 0. By physically changing these states, you can encode the information on your computer. Logic gates consist of a specific arrangement of transistors. A bundle of transistors can form an integrated circuit that can store chunks of data. All of these circuits are interconnected on the surface of the chip.

[Related: The trick to a more powerful computer chip? Going vertical.]

Qubits behave differently than bits, Quantum gate behavior is different From the classic gate. Unlike traditional bits, which can have a value of 1 or 0 under the right conditions Cubit You can stay in a wave-like quantum superposition state. It represents a combination of all possible configurations of 0, 1, or 2 superpositions. When a microwave photon is emitted at a frequency peculiar to a qubit, Researchers who control their behaviorCan hold, modify, or read units of quantum information.

Unfortunately, qubits are extremely fragile, heat sensitive, unstable, and error prone. When qubits communicate with each other or with wiring in the environment, the characteristics of the qubits can be lost and the accuracy of the calculation can be reduced. Experts refer to “coherence time” when describing how long they can stay in superposition. The coherence time and the time it takes to perform the gate set a limit on the amount of quantum computation that can be performed on a set of qubits.

[Related: IBM’s latest quantum chip breaks the elusive 100-qubit barrier]

“The method of designing today’s processors, Falcon, Hummingbird, and Eagle, uses fixed coupling between qubits and a microwave-based two-cubit mutual resonance gate,” Chow said. increase. In those cases, they used different frequencies to talk to the corresponding qubits. Currently, they are adding “individual magnetic field control of couplers between qubits”. This allows you to turn on qubit interactions with different microwave frequencies.

Multiple connected quantum processors

Traditional computers have a core, Grouping of transistors You can run multiple tasks in parallel.You can do it Imagine it Try to open multiple checkout registers in the supermarket instead of having everyone line up in one.CPU that provides multiple cores, or MultithreadYou can divide a large task into smaller pieces and send them to different cores for processing.

IBM now wants to apply this concept to quantum computing as well. Circuit knitting.. It “finds a way to effectively take large quantum circuits and break them down into smaller, more digestible quantum circuits, which can be run almost in parallel across many processors,” Chow said. I will explain. “This classic parallelization increases the types of problems and features that can be addressed.” Parallelization also helps reduce error rates.

Derivatives of this design are separate from the development of Osprey or Condor, which will reach 433 and 1,121 qubits, respectively, over the next few years. “But we also need to have built-in modularity that can be further expanded. At some level, only the number of qubits that can be packed into one chip begins to be limited,” says Chow. “We are currently testing some of these boundaries in Osprey and Condor.”

Heron’s idea is to test how engineers establish quantum links between multiple quantum chips. “We are investigating what we call these modular couplers, which allows us to effectively connect multiple chips,” says Chow. This creates an essentially large quantum coherent processor consisting of three separate quantum chips with the same underlying quantum processor. To this end, IBM hopes to combine the three chips into a 408 qubit system called Crossbill in 2024.

To further expand, IBM is also working on long-range couplers that can connect clusters of quantum processors via 1-meter-long cryogenic cables (superconducting qubits need to be kept very cold. I have). “We call this a quantum communication link,” says Chow, who can extend quantum coherent connections in a shared cryogenic environment.

The combination of parallelization, chip-to-chip connectivity, and long-distance coupling has the potential to achieve the 2025 goal of the 4,158 qubit system, the Kookaburra.

A combination of classical computing and quantum computing

Quantizing does not mean redesigning the entire computer from scratch. Many quantum systems run on traditional computing infrastructure. “The way we usually have our system is that you have your quantum processor in the fridge and you are always talking to the classic infrastructure with it,” Chow. Says. “Traditional infrastructure produces these microwave pulses and produces reads. When you program the circuit, it turns into an orchestration of this gate, the operation sent to the chip.”

However, one controller can feed not only quantum processors, but also traditional processors such as CPUs and GPUs. These processors are connected in parallel with the quantum chip, but not in a quantum way. In this way, you can run threaded applications that take advantage of both classical and quantum computing power.

“Quantum processors offer different resources than GPUs and very large CPUs,” says Chow. “But overall, the whole thing will feel like a supercomputer that is still being tuned together.”

[Related: Recent AWS glitches illustrate the power, and fragility, of cloud computing]

IBM’s vision for the future of computation is that machines include components that can run quantum circuits on quantum hardware. However, this component integrates with traditional memory and traditional infrastructure. This type of hybrid structure can be used for problems like molecular simulation. In molecular simulation, Variational quantum eigenvalue solver..

Quantum software

Quantum circuits are different from classical circuits. Gate logic Is different, the language is algorithm Varies.

For IBM First quantum computer Released in the cloud in 2016, it comes with assembly language. OpenQASM, Used to build programs. Next year, IBM will sayDynamic circuitCan measure qubits and process classical information into the OpenQASM3 library at the same time. It is also a hardware improvement that relies on improved control electronics and better real-time messaging between the control and measurement sides of the circuit. This allows for more error correction and parity checking.

The basic language coding for these types of operations is formed Primitive, Or the basic calculation element of the algorithm. All of these will be part of IBM. Qiskit runtime A programming model for platforms, computing services, and quantum computing. Qiskit includes various levels of assembly language for kernel developers who need to work with code and hardware, as well as APIs in the Qiskit stack for algorithm developers to work with. Serverless..

“At this higher level for algorithm developers, with this cloud environment with co-ordinated access to the CPU, GPU, and QPU, you don’t have to worry about running on a particular backend.” Says Chow. “This allows us to use classical resources in combination with quantum resources to handle some of the problems of larger quantum circuits. This may be driving the benefits of quantum, etc. . “



https://www.popsci.com/technology/ibm-quantum-computing-roadmap/ IBM’s plans for the future of quantum computing

Back to top button