Starting today, customers can explore three new experimental capabilities on QuEra’s Aquila device that provide increased flexibility and control for your workloads running in Amazon Braket, the quantum computing service of AWS. All of these capabilities are available through Braket Direct, a program designed to expand how you explore quantum computing on AWS. Braket Direct provides researchers dedicated access to quantum hardware and the ability to engage directly with scientific experts – all from the same place.

With today’s launch, you can experiment with local detuning, taller geometries with increased lattice height, and tighter geometries with smaller vertical spacing – unlocking new use cases when programming on this device. In this blog, we focus on one of these new experimental capabilities, namely adding control over local detuning. We show how to use it with the Braket Python Software Development Kit (SDK) and highlight some potential use cases.

QuEra’s Aquila on Braket

Amazon Braket’s core proposition is to provide customers a choice between diverse quantum hardware modalities including superconducting, trapped-ion, and neutral atom-based devices, for researching algorithms and use cases.

The Aquila device is a special-purpose neutral atom-based machine suited to study the nature of quantum many-body systems and to research methods for solving certain combinatorial optimization problems. Unlike gate-based systems, this device is programmed using the analog Hamiltonian simulation (AHS) paradigm and simulates the quantum evolution of up to 256 qubits under a customizable Hamiltonian. Using the AHS paradigm, you can specify the exact spatial arrangement of qubits (which translate to pairwise interactions) and the time-dependence of external fields (see Figure 1), that exert coherent control over the qubits.

Figure 1 – An AHS program prescribes the time dependence of different control fields. Starting today, Aquila can run AHS programs with one global driving field and one local detuning. The amplitude and phase of the driving field have the same effect on the AHS qubits as the transverse field on spins; the sum of the detuning of the drive and local detuning produces the same effect on the qubits as the longitudinal field on spins.

To date, this unique paradigm has already been used to study a variety of scientific questions and industry applications on Aquila by mapping a problem of interest to the Hamiltonian supported by this device. These problems range from condensed-matter and quantum many-body physics problems like studying quantum spin liquids, to researching optimization problems in the automotive, aerospace, materials, and financial services industries.

Global driving field

Until today, Aquila could simulate the effect of only a global driving field: this field generates superposition through a coherent mixing between ground and excited states of the atoms.

The code snippet below shows how to instantiate a braket.ahs.driving_field.DrivingField object from braket.timings.time_series.TimeSeries objects using the Braket Python SDK. In Figure 2, we show the corresponding Hamiltonian term (for a complete example see our AHS example notebook).

Omega = TimeSeries(...)
phi = TimeSeries(...)
Delta_global = TimeSeries(...)
drive = DrivingField(
  amplitude=Omega,
  phase=phi,
  detuning=Delta_global
)

Figure 2 – The corresponding Hamiltonian term to match our code snippet.

In this Hamiltonian, k = 1,2, … N, indexes the qubits, |g_k〉 and |r_k〉 are the ground and excited states of qubit k; Ω is the amplitude, φ is the phase, and Δ_global is the detuning time series of the global driving field.

By prescribing the time dependence of amplitude, phase, and detuning over the course of the quantum program, you can study the behavior of the quantum system subject to an external coherent drive with customized time-dependence. This driving field affects all qubits uniformly (i.e. “globally”), by adjusting the spatial arrangements of qubits it is possible to investigate complex quantum behaviors (e.g., studying the emergence of the quantum spin liquid phase) or explore constrained optimization problems that can be applied to industry problems like vehicle routing.

Local detuning

Even though manipulating the global driving field alone is already powerful in its own right, there are certain problems that require even more fine-grained control: The ability to control local detuning coupled with the global driving field enables you to simulate the effect of a time- and position-dependent detuning on Aquila. Let us highlight two categories of use cases that can now be explored with this new capability.

1. Optimization problems: By carefully adjusting the pattern of local detuning, you can map the cost function of an extended maximum independent set problem to Aquila’s Hamiltonian where different vertices contribute different cost amounts. This makes it possible to model more complex combinatorial optimization problems that are relevant across a variety of industry applications, such as financial portfolio optimization. The Braket example notebook “Maximum weight independent set” shows how to find the optimal solution for a small instance of this problem family using quantum annealing.
2. Quantum field theory problems: Shifting specific qubits in and out of resonance, you can single out qubits and subject them to different coherent control schedules. This makes it possible to prepare quantum states with customized excitation patterns and study how they evolve in time. The Braket example notebook “Simulating lattice gauge theory with Rydberg atoms” shows how to do this in the context of studying a lattice gauge theory

Let’s dive into how to get started with using local detuning in your programs. The next code snippet shows how to instantiate the braket.ahs.local_detuning.LocalDetuning object from a braket.ahs.field.Field object containing one braket.timings.time_series.TimeSeries and one braket.ahs.pattern.Pattern object on the Braket Python SDK (for a complete example see our AHS example notebook).

Delta_local = TimeSeries(...)
h = Pattern(...)

shift = LocalDetuning(
  magnitude=Field(
    time_series=Delta_local,
    pattern=h
  )
)

This maps to the Hamiltonian depicted in Figure 3.

Figure 3 – the Hamiltonian represented by the code snippet, reflecting the time series of local detuning and the static strength of its pattern at the location of our qubit.

Here, Δ_local(t) is the time series of local detuning, and h_k, as the static strength of its pattern at the location of qubit k. Although the pattern (h) of the local detuning is static, the overall strength Δ_local can be programmed to change over the course of an AHS program.

From the perspective of any single qubit, changing the magnitude of local detuning (Δ_local x h) is equivalent to changing the detuning (Δ_global) of the global driving field by the same amount. While changing the detuning of the global driving field will affect all qubits equally, by changing the pattern of the local detuning we can fine tune how much it affects each specific qubit. Combining it with the global driving field, makes it possible to study new problems with Aquila.

Conclusion

With Amazon Braket, we offer customers a variety of quantum hardware and the latest set of capabilities to accelerate research and innovation. As part of this mission, we also provide access to emergent new features that enable you to explore the cutting edge of today’s quantum technology and unlock new use cases, as the technology matures. Some experimental capabilities may operate outside of standard device specifications and need hands-on guidance tailored to your use case. They are therefore available by request only via Braket Direct to ensure your workloads are set up for success.

You can get access to all experimental capabilities, including local detuning and expanded geometries on Aquila, through Braket Direct: in the Braket service console, simply open the ‘Braket Direct’ page and navigate to ‘Experimental capabilities’ – then, choose your desired feature and select ‘Get Access’. You can use these features with both on-demand and dedicated device access, at no additional charge. Researchers at accredited institutions can also apply for credits to support experiments on Amazon Braket through the AWS Cloud Credits for Research program.

To get started, check out our two new example notebooks on “Maximum weight independent set” and “Simulating lattice gauge theory with Rydberg atoms”. These notebooks walk through setting up an AHS program, show how to test local detuning on a smaller scale and how to run experiments using this capability on Aquila.

For more information on device capabilities and the AHS paradigm, see Aquila’s page on the Braket console and the Braket Developer Guide. Lastly, if you are running a hybrid variational workload, you can use Amazon Braket Hybrid Jobs and gain priority access to the device.

Feeling inspired? AWS is collaborating with Airbus, BMW Group, and The Quantum Insider on the global Airbus-BMW Group Quantum Mobility Quest to explore how quantum computing may help tackle the most pressing challenges in the aviation and automotive sectors. This quantum computing challenge is open to participants from research groups and companies worldwide. Learn more about the use cases here, register to submit your entries prior to the April 30 2024 deadline, and develop and test your approaches using Aquila and other quantum devices on Braket.