Quantum computing system provider Rigetti Computing — which is working on both quantum hardware circuits as well as the software layers from firmware up to the application layer that will work on them — has released an open source quantum computing environment, Forest 1.0, for developers to start building on, even if they don’t personally have access to quantum computing hardware.
The goal, said Madhav Thattai, chief strategy officer at Rigetti, is to “allow customers and the community to learn about quantum computing and develop real applications.” With Forest 1.0, developers amongst Rigetti’s current customers can start creating quantum computing applications while others in the community can use Rigetti’s quantum virtual machine.
Quantum computing is a recent concept built on quantum physics, drawing on quantum mechanical phenomena such as to how light can be both a particle and a wave. At extremely high speeds, quantum computing can calculate algorithms using large linear algebra that is not possible today because of the multitude of alternative strains of mathematics that must be assessed and calculated simultaneously.
“Quantum systems have very subtle correlations. They have exponentially more correlations than in classical systems, relative to their size. So even a small quantum system can’t be fit on today’s largest supercomputers,” explained Will Zeng, Rigetti’s director of software and applications. “That is exactly what quantum computers use: they use the same correlations that are hard to simulate to do more advanced computations. If these are large linear algebra machines, then quantum computing is a resource to give you access to large linear algebra problems that you could ever have imagined. There is no other technology that has that potential.”
Thattai says that quantum computing uses qubits as a core unit of memory, which can double memory capability every time a qubit is added, making the way memory scales exponentially. Currently, Rigetti Computing is building systems using eight qubits, but with the launch of a new hardware lab, Fab-1, it is expecting to accelerate development significantly.
Zeng explained there aren’t large quantum computers yet that can solve problems faster than classical machines, but that is changing rapidly, especially in fields like quantum chemistry and machine learning, which Zeng says are two “near problems”.
“How do we get there to access these two domains?” Zeng mapped out: “We need to build the first small computers, then we need to program and exploit those machines. There has been lots of academic research in the last 20 years, but it is only in the last few years that industrial opportunities have emerged. But before we have large scale quantum computers, we need different tools to use it. We need to go from physical devices to an API.”
Zeng said that’s a big challenge that Rigetti is addressing in two stages: It has released open source tools so people can contribute early and it is integrating quantum computing into classical computing.
Forest 1.0 as an R&D platform
Zeng said users should think of the recent release of Forest as “an R&D platform for nascent quantum computing.” Forest has several components:
- An instruction set of tools: Forest comes with a language agnostic “quantum instruction language” called Quil. It uses a hybrid shared memory model so that applications can draw on classical computing instructions alongside quantum computations. In some ways, this hybrid model is akin to how serverless is often being used in enterprises. In hybrid serverless systems, instead of a complete replacement to existing cloud or on-prem architecture, specific tasks are sent into a serverless environment and after compute, and the results are returned to the legacy or cloud architecture. For example, new images may be stored in a cloud storage, which triggers a copy of that image being sent to a serverless workspace to run image processing algorithms, which in turn identify objects or faces in the images, and then return the image with the new descriptors to the cloud storage after the algorithm has been processed. Zeng believes this sort of application environment will be available within the next two to five years, requiring a massive uptick in skills development over that time amongst the developer community.
- Client-side libraries: As Quil is language agnostic, developers can write in it directly, much like writing in assembly. But most, Zeng believes, will prefer to use some of the included client-side libraries such as a provided Python library:
- API: While developers working within Rigetti Computing’s current customer base already has some qubit processor hardware they can start programming on, the majority of developers coming to the Forest platform are expected to build applications that connect via API to Rigetti’s Quantum Virtual Machine (QVM). While not suited to production use cases yet, Zeng said the API is “the best way to understand how quantum computing is going to influence programming. QVM is a hyper performance machine, it can simulate 36 qubits on classical computing. It lets you understand how algorithms will work in quantum computing. It is not just high performance, it includes realistic noise models, and all the details that make it a good environment that reflects what the hardware will really look like.”
Quantum Computing in Chemistry Use Cases
One of the most exciting areas where quantum computing can significantly change current science is in computational chemistry. A number of libraries already exist to help developers use Forest in computational chemistry.
“These are fundamentally quantum problems,” said Zeng. “One of the first things we could do is improve the design of catalysts. Catalysts are small molecules that are designed in quantum systems. It could have a huge impact if you make them 1, 2 or 5 percent better. One area we talk a lot about is nitrogen fixation. To create fertilizer we use a catalyst from 1910, and 1-2 percent of the world’s energy goes into this process. If you could make it a little bit better, this one molecule, you could impact the world’s energy budget significantly. You also have catalysts that can do things like pull carbon out of the atmosphere, if we could come up with cheap organic catalysts, we could make better organic batteries.”
Making Quantum Computing Accessible
While quantum chemistry is one area that is seen as low hanging fruit, so too are optimization algorithms. A number of libraries available in the Forest platform are already looking at using quantum computing for large optimization and machine learning use cases.
This is one of the areas where Rigetti’s current customer base are already experimenting with quantum application design.
From an implementation point of view, the key work to do next, according to Zeng, is to make quantum computing work in hybrid environments like the serverless example. Without wanting to share too much of their roadmap, Zeng confirmed “Increasing hybridization is what we are starting to do. On the client side, you have to split off the work to the quantum part and the non-quantum part.” Zeng wants future Forest and hardware development to lead them to a hybrid decision-making process where users can upload their problem and the platform will decide how to carve off what computations should be done on qubit processing and what should be done in classical computing.
More on the Quantum Instructional Language was shared at the start of the year in Seattle: