Unlocking the Potential of Quantum Computing: Hewlett Packard Labs’ Quantum Resurgence
Insider Brief
Hewlett Packard Labs is diving back into quantum computing with a fresh perspective, aiming to solve real-world industrial challenges through a holistic hybrid approach in collaboration with partners, as revealed by the company’s chief architect.
Imagine reducing the computational requirements for complex chemical modeling like benzine from needing 100 million qubits over 5,000 years to just 1 million qubits over a single year. This breakthrough is just the beginning of what Hewlett Packard Labs’ advances in quantum computing could achieve, with practical solutions becoming feasible within the next decade rather than generations.
While significant challenges in scaling qubits and managing error rates persist, the potential for quantum computing to merge with AI opens up exciting possibilities, potentially offering novel acceleration methods for machine learning tasks.
Image: HP
Hewlett Packard Labs has reignited its interest in quantum computing to address practical industrial problems, with a focus on simulating quantum systems to tackle challenges in chemistry and physics, according to the company’s chief architect.
After setting aside an earlier quantum project over a decade ago, Hewlett Packard Labs is leveraging its supercomputing prowess to make quantum computing more accessible and applicable.
“We’re embarking on a new opportunity rather than revisiting our old nitrogen-qubit project. We are focusing on new horizons,” stated Kirk Bresniker, Chief Architect at Hewlett Packard Labs, in an interview with Frontier Enterprise.
About 18 years ago, Hewlett Packard Labs delved into low-level qubit research using a nitrogen vacancy in a diamond lattice, explained Bresniker. While the project involved implanting a nitrogen atom into a lab-grade diamond chip to create an artificial atom functioning as a qubit, it was shelved 12 years ago due to uncertainties about its enterprise value.
“Hewlett Packard Labs has always been about connecting research to tangible business outcomes,” emphasized Bresniker. The team pivoted towards photonics, recognizing the pivotal role of efficient data transfer in enterprise applications.
The company’s acquisitions of SGI and Cray four years ago expanded its footprint in supercomputing, reigniting customer interest in quantum computing.
“As we delivered exascale supercomputers to customers from these acquisitions, they expressed interest in our work and inquired about integrating qubits into the next system,” Bresniker shared.
Hewlett Packard Labs is now embracing a holistic co-design approach, partnering with organizations working on diverse qubit technologies and quantum software. The goal is to simulate quantum systems to solve real-world challenges in solid-state physics, quantum chemistry, exotic condensed matter physics, and industrial applications.
“Our focus is on delivering the optimization promised by quantum computing at an industrial scale. We’re working towards answering the pivotal questions of when and where quantum solutions can have a real impact,” posed Bresniker.
Initially, the team tackled the daunting task of modeling benzine, an exotic chemical derived from benzene. “Our initial approach required 100 million qubits over 5,000 years—a monumental requirement difficult to meet with current quantum capabilities in the tens or hundreds of qubits,” Bresniker shared.
By implementing error correction codes and innovative simulation techniques, the team significantly reduced the computational demands, streamlining the requirement to 1 million qubits for just one year—a staggering 500,000x reduction. This breakthrough indicates the possibility of practical solutions within the next decade rather than far-off generations.
The strategy involves segmenting problems and assigning them to the most suitable computational resources. “Some parts might align well with GPUs, others with superconducting qubits, and some with trapped ion qubits,” Bresniker explained. By amalgamating quantum processing units with traditional classical supercomputing, a hybrid environment is created, embodying the essence of holistic co-design.
While progress is evident, obstacles obstruct the commercialization of quantum computing. “The crux of the matter in quantum lies in coherence time—how long a qubit can remain viable before vanishing. Additionally, addressing error rates adds complexity due to the probabilistic nature of qubits and their entanglement interactions,” Bresniker explained.
Scaling the number of qubits poses a formidable challenge. “Having 32 qubits in a vacuum chamber the size of a phone is not sufficient; we need millions,” emphasized Bresniker.
Technologies like superconducting qubits necessitate cooling to four millikelvin in a dilution refrigerator, presenting substantial engineering hurdles.
“Each technology brings forth significant engineering challenges, and that’s before integrating control systems across a calculation, which might span weeks, months, or even years,” Bresniker shared.
One area of intense exploration is the potential impact of quantum computing on artificial intelligence, an active field of research. “You can perform linear algebra on a quantum system, and there exist HHL algorithms to support this. The question of whether it is superior remains open,” Bresniker remarked.
He highlighted the escalating costs and resources needed to train large AI models. “In about three or four years, the cost of training a single model could surpass the current global IT expenditure,” Bresniker predicted.
The future could involve enhancing GPU efficiency, crafting application-specific accelerators, or delving into the realm of quantum computing. The convergence of quantum computing and machine learning presents an intriguing frontier.
“Our focus is on quantum machine learning, a space where we train machine learning algorithms to model quantum systems without necessitating a physical qubit,” Bresniker shared. This intersection holds great promise for Hewlett Packard Labs, representing a compelling fusion of quantum computing and AI.
“For us, this intersection could spark remarkable innovations,” he concluded.