Improve quantum workload performance with Q-CTRL’s Fire Opal error suppression for IonQ processors on HAQM Braket

This post was contributed by Zia Mohammad, Senior Product Manager at AWS, Rowen Wu, Staff Product Manager at Q-CTRL, Dr. Pranav Mundada, Lead Scientist at Q-CTRL, Ayush Pancholy, scientist at Q-CTRL, Alex Shih, VP of Product at Q-CTRL, Tim Rogers, Quantum Solutions Engineer at IonQ, and Coleman Collins, Senior Director of Product Management at IonQ 

Available today, customers can leverage Q-CTRL’s Fire Opal error suppression technology with IonQ quantum processing units (QPUs) available through HAQM Braket. This integration allows you to improve the performance of your quantum workloads on IonQ devices such as Aria and Forte with just a few lines of code, helping you achieve increased accuracy for your quantum computing workloads.

HAQM Braket is a quantum computing service from AWS that gives you access to different types of quantum computers, including quantum circuit simulators and quantum processing units (QPUs). With Braket, you can explore and experiment with quantum computing by building and testing quantum circuits on simulated and physical quantum computers such as IonQ’s trapped-ion QPUs, which feature up to 36 qubits with long coherence times. With Fire Opal’s error suppression pipeline, you can run deeper and wider circuits with improved fidelity on these devices.

In this post, we’ll show you how Fire Opal improves performance on IonQ hardware, demonstrate benchmark results across various quantum algorithms, and provide step-by-step instructions to get started on HAQM Braket. Over the next few months, Q-CTRL aims to expand on these results with methods like dynamical decoupling and artificial intelligence (AI) driven gate optimization.

How Fire Opal enhances IonQ performance on HAQM Braket

Fire Opal is a performance management software that applies error suppression techniques to quantum circuits before they run on hardware. Unlike post-processing error mitigation methods, Fire Opal modifies circuit instructions to address errors at both gate and circuit levels.

HAQM Braket’s integration with Fire Opal unlocks new possibilities for customers’ quantum applications by achieving higher accuracy while running deeper and wider circuits. This is done by applying error reduction methods across a broad range of use cases to accelerate real-world applications without additional quantum runtime overhead.

Benchmark results show significant performance improvements

Q-CTRL demonstrates these capabilities by running application-oriented benchmarks used to test the performance expected for select use cases on IonQ hardware. The results, seen in Table 1, show performance on application-oriented benchmarks adapted from the QED-C repository, using standardized tests of the technology. In these experiments, the default fidelity is obtained when Qiskit circuits with the standard gateset (CX, SX, RZ, X) are used while the Fire Opal error reduction factor is defined with the following formula:

Table 1 – Benchmark results comparing Fire Opal performance to default implementation on IonQ Forte. The table shows significant error reduction across three algorithms (Bernstein-Vazirani, QFT, and QPE) at various qubit counts, with Fire Opal consistently delivering higher fidelity. The error reduction factor ranges from 1.06x to 2.43x depending on the algorithm and qubit count. Further information on Q-CTRL’s benchmarks is seen on their technical documentation.

Bernstein-Vazirani algorithm

The Bernstein-Vazirani algorithm finds a hidden binary string using a single query to a quantum oracle. As a relatively shallow circuit, Fire Opal provides performance improvements, particularly as the qubit count increases. At 30 qubits on IonQ Forte, the error was reduced 2.43 times, with fidelity reaching 93%.

Quantum Fourier Transform (QFT)

The Quantum Fourier Transform transforms quantum states into their frequency domain faster than classical methods. QFT is more complex than Bernstein-Vazirani, with gate count increasing with qubit count. With this complexity, Fire Opal maintained a fidelity of 61% at 30 qubits on IonQ Forte.

Quantum Phase Estimation (QPE)

Quantum Phase Estimation estimates the phase (eigenvalue) of a unitary operator and is a common building block for many quantum algorithms. It requires a deep circuit. With Fire Opal, QPE success probability reached 68% at 30 qubits on IonQ Forte, with error reduced up to 1.4 times over the default implementation.

Complementing IonQ’s error mitigation toolkit

Fire Opal’s error suppression complements the existing error mitigation tools available on IonQ devices through HAQM Braket—debiasing and sharpening. Debiasing is a method that corrects errors by running the same calculation in different ways—different variants of the physical qubit mapping, gate decomposition, and similar—and then aggregating the results in a way that cancels out certain kinds of systematic error on the device.

Once you have all of the variants, debiasing uses one of two aggregation methods to combine and post-process the results: component-wise averaging and sharpening. Knowing which aggregation method to apply requires understanding in detail the type of output distribution that you’re expecting.

Averaging works best for scenarios where most output states have non-zero probabilities. However, when output distributions are meant to have high contrast, meaning that there are high peaks and low troughs, averaging tends to smooth out the distribution, which makes it harder to distinguish the high-probability outputs from noise.

Sharpening more directly increases the probabilities of the most common outputs. This method is effective for quantum algorithms with a small number of highly probable outputs such as phase estimation or amplitude estimation. For algorithms with complex or non-peaked output distributions, this can provide suboptimal results, which means you have to know what kinds of outputs you’re expecting and tune the approach on a per-algorithm basis.

Fire Opal’s error-suppression pipeline modifies the circuit instructions before the circuit is run on the device, proactively preventing the most likely errors, from the gate level to the circuit level. Because they don’t depend at all on the algorithm’s outputs, Fire Opal’s methods can be applied broadly to any type of algorithm. Because applying error suppression requires no aggregation of results, it comes at no additional cost.

Getting started with Fire Opal on HAQM Braket

You can start using Fire Opal with IonQ hardware on HAQM Braket in just four steps:

1. Set up your HAQM Braket and Fire Opal accounts

Figure 1: Screenshot showing the HAQM Braket console notebook setup page.

Sign into the AWS Management Console and navigate to the HAQM Braket console page. If it is your first time accessing HAQM Braket, refer to the Getting Started documentation to provision your first Jupyter Notebook Instance.

Once your Jupyter notebook instance has been created, create a blank notebook and run the following commands:

%pip install fire-opal
import fireopal as fo
%pip install qiskit-braket-provider

Head over to the FireOpal website and sign up for an account. Note down your Q-CTRL organizational ID, which will be used as part of step 2. Follow the setup instructions to install Fire Opal within the Jupyter Notebook you created with HAQM Braket.

2. Create a cross-account Identity Access Management (IAM) role

Figure 2: Screenshot of the AWS IAM landing page showing you how to create a new role. This will be used to link your FireOpal and AWS accounts.

In a new tab, head to the AWS Management Console to create a cross-account IAM role to allow Fire Opal to use Braket on your behalf. The create role page can be directly accessed via the following link: IAM create role page.

1. For Trusted entity type, select AWS account.
2. Select Another AWS account and type in 522814692014.
3. Select Require external ID and type in your Q-CTRL organization id, which you can retrieve from the Q-CTRL accounts portal. Click Next.
4. For Add permissions, check the HAQMBraketFullAccess Click Next.
5. For Role details, enter an appropriate role name and keep note of it as you will use it later. Click Create role.

3. Configure your organization and generate credentials

Figure 3: Q-CTRL accounts portal showing your organization details setup page.

Retrieve your Fire Opal organization slug from the Q-CTRL accounts portal and configure it to your organization in Fire Opal. Fire Opal will use your organization id an External ID to assume the cross-account AWS role.  One additional command will generate the credentials dictionary needed to execute jobs.

#Configure Fire Opal and create Braket credentials
fo.config.configure_organization("org_slug")
arn = "your_role_arn"
credentials = fo.credentials.make_credentials_for_braket(arn=arn)

4. Run circuits with your chosen IonQ backend

Create your quantum circuit and execute it through Fire Opal, selecting IonQ Forte, IonQ Aria 1, or IonQ Aria 2 as your backend. For example, you can create a basic circuit with a Hadamard and a CNOT gate.

from qiskit import QuantumCircuit

# Create your quantum circuit
qc = QuantumCircuit(2)
qc.h(0)
qc.cx(0, 1)
qc.measure_all()
qc.draw("mpl")
# Convert the circuit to QASM string
circuit_qasm = qc.qasm()

Fire Opal will automatically apply its AI-driven error suppression techniques to optimize performance without requiring any additional configuration.

fire_opal_job = fo.execute(
    circuits=[circuit_qasm],
    shot_count=1024,
    credentials=credentials,
    backend_name="Forte-1"

# Once the workload has finished you can retrieve the optimized result
fire_opal_result = fire_opal_job.result()
bitstring_results = fire_opal_result["results"]

Accelerating quantum optimization with Fire Opal’s Solver

For customers looking to solve optimization problems without dealing with circuit-level complexity, Q-CTRL offers a fully packaged hybrid Optimization Solver through private preview. This solver accepts high-level problem definitions and leverages Fire Opal’s core performance management under the hood.

Accenture Federal Services has already demonstrated the value of this approach, using Fire Opal’s Optimization Solver to implement a network anomaly detection application on IonQ hardware that achieved accuracy three times greater than a classical heuristics-based solution.

“We have seen firsthand how Fire Opal enhances the performance of today’s quantum hardware, delivering measurable improvements in anomaly detection and large-scale optimization challenges. As Fire Opal expands its compatibility with devices like IonQ through HAQM Braket, it opens the door for broader adoption of Q-CTRL’s error suppression techniques. This integration helps accelerate the practical application of quantum solutions and empowers researchers and developers at AFS to achieve more accurate and reliable results in critical areas like cybersecurity and optimization.”

— Raymond Beecham, Security Delivery Specialist at Accenture Federal Services

Conclusion

The integration of Fire Opal with IonQ devices on HAQM Braket provides additional tools for quantum computing users. By combining IonQ’s trapped-ion hardware with Fire Opal’s AI-driven error suppression, you can improve the performance of your quantum workloads.

To get started, sign up for Fire Opal and follow the steps outlined in this blog post. For more information about HAQM Braket and quantum computing on AWS, visit the HAQM Braket homepage.