How can Quantum Elements accelerate scientific discovery today and in the future?

How can Quantum Elements accelerate scientific discovery today and in the future?

Show Video

Hello, I’m doctor Nathan Baker, head of partnerships for chemistry and materials in Azure Quantum at Microsoft. Welcome to season two, episode two of our Quantum Innovator series. In this episode, I’ll discuss how your organization can transform innovation, adapting to today’s evolving pressures with agility while pioneering the products of tomorrow that will help you stay ahead.

I’ll also be sharing how Azure Quantum Elements can accelerate scientific discovery today for chemistry and material science. Using high performance computing and artificial intelligence, as well as future quantum supercomputing. If you have any questions, please drop them in the chat. Members of the Azure Quantum team are standing by to answer them. We will also talk with Dominik Wee, corporate Vice President of Manufacturing and Mobility at Microsoft. We’ll get his perspective on the R&D transformation the manufacturing and chemical industries are undergoing to drive innovation.

If you’ve seen other episodes of our Quantum Innovator series, you know that quantum computing will enable applications in chemistry and material science to solve problems beyond the reach of classical computing. Our society needs innovation in chemistry and material science, now more than ever. Chemistry and material science can help solve important problems facing our planet like food scarcity, pollution, and climate change. With 96% of all manufactured goods relying on chemicals and chemical manufacturers, this is also an enormous business opportunity. At Microsoft, we believe that the future of these businesses requires faster discovery of new molecules and materials that can meet the needs of our society and help us tackle sustainability challenges and beyond.

But many chemistry and material science problems are based on quantum mechanics, making them hard problems to solve for today’s classical computing systems without approximations that can degrade accuracy. To put the challenges of quantum mechanics in practical terms, each additional electron in the system increases the computational time required to solve. Therefore, researchers have to balance compute time and accuracy when they model these systems. Azure quantum has a set of tools to help researchers understand when approximations are being used, and whether or not they’re necessary. Because the language of nature is quantum, quantum computers will be uniquely suited to solving these types of problems in the future. We’ve built Azure Quantum Elements as an on ramp for future chemistry problems solved with future quantum supercomputers.

This includes tools in Azure Quantum Elements to model how quantum computers will remove these approximations. We’ve also included hooks into the existing quantum computers available in Azure to help researchers learn how to use these computational resources in solving complex chemistry and material science problems. This is an example of how both classical and quantum computing will help drive innovation. There’s also an urgency to accelerate innovation today, with businesses under pressure to create sustainable chemicals to comply with shifting regulations and consumer demands for products that help create a safer future.

But sustainability is not only a matter of following regulations, it’s also a source for business growth driven by positive customer perception. When we think about our world economy and the values both businesses and consumers are adopting to create that better future we see business growth is now closely tied to everyone doing their part to build a better future. One way we see that happening is through R&D innovation.

Whether it’s catalysts that can transform carbon dioxide into clean fuel, or compounds that can boost the power of batteries, the discovery and development of new molecules for our toughest challenges is the way forward for business growth. That’s why at Microsoft, our goal is to empower enterprises with the right tools so they can accelerate their discovery pipeline and drive this new growth. As our CEO says, this will happen by accelerating the next 250 years of chemistry into the next 25.

And it’s the mission that we’re here to talk about today. While we continue our progress toward a future scale, the quantum supercomputer, we are able today to innovate faster using progress and high performance computing and artificial intelligence to meet this innovation demand. This progress creates opportunities in quantum chemistry to unlock new growth across almost every industry, enabled by discovering and pioneering the new molecules and materials of the future. This acceleration must happen through a paradigm shift in research and development. We need a digital transformation where computational tools help lead the way, pointing experiments to the most promising leads and maximizing the time and effort spent in the wet lab.

However, there have been challenges in the past that have held back the promise of computational chemistry and limited this digital transformation. And that’s what’s behind the private preview of Azure Quantum Elements, a comprehensive system specifically tailored for chemistry and materials science research and development. Azure Quantum Elements addresses two types of customer challenges in that process: Technical challenges and productivity challenges. On the technical side, the challenges are scale, speed, and accuracy, with elements designed to address these through high performance computing, artificial intelligence, and future quantum computing. On the productivity side, the challenges that most computational chemistry and material science tools are hard to use. The environments that support them can also be very difficult to set up and maintain.

So what is the Azure Quantum Elements platform? Elements uses natural language AI assistance for data and computation. It scales these computations through Azure High Performance Computing, and it accelerates the underlying calculations with artificial intelligence. The hybrid approach of classical computing combined with quantum computing helps in three areas, which I’ll detail more in the upcoming slides. High performance computing helps us scale calculations, while artificial intelligence helps accelerate the underlying workloads. At the same time, future quantum supercomputing will help us address some of the accuracy challenges associated with the quantum mechanical foundations of these problems.

But first, I want to talk about the productivity challenge. As I mentioned earlier, some computational chemistry tools can be hard to use, and besides that, the environments that support them are sometimes challenging to set up and maintain. The goal of Azure quantum elements is to reduce that productivity barrier.

So let’s look at a concrete example of how we keep the needs of researchers front and center on the elements platform. Complexity and using computational tools in chemistry and materials science can be a challenge. Microsoft has invested in artificial intelligence, developing Copilots to help users in their various problem domains.

Our goal with the Azure Copilot was to create a natural language driven environment for asking questions and getting source to answers about problems in chemistry and material science. In addition to information retrieval and helping you understand specific concepts in chemistry and material science, Copilot can help your team visualize molecular structures and their associated molecular orbital. This enables researchers to use conversational interfaces to retrieve information, understand molecular behavior, and generate code to fuel productivity. Another barrier to productivity in chemistry and materials science is configuring and executing the underlying simulations. Copilot helps with this task by generating code that is intended as a starting point. The code is not run automatically, but it can still save scientists tremendous amounts of time from searching on Stack Overflow or perusing through a particular software packages user manual.

This means you can get technical guided answers with sample code through Copilot. In a simple interface, you can query and visualize data, and you can initiate simulations, all of which can reduce productivity barriers and make it easier to configure and use workflow tools. As I mentioned earlier, in addition to the productivity challenges, there are three technical challenges limiting the promise of computational chemistry. These are scale, speed, and accuracy. Let’s start with the scale challenge.

Scale is important in all aspects of computational chemistry. It is essential to scale out the molecular or materials libraries that are screened, and discovery applications. And scale is a core component of being able to simulate large biomolecular assemblies or large material systems. A core component of scale are workflow tools.

These tools are essential to manage the complexity of running many different kinds of calculations, and an end-to-end discovery pipeline, and scaling it across HPC resources to get the best performance for all the calculations in the discovery process. The foundation of Azure Quantum Elements is Azure HPC, which provides fast and secure networking, high performance storage, and top-of-the-line GPUs and CPUs. These are combined with automation and chemistry workflow software that enables scaling up workflows to support large scale calculations. One example application that illustrates the challenge of scale is exploring the steps of a complicated reaction network.

Understanding reaction steps is essential for designing new catalysts or optimizing reaction conditions to produce a desired product or reduce unwanted side products. In collaboration with ETH Zurich and Pacific Northwest National Lab. Microsoft developed a reaction exploration tool that harnesses the power of Azure high performance computing to scale out over the large number of steps involved in many reaction processes. For example, researchers have used this tool to explore over 1.5 million potential steps involved in the urethane production catalysis process.

This reaction network tool was scaled out across Azure and used a variety of computational techniques tuned to different virtual machine architectures. Semi empirical approaches were used to screen initial reaction step candidates, followed by refining those candidates using density functional theory and finally using high level electronic structure methods to get the detailed energetics associated with each step. Azure Quantum Elements enables scale screening of reaction steps from a cost effective and time saving standpoint, especially when compared to other methods.

Now let’s talk about speed. Speed is a central challenge in all aspects of computational chemistry and materials science. Microsoft has been addressing speed challenges through artificial intelligence, working closely with the Microsoft Research AI for science team. Azure Quantum Elements transform simulation and discovery workflows using a combination of Azure high performance computing and advanced artificial intelligence techniques. One example of advanced artificial intelligence techniques includes our large scale transformer models. These models have been trained on hundreds of millions of quantum chemistry data points, which gives them a foundation of information that can then be fine tuned on experimental data to predict molecular properties such as boiling point, solubility, or other characteristics with high accuracy and generalizability.

An additional artificial intelligence capability includes machine learned force fields, which can be used to simulate complex atomic and material systems. These models are based on graph neural networks, which means they’re smaller but can still be generalized to provide up front screening based on additional properties. Artificial intelligence and high performance computing are an inseparable combination for accelerating workflows in materials discovery. Azure Quantum elements artificial intelligence models have been used to significantly increase the number of candidates that can be included in molecular discovery workflows. By accelerating the start of the discovery process with artificial intelligence.

we can quickly screen the pipeline of candidates and winnow down the search to the most promising leads. Following the initial artificial intelligence property based screening, machine learning force fields can be used to run simulations to compute additional characteristics of the candidates to further refine the pipeline. These machine-learned force fields offer a powerful complement to Ab Initio molecular dynamics simulations, giving comparable accuracy while offering up to a half a million times speedup. Once the most promising candidates are identified using artificial intelligence approaches, it’s still important to use traditional physics based simulations.

These high performance computing calculations are important for generating insight and building confidence in the results. The combination of artificial intelligence and high performance computing allowed researchers to compress two decades of calculations into a week, transforming the way new materials are discovered. This is only possible thanks to the acceleration of these workloads with artificial intelligence. Finally, let’s talk about accuracy.

Accuracy is the final technical challenge that Azure Quantum Elements has been designed to address. We expect future quantum supercomputing capabilities to fundamentally transform the way we look at interactions at the electronic level. This transformation will happen through quantum computing, which will change the way we achieve accuracy in quantum mechanical calculations. Classical simulations of quantum mechanical processes often require approximations due to the complexity of running classical computation on quantum mechanical problems. Quantum computing promises to eliminate these approximations and change the way we develop ground truth for quantum mechanical descriptions of chemical and material systems. Azure Quantum Elements has tools today to address the approximations associated with modeling chemical systems.

These tools allow us to study the chemical systems and identify which parts need quantum computing to remove approximations and improve accuracy. For example, elements includes active space methods, which can be used to identify the set of electrons which exhibit entangled or correlated behavior. Not all electrons in every molecule have these properties. However, the existence of correlated electrons often complicates their classical computation. For systems with less than 20 orbitals in the active space, a full configuration interaction calculation can be run using classical Azure HPC resources to obtain accurate results without making approximations.

For larger active space systems, a resource estimator is built into azure quantum elements. This estimator makes it possible to estimate the quantum computing resources needed to execute the underlying quantum circuit that describes the configuration interaction problem. Noisy intermediate scale quantum computers are also offered through Azure Quantum. These resources can be used to study very small chemical problems and experiment with quantum algorithms today.

For small active space systems that contain only a few electrons, quantum curious customers can explore fully integrated quantum classical workflows today using simulators or third-party quantum hardware currently available in Azure Quantum. Azure quantum has a comprehensive stack for quantum computing today. This stack allows you to choose between multiple NISC endpoints, such as Pascal, IonQ, QCI, Rigetti, and Quantinuum, as well as multiple programing languages like Q#, Cirq, and Qiskit.

Overall, the goal of Azure Quantum Elements platform is to provide chemistry and material science research tools for addressing the scale, speed, and accuracy challenges today while building an on ramp to future quantum supercomputing. We’re excited by the potential to deliver high performance computing and artificial intelligence tools that help industry solve the hardest chemistry and materials science challenges today. Now let’s talk to Dominik Wee, Corporate Vice President, Manufacturing and Mobility at Microsoft, to learn what he’s been hearing from real-world customers. Hello, Dominic. Thanks for joining us today. Hey, Nathan. Glad to be here.

So, Dominic, as you’re talking with leaders in the chemical and materials science manufacturing space, what are some of the biggest challenges they’re facing? At its core, manufacturing is about innovation and bringing breakthrough products to market. We’re helping our customers as they try to figure out how to do that, to increase revenue and navigate a once in a century transformation in four areas. The first one is empowering a connected workforce by helping streamline frontline operations, enhance communication and collaboration, improve the employee experience, and strengthen security across devices. The second one is around building resilient and agile supply chains to prevent and minimize interruptions with real time predictive AI insights across materials, inventory and distribution networks.

And the third one is people in AI working together and intelligent factories to drive innovation and efficiency across the design, engineering, manufacturing, and operation lifecycle of their products. And then the fourth one is sustainability. By using renewable energy sources and adopting circular economy principles to reduce waste and optimize production processes, minimizing environmental environmental impact. Of those four areas, sustainability, which also includes creating better products in highly competitive markets, is a strategic business priority for many of them, not surprisingly. At the same time, they’re facing dynamic regulatory changes, increased customer expectations, and nimble startup challenges.

Overall, you can see industries like manufacturing are facing a range of challenges when it comes to innovation. In these times, business as usual isn’t likely to bring the innovation needed to drive top line growth. And that’s why by levering new technologies like AI and supercomputing, investing in R&D transformation and collaborating with partners, companies in this space are finding new ways to meet the imperative to innovate and stay ahead of the competition. Are there specific opportunities or challenges surrounding sustainability that you’re hearing from customers? Yes. It’s, two things. I think, the first one is how to find ways not to only maintain, but to increase growth in these unpredictable times. And the second one is that manufacturers can play a crucial role in tackling climate change by creating entirely new categories of products that can meet the needs of our society, and also help us meet the sustainability challenges of today and in the future.

At Microsoft, we are trying to help with both. We can help our customers build better products that promote and then clean responsibility and comply with regulations, while enabling their future growth. With cloud technologies such as Azure Quantum Elements.

These sustainability imperatives, underscored by financial implications, market dynamics, and the necessity to adapt to a rapidly changing world. We don’t see regulation as a checkbox, but a strategic opportunity for businesses to secure their success and pioneer solutions for a better future. There are clear opportunities for business growth.

Responsible business practices, cost savings, safety, and more for companies to transform their R&D through digitalization with innovative systems such as Azure Quantum Elements. That’s very interesting. So what sorts of impact are you seeing with customers who embrace digital transformation, and especially for their research and development processes? We are seeing several in my recent Relentless Renewal series.

I discussed the potentially disruptive impact of powerful innovations like artificial intelligence, which is connecting to supercomputing capabilities on the manufacturing industry. Specifically, I believe that what is most disruptive, impactful manufacturing industry is the impact it can have on the R&D process. And what you showed earlier. One exciting development in this area, to me is the concept of a copilot in the coding area. This technology has gotten our customers excited because it offers the opportunity to dramatically shorten the development cycle and potentially simplify the process of creating complex products.

And how do you see that happening? But firstly, they are increasing the likelihood of success in finding and designing new products. This is because they’ve experienced the training informed by the power of computer simulations, that helps to identify the best candidates with the ideal properties for the materials they’re looking for, and to apply them for semiconductors, cars, and you name it. And then secondly, they are also broadening their search space to find the next hit product. As you had mentioned earlier, instead of exploring thousands of candidate molecules via traditional methods, they’re able to exploit tens of millions with the support of digital platforms. And this is a game changer. And lastly, it’s about guiding the innovation.

You can’t really improve what you can’t fully understand. And our digital solutions are helping these R&D teams gain insights into areas previously on track to to explore. This is unlocking the ability to truly understand what they can and should change disrupt their market. I just want to add that this impact is driven by innovation and partnership with our customers. Microsoft has multiple and incredible teams of scientists like yourself that are helping companies integrate and up the systems like Azure Quantum Elements into their broader R&D transformation efforts.

This means we are deeply invested in our customer success and as a positive feedback loop, as we learn from our customers and identify opportunities to make the platform even more useful to our customers. And I had a chance to recently see a great example of such innovation from a leader in the space. Very interesting. Could you tell us more?

Yes. And as a customer, I know you’re very familiar with. It’s Johnson Matthey. We’ve teamed up with them to help drive new discoveries in sustainable energy, more specifically in hydrogen fuel cells.

Johnson Matthey has been in business over 200 years and is deeply involved in the transport, energy and chemical processing sectors. Just one example of the reach is the one that 1 in 3 cars on the road today use the Johnson Matthey catalyst in their exhaust system to reduce harmful emissions. We are collaborating with their quantum chemists to develop new predictive modeling tools in Azure HPC, and refined workflows to accelerate chemical simulations, explore the potential of AI and get quantum ready. They’ve been able to accelerate certain quantum chemistry calculations and reduce the turnaround time on their scale workloads from six months to a week.

This impressive, no? And one area they’re focusing on in sustainability, is finding better catalysts for hydrogen fuel cells that power trucks and busses. These electric catalysts help facilitate electrochemical reactions in fuel cells that convert hydrogen to produce electricity. We’ve seen about two fold acceleration in quantum chemistry calculations to run. This is critical for the R&D, as you can see.

So it means they can move much faster to understand and design your electric catalyst, which is a big competitive advantage. Wow, that’s really amazing. Dominik, thank you so much for taking the time to talk with us today.

You’re very welcome, Nathan, and thank you for having me. Auf Wiedersehen. Thank you for taking the time to learn how the Azure Quantum Elements platform can help you accelerate chemistry and materials science discovery today and into the future. If you’d like to learn more, we have a few options. Please sign up to join the Azure Quantum Elements Private Preview.

You can learn more about our customer stories and use cases like Reaction Network Exploration or AI Accelerated Materials Discovery on our blog posts. You can see the links to these materials on the screen, and we’ll also paste them in the webcast chat. Our next episode will be in January 2024, and we’ll be talking about the exciting area of quantum networking. We hope you’ll join us.

On behalf of everyone in Azure Quantum. We appreciate your participation in the Quantum Innovator series today. Thank you very much.

2025-05-18 20:21

Show Video

Other news

Все равно купят?! — Тест RTX 5070 vs RTX 4070 SUPER, RTX 4070 Ti SUPER и RTX 5070 Ti 2025-06-02 18:18
Nvidia Shrugs Off China Concerns With Upbeat Forecast | Bloomberg Technology 2025-05-30 11:26
MIT Robotics - Cecilia Laschi - Methods and technologies for new robotics scenarios 2025-05-30 09:58