The new software development toolkit for quantum natural language processing tested and benchmarked on System Model H1 technology

Telling Alexa to play “Schrodinger’s Cat” by Tears for Fears. Asking Siri for directions to a quantum-themed bar or restaurant. A smart phone autocorrecting a word in a text message.

These are everyday applications of natural language processing – NLP for short – a field of artificial intelligence that focuses on training computers to understand words and conversations with the same reasoning as humans.

NLP technologies have advanced rapidly in recent years with the help of increasingly powerful computing clusters that can run language models that examine reams of text and count how often certain words appear. These models train devices to retrieve information, annotate text, translate words from one language to another, answer questions, and perform other tasks.

The next step is to “teach” computers to infer meaning, understand nuance, and grasp the context of conversations. To do that, however, requires massive computational resources and multiple algorithms or data structures.

A United Kingdom-based quantum computing company believes the answer lies with qubits, superposition, and entanglement.

Cambridge Quantum recently released lambeq, a new open-source software development toolkit, that enables researchers to convert sentences into quantum circuits that can be run on quantum computers. It is the first toolkit developed specifically for quantum natural language processing – or QNLP - and was tested on System Model H1 technology before it was released.

The software takes the text, parses it, and then uses linguistics and mathematics to differentiate between a verb, noun, preposition, adjectives, etc., and label them to understand the relationships between words.

Cambridge Quantum researchers tested 30 sentences on the System Model H1, which was able to classify words correctly 87 percent of the time.

“We deem that a success,” said Konstantinos Meichannetzidis, a member of the CQ team. “We found that our software works well with the Honeywell technology and were able to benchmark the performance of this quantum device.”

The lambeq project also represented a first for Honeywell Quantum Solutions. It was the first QNLP problem run on the System Model H1 hardware.

“We are really excited to be a part of this work and contribute to the development of this important toolkit,” said Tony Uttley, president of Honeywell Quantum Solutions. “Applications like this help us test our system and understand how well it performs solving different problems.”

(Honeywell Quantum Solutions and Cambridge Quantum have a long-standing history of partnering together on research and other projects that benefit end-customers. The two entities announced in June they are seeking regulatory approval to combine to form a new company.)

Why QNLP?

For humans, decoding conversations to understand meaning is a complex process. We infer meaning through tone of voice, body language, context, location, and other factors. For computers, which do not rely on heuristics, decoding language is even more complex.

The only way to create some sort of “meaning-aware” NLP is to explicitly encode compositional, semantic sentence structure into language models. To do this on a classical computer, however, requires massive computational resources, which are costly, and would likely still take months to process.

Quantum computers, on the other hand, run calculations and crunch data very differently.

They harness unique properties of quantum physics, specifically superposition and entanglement, to store and process information. Because of that, these systems can examine problems with multiple states and evaluate a large space of possible answers simultaneously.

What this means in terms of natural language processing is that quantum computers are likely to go beyond counting how often certain words appear or are used together. As noted above, quantum computers can identify words, label them as a noun, verb, preposition, etc., and understand the relationship between words. (lambeq uses the Distributional Compositional Categorical – or DisCoCat – model to do this.)

This enables the computer to infer meaning, and also provides insight into how and why the computer made connections between words. The latter is important for validating data and also expanding the use of QNLP in regulated sectors such as finance, legal, and medicine where transparency is critical.

Built upon previous work

The Cambridge Quantum team has long explored how quantum computing can advance natural language processing, and has published extensively on the topic.

In December 2020, researchers released two foundational papers that demonstrated that QNLP is inherently meaning-aware and can successfully interpret questions and respond.

Earlier this year, the team performed the first NLP experiment conducted on a quantum computer by converting more than 100 sentences into quantum circuits using an IBM technology. Researchers successfully trained two NLP models to classify words in sentences.

The release of lambeq and the testing of the open-source toolkit on the Honeywell System Model H1 represents the next steps in their QNLP efforts.

“Our team has been involved in foundational work that explores how quantum computers can be used to solve some of the most intractable problems in artificial intelligence,” said Bob Coecke, Cambridge Quantum’s chief scientist.

“In various papers published over the course of the past year,” Coecke added, “We have not only provided details on how quantum computers can enhance NLP but also demonstrated that QNLP is ‘quantum native,’ meaning the compositional structure governing language is mathematically the same as that governing quantum systems. This will ultimately move the world away from the current paradigm of AI that relies on brute force techniques that are opaque and approximate.”

Honeywell Quantum Solutions notched another important milestone this week with its trapped-ion quantum computing technology.

The System Model H1 became the first quantum computer to pass the Quantum Volume 1024 benchmark, a metric introduced by IBM to measure the overall capability and performance of a quantum computing system regardless of technology. (Calculating quantum volume requires a complex set of statistical tests.)

The achievement marks a new record for performance in terms of quantum volume, and the third set by the System Model H1 since it was launched in fall 2020. It also fulfills a promise made last summer that Honeywell Quantum Solutions would increase the quantum volume of its commercial offerings by an order of magnitude each year for the next five years.

“We achieved what we set out to do,” said Tony Uttley, president of Honeywell Quantum Solutions. “Our goal is to provide users with the most powerful hardware as they work on solving real world problems. We believe that being able to quantify the increases in capability is important."

This is the latest in a string of accomplishments for Honeywell Quantum Solutions, which recently announced it was combining with Cambridge Quantum Computing to form the largest stand-alone quantum computing company in the world.

Over the past year, the Honeywell team:

• Launched two commercial computing systems. The System Model H0 was released in June 2020 followed by the System Model H1 four months later.
• Set four industry records for quantum volume. The System Model H0 debuted with a then-record quantum volume of 64. The System Model H1 launched with a quantum volume of 128, a new record, and through system upgrades, passed the quantum volume benchmarks of 512 in March and now 1024 in July.
• Developed and demonstrated the holographic quantum dynamics (holoQUADS) algorithm, which can accurately simulate a quantum dynamics model with fewer qubits than traditional methods. The algorithm could lead to quantum computers running more complex scientific simulations sooner than expected.
• Completed repeated rounds of quantum error correction and demonstrated it can detect and correct quantum errors in real-time while a computation is running.
• Forged several new collaborations with enterprise partners and businesses, including DHL, BMW, Nippon Steel, Samsung, and others.
  

An algorithm developed by researchers at Honeywell Quantum Solutions could lead to quantum computers running more complex scientific simulations sooner than expected.

The Honeywell team recently demonstrated that its holographic quantum dynamics (holoQUADS) algorithm accurately simulated a quantum dynamics model with fewer qubits than traditional methods. The algorithm used nine qubits to simulate 32 “spins” – or localized electrons. Traditional methods require one qubit per spin.

The demonstration, led by Eli Chertkov, has important implications. Simulating quantum dynamics is a promising application for quantum computers. However, many predict quantum computers will need hundreds or thousands of qubits to run simulations too complex for classical computers.

The holoQUADS algorithm could change that.

“This algorithm allows us to run more complex simulations with less than a third of the qubits,” said Tony Uttley, president of Honeywell Quantum Solutions. “This is an exciting achievement that gets us closer to quantum computers solving real-world problems that classical computers cannot.”

Borrowed from the classical world

Scientists have long sought to better understand how atoms and subatomic particles move, behave, and interact (known as quantum mechanics) and react when disturbed (quantum dynamics).

Such knowledge is critical to the development of new vaccines and gene therapies, and the discovery of novel materials that are stronger, longer lasting, or better conductors of heat or electricity.

Currently, it is impossible to fully simulate the quantum dynamics of systems larger than a few atoms, and many believe it always will be. Classical computers crunch data by manipulating ones and zeroes and represent states as “off” or “on.” Atoms and subatomic particle exist in multiple states and move and behave in different ways.

This is what led to famed American physicist Richard Feynman postulating in the 1980s that only computers that are quantum in nature can adequately simulate quantum dynamics. 

That is not to say computational scientists do not have tricks to model some aspects of quantum dynamics on classical computers. They have developed powerful algorithms such as tensor networks to approximate quantum states.

In fact, the holoQUADS algorithm is based on tensor networks. These mathematical tools compress data and scientists use them to study the quantum nature of different materials.

The Honeywell team published a paper last May detailing the steps necessary to adapt tensor networks for a quantum computer and how to extend them to simulate dynamics. They published a second paper explaining how quantum tensor networks can measure the degree to which parts of a system are entangled, or entanglement entropy, which is used for studying topological properties of materials. 

The recent demonstration showed the dynamics algorithm described in the original paper is not only efficient but can return quantitatively accurate results with trapped-ion hardware available right now. 

Tested and verified

The Honeywell team tested the algorithm by simulating the chaotic dynamics of the “kicked” Ising model, a mathematical framework used to study chaos and thermalization in strongly interacting quantum systems. The results mirrored those generated by simulations on classical computers.

The demonstration served as an important benchmark and will help the team verify performance and accuracy as they scale the algorithm and quantum hardware.

“The model we simulated is a perfect test of the algorithm because it behaves in many ways like a typical chaotic quantum system, but it has a very special feature that lets us check the results classically,” said Dr. Michael Foss-Feig, a physicist who helped develop the algorithm.

Chertkov, Foss-Feig, and the other co-authors are excited by how well the algorithm worked in the real world, and by the performance of the System Model H1. The algorithm relies on mid-circuit measurement and qubit reuse, techniques first demonstrated by Honeywell. The H1 is adept at both.  And because of the H1’s high fidelities, the raw data had less “noise” than other state-of-the art simulations.

“The QCCD architecture at the heart of System Model H1 enables high-fidelity qubit reset and mid-circuit measurements with very low crosstalk errors,” said Justin Bohnet, one of the co-authors who led the hardware team. “Those features, along with the long coherence times and high-fidelity gates provided by trapped-ion qubits, are enabling creative advances in the study of quantum systems, as shown by this the holoQUADS demonstration.”

The online tool adapts quantum circuits to run optimally on different quantum computing technologies, with easy switching between Honeywell’s system and other hardware platforms.

Cambridge Quantum recently announced it has made the source code for TKET, its quantum software development kit, fully open to the quantum software community.

The move, which comes just months after the company began providing free access to TKET, is expected to benefit software developers as well as Honeywell Quantum Solutions and other hardware providers.

Most of the programming languages or quantum software development kits available were designed initially to run on certain hardware platforms, creating compatibility issues. Software developers who wanted to test circuits or algorithms on different quantum technologies had to rewrite or tweak code to run on a new system.

Providing access to TKET and its source code makes it easier for developers to do that.

“Users need only to focus on developing their quantum applications, not rewriting code around the idiosyncrasies of any particular hardware,” said Dr. Ross Duncan, head of software at Cambridge Quantum.

And for Honeywell Quantum Solutions and other hardware providers, TKET broadens access to their technologies by enabling developers to move more seamlessly between systems. The software development kit is optimized for each commercial hardware system, including Honeywell’s trapped-ion quantum computer.

“We want to make it as easy as possible for the quantum software community to run circuits and algorithms on our trapped-ion quantum computers,” said Tony Uttley, president of Honeywell Quantum Solutions. “The System Model H1 technology is the highest performing quantum system available, and we want them to experience that.”

Honeywell Quantum Solutions and Cambridge Quantum have a long-standing history of partnering together for the benefit of end-customers. (The two entities announced in June they are seeking regulatory approval to combine to form a new company.)

"Having TKET fully open-sourced provides an incredible tool to the world’s quantum algorithm developers, including Honeywell," Uttley said.

“Our products and offerings have always been complementary and continue to be,” he said. “(Cambridge Quantum) has developed a suite of tools and programs that interface well with our hardware.”

How TKET works

If you have ever traveled to another country and tried plugging something in, you’ve likely discovered the need for an adapter. Electrical outlets vary and the plug-ins used in the United States don't always work in Europe or other countries and vice versa.

The same is true with today’s early-stage quantum computers. Each technology has its own performance specifications, API (an interface that enables different computing systems to talk to one another), and compiler (a program that translates code written in one computing language to another).

TKET is versatile. Developers can use it to create circuits or algorithms and also to serve as a universal connector or adapter between hardware and software platforms.

Cambridge Quantum has developed extensions, which are Python modules, for each commercial quantum hardware platform available. These extensions enable developers to code in Qiskit, Cirq, or another language and automatically adapt their circuits or algorithms to work on different quantum devices or simulators without having to tweak it themselves.

And now that TKET is open source, developers can create their own extensions to the codebase and bridge between platforms.

Cambridge Researchers at Honeywell Quantum Solutions have turned problematic micromotion that jostles trapped ion qubits out of position into a plus.

The team recently demonstrated a technique that uses micromotion to shield nearby ions from stray photons released during mid-circuit measurement, a procedure in which lasers are used to check the quantum state of certain qubits and then reset them.

Mid-circuit measurement is a key capability in today’s early-stage quantum computers. Because the qubit’s state can be checked and then re-used, researchers can run more complex algorithms – such as the holoQUADS algorithm – with fewer qubits.

By “hiding” ions behind micromotion, Honeywell researchers significantly reduced the amount of “crosstalk” – errors caused by photons hitting neighboring qubits – that occurred when measuring qubits during an operation. (Details are available in a pre-print publication available on the arXiv.)

“We were able to reduce crosstalk by an order of magnitude,” said Dr. John Gaebler, Chief Scientist of Commercial Products at Honeywell Quantum Solutions, and lead author of the paper. “It is a significant reduction in crosstalk errors. Much more so than other methods we’ve used.”

The new technique represents another step toward reducing errors that occur in today’s trapped-ion quantum computers, which is necessary if the technology is to solve problems too complex for classical systems.

“For quantum computers to scale, we need to reduce errors throughout the system,” said Tony Uttley, President of Honeywell Quantum Solutions. “The new technique the Honeywell team developed will help us get there.”

Eliminating errors

Today’s quantum computing technologies are still in the early stage and are prone to “noise” - or interference - caused by qubits interacting with their environment and one another.

This noise causes errors to accumulate, corrupts information stored in and between physical qubits, and disrupts the quantum state in which qubits must exist to run calculations. (Scientists call this decoherence.)

Researchers are trying to eliminate or suppress as many of these errors as possible while also creating logical qubits, a collection of entangled physical qubits on which quantum information is distributed, stored, and protected.

By creating logical qubits, scientists can apply mathematical codes to detect and correct errors and eliminate noise as calculations are running. This multi-step process is known as quantum error correction (QEC). Honeywell researchers recently demonstrated they can detect and correct errors in real-time by applying multiple rounds of full cycles of quantum error correction.

Logical qubits and QEC are important elements to improving the accuracy and precision of quantum computers. But, Gaebler said, those methods are not enough on their own.

“Everything has to be working at a certain level before QEC can take you the rest of the way,” he said. “The more we can suppress or eliminate errors in the overall system, the more effective QEC will be and the fewer qubits we need to run complex calculations.”

Cutting out crosstalk

In classical computing, bit flip errors occur when a binary digit, or bit, inadvertently switches from a zero to one or vice versa. Quantum computers experience a similar bit flip error as well as phase flip errors. Both errors cause qubits to lose their quantum state – or to decohere. In trapped ion quantum computing, one source of errors comes from the lasers used to implement gate operations and qubit measurements.

Though these lasers are highly controlled, unruly photons (small packets of light) still escape and bounce into neighboring ions causing “crosstalk” and decoherence.

Researchers use a variety of methods to protect these ions from crosstalk, especially during mid-circuit measurement where only a single qubit or a small subset of qubits is meant to be measured. With its quantum charged-coupled device (QCCD) architecture, the Honeywell team takes the approach of moving neighboring ions away from the qubit being fluoresced by a laser. But there is limited space along the device, which becomes even more compact as more qubits are added.

“Even when we move them more than 100 microns away, we still get more crosstalk than we prefer,” said Dr. Charlie Baldwin, a senior advanced physicist and co-author of the paper. “There is still some scattered light from the detection laser.”

The team hit on hiding neighboring ions from stray photons using micromotion potentials, which are caused by the oscillating electric fields used to “trap” these charged atoms. Micromotion is typically thought of as a nuisance with ion trapping, causing the ions to rapidly oscillate back and forth, and occurs when the ions are pushed out of the center of the trap by additional electric fields.

“Usually, we are trying to eliminate micromotion but in this case, we were able to use it to our benefit,” said Dr. Patty Lee, chief scientist at Honeywell Quantum Solutions.

The team’s goal is to reduce by 10 million the probability of a neighboring ion absorbing photons at 110 microns away. By moving neighboring ions and hiding them behind micromotion the Honeywell team is approaching that mark.

How and why the technique works

In their paper, Honeywell researchers delved into how and why hiding ions with micromotion works, including the ideal frequency of the oscillations. They also identified and characterized errors. (The basic physics behind the concept of hiding ions was first explored by the ion storage group at the National Institute of Standards and Technology.)

“Mid-circuit operations are a new feature in commercial quantum computing hardware, so we had to invent a new way to validate that the micromotion hiding technique was achieving the low level of crosstalk errors that we predicted,” said Dr. Charlie Baldwin.

Though the new method resulted in a significant reduction of crosstalk errors, the Honeywell team acknowledged there is further to go.

“Crosstalk is one of those scary errors for scaling,” Gaebler said. “It has to be controlled because it becomes more of a problem as you scale and add qubits. This is another tool that will help us scale and help us compact our systems and pack in as many qubits as we can.”

Cambridge Researchers at Honeywell Quantum Solutions have taken a significant step toward demonstrating the viability of large-scale quantum computing on its trapped-ion quantum computing technology. 

The Honeywell team can now perform quantum error correction (QEC), which are protocols necessary to detect and correct errors in real time on a quantum computer. They demonstrated the ability to “protect” quantum information (prevent a quantum computation from being quickly corrupted by imperfections and noise) on the System Model H1. This is an important first in the quantum computing industry. Currently, most demonstrations of quantum error correction involve correcting errors or “noise” after the procedure has finished running, a technique known as post-processing.

In a paper published this week on arXiv, researchers detailed how they created a single logical qubit (a series of entangled physical qubits) and applied multiple rounds of quantum error correction. This logical qubit is protected from two main types of errors that occur in a quantum computer: bit flips and phase flips.

Previously, groups have looked at codes that only are capable of correcting a single type of error (bit or phase but not both) {Google, IBM/Raytheon, IBM/Basel}. Others have looked at quantum error detecting codes, which can detect both types of errors but not correct them {ETH, Google, Delft}. Further still, groups have demonstrated pieces of the quantum error correcting process {Blatt, Monroe}. 

“All of today’s quantum technologies are at an early stage where they must combat errors that accumulate during computations,” said Tony Uttley, president of Honeywell Quantum Solutions. “What the Honeywell team accomplished is groundbreaking. It proves what was once only theoretical, that quantum computers will be able to correct errors in real time, paving the way for precise quantum computations.”

Though the achievement represents progress toward large-scale quantum computing, Honeywell researchers are still working to cross the break-even point at which the logical error rate is less than the physical error rate. 

The need for logical qubits

To appreciate this achievement, it is important to understand how difficult it is to detect and then correct a quantum error.

Quantum bits, or qubits, are fragile and finicky. They pick up interference or “noise” from their environment. This noise causes errors to accumulate and corrupts information stored in and between physical qubits. Scientists call this decoherence.

Attempts to directly detect and correct errors on a physical qubit also corrupts its “quantumness.” And cloning this data, a method used in classical computing that involves making multiple exact copies of the information, does not work in quantum (as prohibited by “The No Cloning Theorem”.)

To overcome these concerns, several scientists, most notably Peter Shor, Robert Calberbank, and Andrew Steane, found a way around this, at least in theory, after studying how quickly qubits experience decoherence.

They demonstrated that by storing information in a collection of entangled qubits, it was possible to detect and correct errors without disrupting quantum information. They called this assortment of entangled qubits a logical qubit. 

Scientists have spent years developing codes and methods that could be applied to logical qubits to protect quantum information from errors.

What’s next

The next step is to break even, crossing the point at which the logical qubit error rate is lower than the error rate for physical qubits. (Creating logical qubits and applying quantum error correction codes also can inject noise into a system).

The Honeywell team is closing in on that mark. To definitively demonstrate passing the break-even point, the error rate per QEC cycle needs to be lower than the largest physical error rate associated with the QEC protocol. 

“In the technical paper, we point to key improvements we need to make to reach the break-even point,” said Dr. Ciaran Ryan-Anderson, an advanced physicist and lead author of the paper. “We believe these improvements are feasible and are pushing to accomplish this next step.”

From there, the goal is to create multiple logical qubits, which depending on the quantum technology, requires better fidelities, more physical qubits, better connectivity between qubits, and other factors.

An increase in logical qubits will usher in a new era of fault-tolerant quantum computers that can continue to function even when some operations fail. (Fault tolerance is a design principle that prevents errors from cascading throughout a system and corrupting circuits.)

“The big, enterprise-level problems we want to solve with quantum computers require precision and we need error-corrected logical qubits to scale successfully,” Uttley said.