The new solution offers support for running hybrid quantum-classical workloads on Amazon Braket.
This week Amazon Web Services(AWS)announced the launch of Amazon Braket Hybrid Jobs. This is a new capability of Amazon Braket that simplifies the process of setting up, monitoring, and efficiently executing hybrid quantum-classical algorithms.
AWS first announced this solution at re:Invent 2021. Since then, many quantum computing researchers have shifted their focus to using so-called hybrid algorithms that use classical computers to optimize quantum algorithms. This is a process similar to training machine learning models.
But setting up and running those hybrid quantum-classical algorithms can be a monumental task. With the release of Amazon Braket Hybrid Jobs, Amazon is trying to make that task easier.
Helping users “innovate more quickly”
Danilo Poccia, Chief Evangelist (EMEA) at AWS, detailed the new release in a blog post this week. With Amazon Braket Hybrid Jobs, users can avoid extensive infrastructure and software management and confidently execute their algorithms quickly and predictably. It also provides on-demand priority access to QPUs (quantum processor units), Poccia says.
“When you create a job, Amazon Braket spins up the job instance…executes the algorithm (using quantum hardware or simulators), and releases the resources once the job is completed,” he explains. This is importsant becauswe then the customer only has to pay for what they use.
Users can also define custom metrics for algorithms. Amazon CloudWatch automatically logs them and displays them in near real-time in the Amazon Braket console as the algorithm runs. “This,” Poccia says, “provides you with live insights into how your algorithm is progressing.” It also creates the opportunity for the user to adjust the algorithm as necessary and innovate more quickly.
“Instead of relying on theoretical studies, you can start to use quantum computers as the primary tool to understand and improve hybrid algorithms,” Poccia explains. “In this way, you can focus on your research and not deal with setting up and coordinating these different compute resources for your experiments.”