Growing the quantum ecosystem on Forest

Forest users have run more than 65 million experiments on our platform to date, including on our QVM, 8- and 19-qubit processors. Here’s a snapshot of the latest papers and projects resulting from those experiments, spanning quantum chemistry, machine learning, and quantum information.

Quantum simulation

Understanding properties of atoms and molecules is key for solving problems like drug discovery, catalyst design, even understanding the earliest moments of the universe. These problems become intractable on classical computers because the computational cost scales exponentially with the size of the system. With Forest, several teams have tackled small-scale versions of these types of problems on Rigetti’s near-term quantum computers.

Using constrained VQE for electronic structure calculations
Ilya Ryabinkin and Artur F. Izmaylov, University of Toronto & 
Scott Genin, OTI Lumionics

The variational quantum eigensolver (VQE) is a hybrid quantum-classical algorithm for efficiently calculating the ground state of a molecule. However, it falls short when calculating higher energy ionic states. In this new take on VQE, the team is able to search for specific states with a certain number of electrons, spin, or other electronic properties. They used this constrained VQE method to simulate various states of H2 and H2O on our 19Q-Acorn processor. The approach also serves to compensate for noise on near-term quantum devices, improving the quality of the electronic structure calculations.

Simulation of non-equilibrium dynamics
Henry Lamm and Scott Lawrence, University of Maryland

Understanding how quantum systems evolve over time remains an open challenge in physics. For example, it could one day lead to a deeper explanation for how matter was created in the instants after the Big Bang. Lamm and Lawrence introduced Evolving Density Matrices on Qubits (EρOQ), a new hybrid quantum-classical algorithm for simulating non-equilibrium dynamics of quantum systems. They implemented EρOQ on our 8Q-Agave processor and QVM to calculate real-time evolutions of small systems, a demonstration that could scale to more complex problems as larger quantum computers come online.

Lattice-model protein folding on universal gate-based quantum computers
Tomas Babej, Mark Fingerhuth, and Christopher Ing, ProteinQure

Predicting how proteins fold into their three-dimensional shapes could unlock new treatments for incurable diseases. Lattice protein models are coarse representations of proteins that can be used to explore the vast number of possible configurations, but these classical computational methods become limited for proteins with more than 100 amino acids. The ProteinQure group developed a hybrid algorithm for encoding the lattice protein folding problem on fixed qubit architectures, using the 19Q-Acorn processor to fold a small amino acid sequence. This proof-of-principle experiment is the first time a lattice protein has been folded on a gate-based quantum computing architecture.

Quantum machine learning

Another active area of research in quantum computing applications is finding a quantum advantage — such as exponential speedup or reducing sample complexity — to solve machine learning problems. In particular, the Bayesian approach to machine learning has the potential to achieve these advantages when applied on quantum devices. These recent papers explore Bayesian machine learning methods on our QPU and QVM:

Generative learning on quantum circuits
Yuxuan Du, Tongliang Liu, and Dacheng Tao, University of Sydney

Deep learning on a quantum computer
Zhikuan Zhao, Singapore University of Technology and Design & National University of Singapore; Alejandro Pozas-Kerstjens, The Barcelona Institute of Science and Technology; Patrick Rebentrost, & Peter Wittek, University of Toronto, Creative Destruction Lab, Vector Institute for Artificial Intelligence, & Perimeter Institute for Theoretical Physics

Quantum information

Current quantum computers are small and imperfect. Characterizing the performance of a device and finding efficient ways to account for errors are important for extracting the highest quality information out of these near-term devices.

Benchmarking quantum processors with random circuits
James R. Wootton, University of Basel

The effectiveness of a quantum device may vary between tasks and is not always easy to predict. Benchmarking is a useful way to assess the quality of a quantum processor, but many techniques depend on the size and connectivity of a given device. Wootton proposes a more universal benchmark that can be implemented on devices of any size and connectivity. Sampling random gates and measuring the difference between the expected and measured output of increasingly complex circuits indicates the level of noise of a particular device. Wootton ran this experiment on both Rigetti and IBM quantum processors.

Error mitigation on a quantum computer
Matthew Otten and Stephen Gray, Argonne National Laboratory

One approach to quantum error correction is encoding a logical qubit into many physical qubits. But near-term devices have on the order of just tens of qubits, and error correction remains one of the biggest challenges in processing quantum information. By combining the results of many slightly different experiments, the Argonne team was able to estimate a noise-free answer without the need for additional qubits.

Try it yourself!

We’re always excited when Forest users open-source their projects for other community members to build on top of and experiment with. Here are a few recent ones worth checking out:

Quantum autoencoder
Hannah Sim, Harvard University & Zapata Computing

Expressing quantum data, such as wavefunctions of molecules, using a small number of qubits is crucial when trying to maximize the capabilities of current and near-term quantum devices. During our April hackathon, Sim and her teammates built a simulation tool called QCompress, which can be used to compress data into low-dimensional representations. Their open source project includes a demo jupyter notebook for compressing the ground state of molecular hydrogen.

Quantum Music Composer
James Weaver, Pivotal Software

The Quantum Music Composer uses a quantum computer — our 8Q-Acorn processor in this example — to improvise a musical performance. Supply the desired probabilities for a given pitch to follow another both melodically and harmonically, and the quantum computer outputs a unique composition. Listen to one of the clips below or try it for yourself on GitHub.


If you’re new to quantum programming, get your Forest API key and download pyQuil to get started, and join our community.