Compiler Technology for Quantum Computers

We aim to develop innovative quantum compiler technology to harness the power of quantum computing. Quantum compilers are designed to translate high-level programmed quantum circuits into quantum instructions and optimize them to run on quantum computers with improved performance. We are working on state-of-the-art compiler technology to resolve the challenges of quantum noise vulnerability, qubit resource limitation, and so on.

Compiler Technology for AI Accelerators

We strive to study AI compiler technology to enhance the performance and efficiency of AI hardware accelerators. AI compilers analyze the input model and generate optimized executable code for the accelerators to diminish execution time, power consumption, and hardware requirements. Advanced AI compilers aim to fully unlock the potential of AI hardware, rapid adoption in diverse applications, and enable more efficient AI-powered solutions.

Quantum Computer Simulation Acceleration

Due to the limitations of real quantum machines, quantum computer simulations are still mainly employed to study quantum algorithms. However, large-scale quantum circuit simulations severely suffer from massive vector-matrix multiplications, exponentially increasing with the number of qubits. We are investigating the techniques to accelerate the simulation of large-scale quantum circuits by using emerging memory and PIM technologies.

Quantum Random Access Memory

Quantum random access memory (QRAM) is essential for running Oracle-based quantum algorithms. Unlike classical memory, a QRAM organizes memory cells into qubits, so we can access all possible data combinations stored in a quantum memory cell at once by using quantum superposition. The QRAM is a promising quantum technology, but many issues still need to be solved for practical adoption. We are conducting research on the architecture optimization and error mitigation of the QRAM.

Quantum Machine Learning

Quantum machine learning (QML) is an exciting and innovative field of research that explores the intersection of quantum computing and machine learning with the goal of revolutionizing the field of artificial intelligence. QML has the potential to significantly improve the efficiency and effectiveness of machine learning algorithms. We conduct research using compiler technology to maximize the performance and mitigate errors of QML in fields such as drug discovery, finance, and optimization.

Quantum Error Mitigation

Our study is centered on enhancing error mitigation for Noisy Intermediate-Scale Quantum (NISQ) machines. Recent quantum systems are particularly susceptible to errors due to environmental noise and qubit imperfections. By studying advanced architectures and algorithms tailored explicitly for NISQ machines, we aim to significantly improve their computational stability and accuracy. This work is pivotal for making NISQ technology more reliable in the short term and developing a foundational framework for the broader field of quantum computing.

Embedded System Optimizations

We also perform optimization research on various embedded systems, including eyewear for AR/VR, electric power steering (EPS), and advanced power systems. This enables us to improve the performance of those embedded systems and increase their resource efficiency. Our researchers have rich development and research experience in developing and optimizing embedded systems in both hardware and software.