Microsoft announced three announcements specifically aimed at advancing computational chemistry and materials science discovery using their Azure, HPC, AI, and quantum technology services. The goal of this effort, as Microsoft CEO Satya Nadella put it, is to condense 250 years of chemistry into the next 25 years.
The first announcement is for a product called Microsoft Azure Quantum Elements which uses HPC and AI to accelerate certain chemical simulations by factors as high as 500,000. A key component of the software is using the GPT-4 model which has been enhanced by providing it with additional data in the fields of quantum computing and chemistry. A common problem in drug and substance discovery is screening compounds from a potential library of millions of compounds to find potential candidates for further testing. Quantum Elements can simulate 1.5 million potential molecular configurations in complex reactions that can have as many as 50,000 elementary steps to find the candidate. Quantum Elements also supports a number of popular computational chemistry tools including Aiida, Gromacs, NAMD, Scine, Lampmps, and NWChem. The company indicated that it already had early adopters including BASF, AkzoNobel, AspenTech, Johnson Matthey, SCGC, and 1910 Genetics successfully working with Quantum Elements and adopting the tool for use as part of their R&D. Currently, Quantum Elements does not use a real quantum computer for this production work. But Microsoft predicts that when higher-performance quantum computers become available, they will allow increasing the accuracy of predictive chemistry by a factor of 100-fold. Microsoft opened a private preview for Quantum Elements which is scheduled to start on June 30th. Interested users can register to participate in this preview by registering their interest Here. A webpage giving an overview of the Quantum Elements is available Herevideos about Quantum Elements can be seen Hereand a blog post providing additional details can be found Here.
Microsoft’s second announcement is called Copilot on Azure Quantum. This capability is an AI-based assistant that can communicate with users in natural language to reason through complex chemistry and materials science problems. It is a productivity tool that can provide tutorials on quantum and chemistry topics and even write code from a given problem statement. One of the biggest challenges in quantum computing is enabling subject matter experts, in this case chemists, to make the best use of quantum computing without requiring them to go through extensive training in quantum physics or quantum programming. The Co-pilot is a new tool that will help them do just that and can increase productivity significantly. Copilot is currently available for free and includes a built-in code editor, quantum simulator, unlimited code compilation, and data query and visualization capabilities along with the aforementioned capabilities. A user can register to use Copilot together with Azure Quantum on a web page Here.
Microsoft’s latest announcement is a conceptual roadmap for a quantum supercomputer. They have defined six stages of progress as shown in the diagram below:
Microsoft believes that the widespread use of quantum computing for commercial and scientific benefit will require a large number of error-corrected qubits. To measure progress towards this goal, they have devised another metric they call rQOPS (Reliable Quantum Operations Per Second). Their original goal was to build a quantum computer that could reach 1 million rQOPS with a logical error rate of 10 qubits.-12 and then build a machine beyond that which can support 1 billion rQOPS with an error rate of 10-18. They had reached the first milestone shown in the chart above, “Create and control Majoranas”, last year and posted a preprinted paper on arXiv reported their results. Now, the paper has gone through peer review and has just been published on APS physical review B publication under the title The InAs-Al hybrid device bypasses the topology gap protocol. Microsoft is currently working on their second foray to demonstrate hardware protected qubits. While Microsoft didn’t provide specific forecasts for when they would reach each of these milestones, they did point out that it will take years, but not decades, to work on these developments. Microsoft has published a blog explaining this roadmap that can be found Here. They also posted a helpful video visualizing how the Majorana particle and topological gaps work and it can be seen Here.
Finally, Microsoft has posted a web page containing a video presentation from their executives regarding the announcement along with links and other accessible resources. Here.
June 21, 2023