Amazon Braket provides access to a choice of different quantum computers from third-party hardware providers including quantum annealing from D-Wave, ion trap devices from IonQ, and gate-based superconducting machines from Rigetti. Here’s how each of our hardware providers describes their technology and approach to quantum computing.

## D-Wave

D-Wave is a leader in the development and delivery of quantum computing systems, software, and services, and is the world's first commercial supplier of quantum computers. Our mission is to unlock the power of quantum computing by delivering customer value through practical applications. Our systems are being used by some of the world’s most advanced organizations, from global enterprises such as Lockheed Martin, DENSO, and Volkswagen, to national research centers such as NASA Ames, Los Alamos National Lab, Oak Ridge National Lab, and Forschungszentrum Jülich.

The D-Wave quantum computer leverages quantum dynamics to accelerate and enable new methods for solving complex discrete optimization, constraint satisfaction, artificial intelligence, machine learning, materials science, and simulation problems. These problem types are applicable to a broad range of applications in areas as diverse as financial modeling, airline scheduling, election modeling, quantum chemistry, physical simulation, automotive design, preventative healthcare, logistics, and more.

Our system uses quantum annealing to solve problems represented as mathematical functions (resembling a landscape of peaks and valleys) by harnessing quantum mechanical effects, including superposition, entanglement, and tunneling, to find their global minima, corresponding to optimal or near optimal solutions. Our device has a footprint of approximately 10’ x 7’ x 10’. Its physical enclosure houses sophisticated cryogenic refrigeration, shielding, and I/O systems to support one quantum processing unit (QPU). For quantum effects to play a role in computation, the QPU requires an isolated environment. The refrigerator and layers of shielding create an internal high-vacuum environment with a temperature close to absolute zero that is isolated from external magnetic fields, vibration, and RF signals.

The QPU itself is built from a network of interconnected superconducting flux qubits. Each qubit is made from a tiny loop of metal interrupted by a Josephson junction. At the low temperatures in our system these loops become superconductors and exhibit quantum mechanical effects. When a qubit is in a quantum state, current flows in both directions simultaneously, which means that the qubit is in superposition - that is, in both a 0 and a 1 state at the same time. At the end of the problem-solving process, this superposition collapses into one of the two classical states, 0 or 1.

Going from a single qubit to a multi-qubit QPU requires that the qubits be interconnected to exchange information. Qubits are connected via couplers, which are also superconducting loops. The interconnection of qubits and couplers, together with control circuitry to manage the magnetic fields, creates an integrated fabric of programmable quantum devices. When the QPU arrives at a solution to a problem, all qubits settle into their final states and the values they hold are returned to the user.

Customers of Amazon Braket will have live, real-time access to D-Wave quantum computers. To optimize the use of our system, AWS customers will use the D-Wave Ocean SDK, a set of Python tools that support problem mapping by translating application objectives into a form suitable for solution on the D-Wave quantum computer, and then return solutions suitable for the original application. The SDK also includes a Uniform Sampler API, an abstraction layer that represents the problem in a form that can be used by the quantum computer, and sampler selection tools that allow the user to direct which of several methods (called “samplers”) to use to solve problems. Methods include running quantum annealing, classical computer hardware running classical algorithms, or potentially custom-designed samplers.

While users can submit problems to the system in a number of different ways, ultimately a problem is represented as a set of values that correspond to the weights of the qubits and the strength of the couplers. Problem solutions correspond to the optimal configuration of qubits found; that is, the lowest points in the energy landscape. These values are returned to the user. Because quantum computers are probabilistic rather than deterministic, multiple values can be returned, representing a set of good, if not optimal, solutions to a problem.

## IonQ

IonQ is a leader in universal quantum computing. We believe the best way to build a quantum computer is by starting with nature: IonQ uses individual atoms as the heart of our quantum processing units. We levitate them in space with semiconductor-defined electrodes on a chip. Then we use lasers for initial preparation, gate operations, and final readout. To bring it all together requires counterintuitive physics, precision optical and mechanical engineering, and fine-grained firmware control over a variety of components. IonQ was founded in 2015 by Jungsang Kim and Christopher Monroe.

The execution of computational tasks on our quantum computer is accomplished by programming the sequence of laser pulses used to implement each quantum gate operation. Our system architecture enables gate operations between an arbitrary set of quantum bits, or qubits, in the system, making it a highly versatile computing machine that can efficiently run a wide range of quantum algorithms. Our system is able to execute a broad range of quantum algorithms designed to tackle problems in chemistry and materials simulation, logistics and optimization, pharmaceutical, and security applications.

IonQ’s trapped-ion approach to quantum computing starts with ionized ytterbium atoms. Two internal states of these identical atoms make up the qubits, the most important part of any quantum computer. Each ytterbium atom is perfectly identical to every other ytterbium atom in the universe. We first strip an electron from the atom to turn our atom into an ion, and use a specialized chip called a linear ion trap to hold it precisely in 3D space. The trap features around 100 tiny electrodes precisely designed, fabricated, and controlled to produce electromagnetic forces that hold our ions in place, isolated from the environment to minimize environmental noise and decoherence.

Once the first ion is in place, we can then load any number of ions into a linear chain. This on-demand reconfigurability allows us to theoretically create anything from a one-qubit system to a 100+ qubit system (not currently available) without having to fabricate a new chip or change the underlying hardware. Once the atoms are trapped, we can prepare them in any quantum state, and they remain in that state indefinitely as long as the qubits are adequately isolated from the environment. Before we can use our ions to perform quantum computations, we have to prepare them for the task. This involves two steps: cooling, to reduce computational noise, and state preparation, which initializes each ion into a well-defined “zero” state, ready for use.

We perform gate operations with an array of individual laser beams, each imaged onto an individual ion, plus one “global” beam. The interference between the two beams produces a control signal that can kick the qubits into a different state. We can manipulate the state of the ions to create single and two-qubit gates. To date, we’ve run single-qubit gates on a 79-ion chain, and complex algorithms consisting of multiple two-qubit gates on chains of up to 11 ions. Once the computation has been performed, reading the result is done by shining a resonant laser on all of the ions to collapse the quantum information to one of two states. Collecting and measuring this light allows us to simultaneously read the collapsed state of every ion — one of these states glows in response to the laser light, the other does not. We interpret the result as a binary string. In order to isolate the atomic ion qubits from the environment, we put the trap inside an ultra-high vacuum chamber, pumped down to pressures of around 10^{-11} Torr. At this pressure, there are fewer molecules in a given volume than in outer space.

## Rigetti

Rigetti Computing builds and deploys integrated quantum computing systems leveraging superconducting qubit technology. These systems enable organizations to augment existing computational workflows with quantum processors. Rigetti serves customers in finance, insurance, pharmaceuticals, defense, and energy with custom software and full-stack solutions focused on simulation, optimization, and machine learning applications. The company is headquartered in California, with offices in Washington, DC, Australia, and the UK.

Rigetti quantum processors are universal, gate-model machines based on superconducting qubits. The Rigetti Aspen series chips feature tileable lattices of alternating fixed-frequency and tunable superconducting qubits within a system architecture that is scalable to large qubit counts. Parametric entangling logic gates on these chips also offer fast gate times and program execution rates. The Aspen series quantum chips are fabricated in Rigetti’s dedicated device foundry using state-of-the-art manufacturing techniques for superconducting circuits. The result is a combination of precision, scale, and speed.

Rigetti processors consist of three principal subsystems. First, user programs are optimized into machine-native instructions through an efficient compiler tool chain. Then, a low-latency hardware controller sequences these instructions as calibrated electrical signals. Finally, qubits that are made from coherent superconducting circuit elements transduce these electrical signals logically as digital quantum gates and measurement instructions.

Universal, gate-based quantum computers enable applications in areas including chemical simulation, combinatorial optimization, and machine learning. Because coherent superconducting qubits operate on the same quantum mechanical principles that govern nature, they can be used to efficiently simulate and understand the behavior of biochemical mechanisms, for example photosynthesis and protein folding. In addition, quantum computers can be used to navigate exponentially more state space to find more optimal solutions among an incalculable set of possibilities, for example in global logistics management. Early application demonstrations on Rigetti processors include the first simulation of an atomic nucleus as well as the largest implementation of a linear system solver on quantum hardware.

The Aspen chip connectivity graph is octagonal with 3-fold (2-fold for edges) connectively. The Rigetti quilc compiler maps an abstract quantum algorithm onto this network of physical connections. SWAP gates shuttle quantum information across the Aspen processor to link non-nearest neighbor qubits. Because these can be costly operations, the quilc compiler is highly optimized for the “layout problem.” Specifically, the compiled depths for programs on Aspen graphs are often estimated to be less than or approximately equal to an all-to-all, serial program construction.

**Figure 1 - Scalable Aspen chip architecture**:

*Distinguishing characteristics of Aspen series chips from Rigetti include direct coupling between one qubit and its three nearest neighbors; entangling gates actuated via frequency modulation (parametric) control; rapid sampling via active register reset.*

Rigetti’s two-qubit entangling gate mechanism (typically a controlled-phase gate or “CZ” gate) is actuated by electrical frequency modulation (FM) control. For these “parametric gates” to execute this instruction, the radio dial of one qubit is flipped into resonance with a nearest-neighbor qubit for a precisely defined duration. On the Aspen graph, this scheme requires that at least half of the qubits are FM-tunable. Within the integrated circuit, through-silicon vias and a flip-chip bonded superconducting shield minimizes spurious control signal crosstalk on the chip. To estimate the effect of decoherence on an algorithm, the lifetimes of the qubits (~25-50 μs) should be compared to the gate duration (~50-200 ns) multiplied by the circuit depth of the algorithm. These times from the latest Rigetti processor (Aspen-8) is shown in Table 1.

Rigetti Aspen-8 |
Median Time Duration (μs) |
---|---|

T1 Lifetime |
29 |

T2 Lifetime |
18 |

Single qubit gate operation |
0.060 |

Two qubit gate operation |
0.144 |

Readout operation |
1.68 |

Register reset operation |
10 |

*Table 1 - Lifetimes and operation speeds for state-of-the-art superconducting qubit processors:** Median values for T1 and T2 and instruction execution times for the Aspen-8 processor at the time of launch.*

Rigetti Aspen-8 |
Median Fidelity (per op.) |
---|---|

Single-qubit gates |
99.79 % |

Two-qubit gates |
95.66 % |

Readout operation |
94.51 % |

Active reset operation |
99.25 % |

*Table 2 - Fidelity per operation for state-of-the-art superconducting qubit processors: **Median values for operation fidelity for the Aspen-8 processor at the time of launch.*

For more information, visit https://www.rigetti.com.