Quantum Computing News

Qiskit SDK v2.2 Propels Quantum-Centric Supercomputing with Qiskit C API Advancements

The Qiskit SDK v2.2, which was released by the Qiskit Team and IBM Quantum Research, offers important features that facilitate smooth quantum-centric supercomputing (QCSC) operations in high-performance computing (HPC) settings. This minor version provides a crucial component for implementing end-to-end quantum-centric supercomputing applications written natively in compiled languages like C++. It is distinguished by significant performance enhancements and long-awaited features. By the end of 2026, developers anticipate that QCSC will allow for the first quantum advantage demonstrations.

You can also read Introduction To Loop Quantum Gravity & String Theory Vs LQG

Qiskit C API Unlocks End-to-End Hybrid Workflows

To fully utilize QCSC, the most important upgrade in Qiskit SDK v2.2 focusses on Qiskit’s C API, which is its first compiled language interface. Up until now, the C API has made it possible to create a complete end-to-end quantum + HPC workflow with Qiskit versions that only support compiled languages.

A stand-alone transpiler function that can be called directly from C is introduced in Qiskit v2.2. By combining this new transpiler function with the C API’s pre-existing support for circuits (added in v2.1) and observables (introduced in v2.0), users can create fully functional QCSC applications. The four standard steps of a quantum computing workflow mapping, optimization, quantum circuit execution, and post-processing on traditional HPC infrastructure can all be covered by these applications.

The C API’s broad compatibility with other programming languages is one of its main advantages; it enables programmers to create wrappers for other compiled languages, such as C++ and Fortran, the two most used languages in contemporary HPC environments.

Showcasing Hybrid Power: The SQD Workflow Demo

IBM Quantum unveiled a new quantum + HPC workflow demo to highlight the capability of creating end-to-end quantum + HPC workflows natively in compiled languages. This demonstration offers working code and detailed instructions for executing an actual QCSC process using the Sample-based Quantum Diagonalization (SQD) technique. One interesting option for demonstrating quantum advantage in the near future is SQD.

Using MPI (Message Passing Interface), the sample integrates data preparation, circuit execution, and parallel classical post-processing on HPC equipment to perform the entire SQD workflow as a single application. This HPC-ready solution makes use of MPI and OpenMP, two standardised parallel computing frameworks. In order to estimate the ground state energy of Fe₄S₄, a molecular cluster present in biological beings, the C API demo shows how to post-process noisy quantum samples.

The new HPC-ready SQD addon, which replicates the main features of the original Python-based SQD addon in a compiled language (C++), is a crucial part of the demonstration. Written as a C++17 template library, this new tool is intended for scalable HPC cluster execution. Users can compile their parallel QCSC program into a single binary executable that is designed to function with standardised parallel job launching commands like mpiexec or mpirun by writing it in a compiled language like C++.

Qiskit v2.2 included a QkTranspileLayout object to store qubit mappings and permutations generated by the transpiler, further supporting these native built workflows. To apply these resultant qubit layouts to an observable (QkObs), a new function was also added qk_obs_apply_layout().

You can also read Quantum Key Distribution Protocols: BB84, E91, And B92

Advancing Toward Fault Tolerance

New capabilities targeted at the long-term objective of fault-tolerant quantum computing are also included in Qiskit v2.2. A particular transformation that is helpful in compilation for some fault-tolerant architectures is implemented by the new transpilation pass, LitinskiTransformation,. This pass transforms input circuits with single-qubit RZ-rotation and Clifford gates into a circuit with Clifford gates after multi-qubit Pauli rotations (executed as PauliEvolutionGate gates).

Additionally, new ancilla-free synthesis techniques for Adders and MCX gates are introduced in Qiskit v2.2. Compared to earlier approaches that were tailored for continuous, near-term gate sets, these new techniques are more effective for fault-tolerant architectures. The HighLevelSynthesis pass, which uses these synthesis techniques, now has an optimization_metric keyword that lets users target 2-qubit gate count minimization (for short-term goals) or T count minimization (for fault tolerance).

Expanded Hardware Modeling and Performance Boost

An enhanced Target model in the most recent SDK release allows backends to specify parameter limitations in far greater depth. Importantly, Qiskit can now realistically represent hardware limits like fractional gates found on systems like IBM Quantum Heron since the Target now supports bounds on gate parameters.

By converting out-of-range angles into comparable, hardware-valid sequences, the new WrapAngles transpiler pass automatically enforces these constraints. Furthermore, a seconds_to_dt() function has been added to the Target model to translate physical seconds into device time steps. This function streamlines device time handling and facilitates the integration of calibration-informed durations with scheduling passes.

In addition to new capabilities, Qiskit v2.2 offers quantifiable circuit translation performance gains, with an average speedup of 10–20% throughout the whole benchmark suite. According to preliminary findings, the continuous conversion of Qiskit code to Rust may be the cause of this long-term efficiency increase.

Upgrade Considerations

Users planning to adopt Qiskit v2.2 should be aware of several platform changes:

• The minimum supported Rust version has increased from 1.79 to 1.85.

• Qiskit SDK v2.2 is the final release to support Python 3.9, as it has reached end of life; Qiskit v2.3 will require Python 3.10 or higher.

• Support for Intel Macs (x86-64) running macOS will be downgraded starting with Qiskit v2.3, driven by the manufacturer’s withdrawal of support and subsequent deprecation of Intel Mac CI runners.

• Several classes in the circuit library, such as PhaseOracle and QuantumVolume, have been deprecated in Qiskit v2.2 and will be removed in Qiskit 3.0, replaced by modern gate equivalents or Rust-accelerated circuit-constructor functions.

You can also read What Is QMA? Quantum Proof System, QMA Complete Problems