The new technique of researchers at Princeton University, the University of Chicago and IBM, significantly improves the reliability of quantum computers by utilizing the data for the noisy operation of genuine hardware. In a paper presented this week, researchers describe a new compilation method that increases the ability of limited resources and "noisy" quantum computers to produce useful answers. Namely, the researchers demonstrated almost three times the average enhancement of reliability for the real system of IBM's 16-kubit quantum computer IBM, which improved some executions programs eighteen times.
A joint research group includes computer scientists and physicists from EPiQC (Enabling Practical Quantum Computing) Collaboration, a NSF expedition to a computer that started in 2018. EPiQC aims to bridge the gap between theoretical quantum applications and programs for practical quantum computer architectures in close terms. EPiQC researchers collaborated with IBM quantum computing experts for this study, which will be presented at the 24th AMM International Conference on Language and Operating Systems Support for Architecture (ASPLOS) in Providence, Rhode Island on April 17.
Adjustment of noise loss programs
Quantum computers are composed of qubits (quantum bits) that are equipped with special properties of quantum mechanics. These special features (superposition and entanglement) allow the quantum computer to represent a very large space of opportunity and comb them through for the right answer, finding solutions much faster than the classic computers.
However, quantum computers today and the next 5-10 years are limited by noisy operations, where quantum computer processing operations generate inaccuracies and errors. As you run the program, these errors accumulate and potentially can lead to wrong answers.
To offset these errors, users run quantum programs thousands of times and choose the most common response as the correct answer. The frequency of this response is called the success rate of the program. In an ideal quantum computer, this success rate would be 100% – each hardware run will produce the same response. However, in practice, success rates are much less than 100% due to noisy operations.
The researchers noted that on real hardware, such as the 16-qubit IBM system, the quantum operation error rates have very large differences in the different hardware resources (qubits / gates) in the system. These error rates can vary from day to day. The researchers found that the operation error rates could have up to 9 times more variations depending on the time and location of the operation. When it comes to a program on this machine, the hardware cubes selected for Run determine the success rate.
"If we want to run the program today, and our compiler chooses a hardware port (operation) that has a bad mistake, the success rate of the program dramatically decreases," said researcher Prrakash Murali, a graduate student at Princeton University. "Instead, if we bring awareness of this noise and perform our programs using the best cubits and operations in the hardware, we can significantly increase the success rate."
To exploit this idea of adapting the program's execution to hardware noise, the researchers have developed "noise adjusting" compilers that use detailed data to characterize the noise for target hardware. Such noise data is routinely measured for IBM quantum systems as part of the daily operational calibration and includes error rates for each type of work capable of hardware. Using these data, the compiler displays cubic programs of hardware cubes that have low error rates and allocate ports quickly to reduce the chances of decomposition of the country from disorientation. In addition, it also minimizes the number of communication operations and performs them using reliable hardware operations.
Improving the quality of the duration of a real quantum system
To demonstrate the impact of this approach, the researchers compiled and performed a series of benchmark programs on the 16 quantum IBM quantum computer, comparing the success rate of their new Compatible Execution Noise Adapter from the IBM Qiskit Compiler, the default compiler for this machine. Through benchmarks, they recorded almost three times average improvement in success, with up to eighteen times improvements to some programs. In several cases, the IBM compiler produces faulty responses to executions due to its uncertainty in noise, while the adapted noise compiler provides accurate responses with high success rates.
Although the team's methods were demonstrated on a machine with 16 kubits, all quantum systems in the next 5-10 years are expected to have noisy operations due to difficulty in performing precise gates, defects caused by lithographic production, temperature fluctuations, and other sources. Adjustability of noise will be key to exploiting the computing power of these systems and pave the way for large quantum calculations.
"When working with large-scale programs, we want successful rates to be high so they can distinguish the right response from the noise and also reduce the number of repeat things needed to get an answer," Murali said. "Our evaluation clearly shows that adapting noise is crucial to achieving the full potential of quantum systems."
The full document of the team, "Mapping the Compiler for Forest Quantitative Computer with Noise Computing", is now released on arXiv and will be presented at the 24th AMM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) conference in Providence , Rhode Island on April 17.
Studies allow increasing the speed of quantum computers
Pray Murali et al. Mapping the compiler to a mid-size noisy quantized computer compiler, arXiv: 1901.11054 [quant-ph] arxiv.org/abs/1901.11054
The researchers use noise data to increase the reliability of quantum computers (2019, April 17)
received April 17, 2019
This document is subject to copyright. Apart from any fair dealing with a private study or research, no
part can be reproduced without written permission. Content is provided for informational purposes only.