Thanks to a combination of reworked algorithms, improved control systems, and a new quantum service called Qiskit Runtime, IBM researchers were able to solve a quantum problem 120 times faster than in their previous attempts.
In 2017, Big Blue announced that its researchers had successfully simulated the behavior of a small molecule called lithium hydride (LiH) using a seven-qubit quantum processor. At the time, the operation had lasted 45 days. Four years later, the IBM Quantum team announced that the same problem had just been solved in just nine hours.
The simulation was run entirely in the cloud, through IBM’s Qiskit platform – an open-source tool library that allows developers around the world to create quantum programs and run them on prototypes of quantum devices that IBM makes available in the cloud. The observed acceleration was largely made possible by a new quantum service, Qiskit Runtime, which played a key role in reducing latencies during simulation.
IBM introduced Qiskit Runtime earlier this year as part of its strategic plan for the company’s software for quantum computing. The American giant estimated at the time that the new service would accelerate workloads by 100 times. With 120x acceleration, it looks like Big Blue has surpassed its own goals.
Classical computing remains a fundamental element of Qiskit, and of any quantum operation performed in the Cloud. A quantum program can effectively be broken down into two parts: Using conventional hardware, such as a laptop, developers send queries through the cloud to quantum hardware – in this case, IBM’s quantum computing center in the States. -United.
“The quantum method is not just a quantum circuit that you run,” says Blake Johnson, manager of the quantum platform at IBM Quantum interviewed by . “There is an interaction between a classic computing resource that makes queries to quantum hardware and then interprets those results to make new queries. This conversation is not a one-time thing – it happens over and over again, and you need to let her be quick. ”
For each request sent, a few tens of thousands of quantum circuits are executed. To simulate the small molecule LiH, for example, 4.1 billion circuits have been executed, which corresponds to millions of requests going back and forth between the classical resource and the quantum resource. When that conversation takes place in the cloud, over an internet connection, between a user’s laptop and IBM’s US-based quantum processors, latency can quickly become a big hurdle.
Case in point: if solving a problem as complex as molecular simulation in 45 days is a start, it is not enough to achieve the quantum advances that scientists are excited about. “We currently have a system whose architecture is not intrinsically linked to the fact that real workloads have these quantum and classical loops,” explains the IBM researcher.
Based on this observation, IBM’s quantum team set out to build Qiskit Runtime – a system designed to natively speed up the execution of a quantum program by removing some of the friction associated with going back and forth between the world. quantum and the classical world.
Qiskit Runtime creates a containerized runtime environment located next to quantum hardware. Rather than sending many requests from their device to the cloud-based quantum computer, developers can therefore send entire programs to the Runtime environment, where the IBM hybrid cloud downloads and performs the work for them.
In other words, the loops that occur between the classical environment and the quantum environment are contained in Runtime, which itself is close to the quantum processor. This helps reduce latency times associated with communication between the user’s computer and the quantum processor.
“The classic part, which generates queries for quantum hardware, can now be performed in a container platform that sits in the same location as quantum hardware,” says Johnson. “The program running there can ask a question of the quantum hardware and get an answer very quickly. This is a very low cost interaction, so these loops are suddenly much faster.”
The many faces of quantum computing
Improving the precision and scale of quantum calculations is no easy task. So far, says Blake Johnson, most of the research effort has focused on improving the quality of the quantum circuit. In practice, it was about developing software to correct errors and add fault tolerance to quantum hardware. In this sense, Qiskit Runtime marks a change in mentality: instead of working on the quality of quantum hardware, the system increases the overall capacity of the program, says the researcher.
Still, the 120-fold acceleration wouldn’t have been possible without additional hardware performance tweaks. Algorithmic improvements, for example, have made it possible to reduce from two to ten times the number of iterations of the model necessary to obtain a final answer; while improved processor performance means that each iteration of the algorithm requires fewer passes through the circuit.
At the same time, system software upgrades and control systems have reduced the execution time of each circuit for each iteration. “Quality is a critical ingredient that also keeps the whole system running faster,” says Johnson. “It is the harmonious improvement in quality and the ability to work together that makes the system faster.”
A new classification task
Now that the acceleration has been demonstrated in the simulation of the LiH molecule, Blake Johnson hopes developers will use this improved technology to experiment with quantum applications in a variety of fields other than chemistry. In another demonstration, for example, the quantum team at IBM used Qiskit Runtime to run a machine learning program for a classification task.
The new system was able to run through the workload and find the optimal pattern to label a data set within a timeframe that Johnson called “significant.” Qiskit Runtime will initially be released in beta, for a limited number of users on the IBM Q network, and will ship with a fixed set of configurable programs. IBM expects the system to be available to all users of the company’s quantum services in the third quarter of 2021.
In combination with the 127-qubit quantum processor, called the IBM Quantum Eagle, which is slated for release later this year, Big Blue is hoping that the acceleration allowed by Runtime will mean that many tasks that were previously considered impractical on quantum computers will be now achievable. The system certainly puts IBM on track to meet the goals set out in the company’s quantum software roadmap, which predicts frictionless quantum computing will be used in a number of applications by 2025. .