AWS braket documentation change
Summary
Added documentation for dynamic circuits on IQM devices with mid-circuit measurements and feed-forward operations, updated constraints/limitations formatting, and expanded experimental capability descriptions
Security assessment
Changes focus on new quantum computing capabilities (dynamic circuits) and formatting improvements. No security vulnerabilities, mitigations, or security features are mentioned. Constraints relate to technical limitations rather than security controls.
Diff
diff --git a/braket/latest/developerguide/braket-experimental-capabilities.md b/braket/latest/developerguide/braket-experimental-capabilities.md index b7962114a..256fc27d8 100644 --- a//braket/latest/developerguide/braket-experimental-capabilities.md +++ b//braket/latest/developerguide/braket-experimental-capabilities.md @@ -5 +5 @@ -Access to local detuning on QuEra AquilaAccess to tall geometries on QuEra AquilaAccess to tight geometries on QuEra Aquila +Access to local detuning on QuEra AquilaAccess to tall geometries on QuEra AquilaAccess to tight geometries on QuEra AquilaAccess to dynamic circuits on IQM devices @@ -9 +9 @@ Access to local detuning on QuEra AquilaAccess to tall geometries on QuEra Aquil -To advance your research workloads, it is important to get access to new innovative capabilities. With Braket Direct, you can request access to available experimental capabilities, such as new quantum devices with limited availability, directly in the Braket console. +To advance your research workloads, it is important to get access to new innovative capabilities. With Braket Direct, you can request access to available experimental capabilities, such as new quantum devices with limited availability or software features, directly in the Braket console. @@ -29,0 +30,2 @@ To advance your research workloads, it is important to get access to new innovat + * Access to dynamic circuits on IQM devices + @@ -37 +39,5 @@ Local detuning (LD) is a new, time-dependent control field with a customizable s -**Constraints:** The spatial pattern of the local detuning field is customizable for each AHS program, but it is constant over the course of a program. The time series of the local detuning field must start and end at zero with all values being less than or equal to zero. Additionally, the parameters of the local detuning field are limited by numerical constraints, which can be viewed through the Braket SDK in the specific device properties section \- `aquila_device.properties.paradigm.rydberg.rydbergLocal`. +**Constraints:** + +The spatial pattern of the local detuning field is customizable for each AHS program, but it is constant over the course of a program. The time series of the local detuning field must start and end at zero with all values being less than or equal to zero. Additionally, the parameters of the local detuning field are limited by numerical constraints, which can be viewed through the Braket SDK in the specific device properties section \- `aquila_device.properties.paradigm.rydberg.rydbergLocal`. + +**Limitations:** @@ -39 +45 @@ Local detuning (LD) is a new, time-dependent control field with a customizable s -**Limitations:** When running quantum programs that use the local detuning field (even if its magnitude is set to constant zero in the Hamiltonian), the device experiences faster decoherence than the T2 time listed in the performance section of Aquila’s properties. When unnecessary, it is best practice to omit the local detuning field from the Hamiltonian of the AHS program. +When running quantum programs that use the local detuning field (even if its magnitude is set to constant zero in the Hamiltonian), the device experiences faster decoherence than the T2 time listed in the performance section of Aquila’s properties. When unnecessary, it is best practice to omit the local detuning field from the Hamiltonian of the AHS program. @@ -45 +51 @@ Local detuning (LD) is a new, time-dependent control field with a customizable s - 1. **Simulating the effect of non-uniform longitudinal magnetic field in spin systems.** + 1. **Simulating the effect of non-uniform longitudinal magnetic field in spin systems** @@ -49 +55 @@ While the amplitude and phase of the driving field have the same effect on the q - 2. **Preparing non-equilibrium initial states.** + 2. **Preparing non-equilibrium initial states** @@ -53 +59 @@ The example notebook [Simulating lattice gauge theory with Rydberg atoms](https: - 3. **Solving weighted optimization problems.** + 3. **Solving weighted optimization problems** @@ -64 +70 @@ The tall geometries feature allows you to specify geometries with increased heig -**Constraints:** The max height for tall geometries is 0.000128 m (128 um). +**Constraints:** @@ -66 +72,5 @@ The tall geometries feature allows you to specify geometries with increased heig -**Limitations:** When this experimental capability is enabled for your account, the capabilities shown on the device properties page and the `GetDevice` call will continue to reflect the regular, lower limit on the height. When an AHS program uses atom arrangements that go beyond the regular capabilities, the filling error is expected to increase. You will find an elevated number of unexpected 0s in the `pre_sequence` part of the task result, in turn, lowering the chance to get a perfectly initialized arrangement. This effect is strongest in rows with many atoms. +The max height for tall geometries is 0.000128 m (128 um). + +**Limitations:** + +When this experimental capability is enabled for your account, the capabilities shown on the device properties page and the `GetDevice` call will continue to reflect the regular, lower limit on the height. When an AHS program uses atom arrangements that go beyond the regular capabilities, the filling error is expected to increase. You will find an elevated number of unexpected 0s in the `pre_sequence` part of the task result, in turn, lowering the chance to get a perfectly initialized arrangement. This effect is strongest in rows with many atoms. @@ -72 +82 @@ The tall geometries feature allows you to specify geometries with increased heig - 1. **Bigger 1d and quasi-1d arrangements.** + 1. **Bigger 1d and quasi-1d arrangements** @@ -76 +86 @@ Atom chains and ladder-like arrangements can be extended to higher atom numbers. - 2. **More room for multiplexing the execution of tasks with small geometries.** + 2. **More room for multiplexing the execution of tasks with small geometries** @@ -87 +97,3 @@ The tight geometries feature allows you to specify geometries with shorter spaci -**Constraints:** The minimal row spacing for tight geometries is 0.000002 m (2 um). +**Constraints:** + +The minimal row spacing for tight geometries is 0.000002 m (2 um). @@ -89 +101,3 @@ The tight geometries feature allows you to specify geometries with shorter spaci -**Limitations:** When this experimental capability is enabled for your account, the capabilities shown on the device properties page and the `GetDevice` call will continue to reflect the regular, lower limit on the height. When an AHS program uses atom arrangements that go beyond the regular capabilities, the filling error is expected to increase. Customers will find an elevated number of unexpected 0s in the `pre_sequence` part of the task result, in turn, lowering the chance to get a perfectly initialized arrangement. This effect is strongest in rows with many atoms. +**Limitations:** + +When this experimental capability is enabled for your account, the capabilities shown on the device properties page and the `GetDevice` call will continue to reflect the regular, lower limit on the height. When an AHS program uses atom arrangements that go beyond the regular capabilities, the filling error is expected to increase. Customers will find an elevated number of unexpected 0s in the `pre_sequence` part of the task result, in turn, lowering the chance to get a perfectly initialized arrangement. This effect is strongest in rows with many atoms. @@ -95 +109 @@ The tight geometries feature allows you to specify geometries with shorter spaci - 1. **Non-rectangular lattices with small lattice constants.** + 1. **Non-rectangular lattices with small lattice constants** @@ -99 +113 @@ Tighter row spacing allows the creation of lattices where the closest neighbor t - 2. **Tunable family of lattices.** + 2. **Tunable family of lattices** @@ -105,0 +120,61 @@ In AHS programs, interactions are tuned by adjusting the distance between pairs +## Access to dynamic circuits on IQM devices + +Dynamic circuits on IQM devices enable mid-circuit measurements (MCM) and feed-forward operations. These features allow quantum researchers and developers to implement advanced quantum algorithms with conditional logic and qubit reuse capabilities. This experimental feature helps explore quantum algorithms with improved resource efficiency and study quantum error mitigation and error correction schemes. + +**Key instructions:** + + * `measure_ff`: Implements measurement for feed-forward control, measuring a qubit and storing the result with a feedback key. + + * `cc_prx`: Implements a classically-controlled rotation that applies only when the result associated with the given feedback key measures a |1⟩ state. + + + + +Amazon Braket supports dynamic circuits through OpenQASM, the Amazon Braket SDK, and the Amazon Braket Qiskit Provider. + +**Constraints:** + + 1. Feedback keys in the `measure_ff` instructions must be unique. + + 2. A `cc_prx` must happen after `measure_ff` with the same feedback key. + + 3. In a single circuit, the feed-forward on a qubit can only be controlled by one qubit, either by itself or by another qubit. In different circuits, you can have different pairs of control. + + 1. For example, if qubit 1 is controlled by qubit 2, it cannot be controlled by qubit 3 in the same circuit. There is no constraint on how many times the control is applied between qubit 1 and qubit 2. Qubit 2 can be controlled by qubit 3 (or qubit 1), unless an active reset was performed on qubit 2. + + 4. Control can only be applied to qubits within the same group. + + 5. Programs with these capabilities must be submitted as verbatim programs. To learn more about verbatim programs, see [Verbatim compilation with OpenQASM 3.0](https://docs.aws.amazon.com/braket/latest/developerguide/braket-openqasm-verbatim-compilation.html). + + + + +**Limitations:** + +Currently, MCM can only be use for feed-forward control in a program. The MCM outcomes (0 or 1) are not returned as part of a task result. + + + +**Examples:** + + 1. **Qubit reuse through active reset** + +MCM with conditional reset operations enable qubit reuse within a single circuit execution. This reduces circuit depth requirements and improves quantum device resource utilization. + + 2. **Active bit flip protection** + +Dynamic circuits detect bit flip errors and apply corrective operations based on measurement outcomes. This implementation serves as a quantum error detection experiment. + + 3. **Teleportation experiments** + +State teleportation transfers qubit states using local quantum operations and classical information from MCMs. Gate teleportation implements gates between qubits without direct quantum operations. These experiments demonstrate foundational subroutines in three key areas: quantum error correction, measurement-based quantum computing, and quantum communication. + + 4. **Open quantum systems simulation** + +Dynamic circuits model noise in quantum systems through data qubit and environment entanglement, and environmental measurements. This approach uses specific qubits to represent data and environment elements. A Noise channel can be designed by the gates and measurements applied on the environment. + + + + +For more information on using dynamic circuits, see additional examples in the [Amazon Braket notebook repository](https://github.com/amazon-braket/amazon-braket-examples/tree/main/examples/experimental_capabilities/dynamic_circuits). +