February 3, 2017
The annual international Quantum Information Processing (QIP) series is the premier meeting for theoretical quantum information research. Hosted by Microsoft’s QuArC Group, it features researchers from leading academic institutions. The list of attendees this year also included several corporations, such as 1QBit, Rigetti, IBM, Microsoft, and Google.
The three overarching themes especially relevant to the work of 1QBit included research on multiple quantum architectures, quantum machine learning, and quantum supremacy.
1. This year we were excited about the wide variety of quantum computing architectures and simulations that were presented. As researchers and software developers, the near-term results of various architectures and simulations are particularly interesting for us to explore and integrate in the development of our hardware-agnostic applications.
Here are a couple of highlights from the talks that focused on this topic:
Simulated quantum annealing can be exponentially faster than classical simulated annealing – Elizabeth Crosson and Aram Harrow
Adiabatic optimization versus diffusion Monte Carlo – Michael Jarret, Stephen Jordan, and Brad Lackey
2. Interesting viewpoints presented on quantum machine learning included Srinivasan Arunachalam and Ronald de Wolf’s presentation describing an algorithm having no quantum advantage over classical methods in the context of sample complexity, and Iordanis Kerenidis and Anupam Prakash’s development of a quantum recommendation system which showcase exponential speedup over known classical methods.
Quantum recommendation systems – Iordanis Kerenidis and Anupam Prakash
Optimal quantum sample complexity of learning algorithms – Srinivasan Arunachalam and Ronald de Wolf
Spectrahedral lifts and quantum learning – James Lee
3. The concept of quantum supremacy was discussed by multiple presenters. Sergio Boixo’s team from Google claims that “quantum supremacy can be achieved in the near-term with approximately fifty superconducting qubits”. Scott Aaronson and Shalev Ben-David developed a method to find subproblems within a problem for which quantum algorithms provide an exponential advantage.
Characterizing quantum supremacy in near-term devices – Sergio Boixo, Sergei Isakov, Vadim Smelyanskiy, Ryan Babbush, Nan Ding, Zhang Jiang, Michael Bremner, John Martinis, and Hartmut Neven
Threshold theorem for quantum supremacy – Keisuke Fujii
Sculpting quantum speedups – Scott Aaronson and Shalev Ben-David
Along with the talks and plenary sessions, the poster sessions were also a popular venue for exploring new architectures and quantum information theory. 1QBit researcher Raouf Dridi presented the poster, “Two Topos Interpretations for Measurement Based Quantum Computations”. This work investigates the nature of speedup or computational advantage in measurement based quantum computation.
During nightly tutorial events, both Microsoft and Rigetti showcased their quantum gate model-based programming frameworks. Rigetti has designed an assembly-style language called Quil to interact with a gate model quantum device, along with a Python implementation called pyQuil. Microsoft demonstrated LIQUi|>, a software architecture and tool suite for quantum computing. 1QBit provided personalized introductions to 1QBit’s quantum software development kit, for which limited-version access was made available last spring in the form of tutorials, documentation, and Jupyter notebooks. A limited version of these tools and tutorials can also be found at http://qdk.1qbit.com, to explore applications of both high-level algorithms and low-level tools to benchmark and build hardware-agnostic applications that easily interface with quantum annealing processors, quantum simulators, and classical processors.
Takeaways and implications for quantum information processing in 2017
We believe that research related to the concept of quantum supremacy now involves recognizing and characterizing which problems best exploit quantum computing. Machine learning, a topic increasingly studied by researchers in this field, has several candidate problems for demonstrating quantum supremacy. Additional research has shown that certain problems that could allow for quantum supremacy might also benefit from classical simulations of quantum algorithms. In general, the quantum computing community is discovering the types of problems that can benefit from advances in quantum hardware and quantum algorithms.
1QBit’s latest research in quantum machine learning (see corresponding paper, “Reinforcement Learning Using Quantum Boltzmann Machines”) investigates reinforcement learning using simulated quantum annealing, exploring the themes discussed above.