qiskit-hardware

Qiskit - Real Hardware & Pulse Control

Moving from simulators to real hardware requires a shift in mindset. You are no longer working with perfect "ideal" qubits, but with superconducting circuits that suffer from decoherence, readout errors, and crosstalk. This guide covers how to get the most science out of noisy intermediate-scale quantum (NISQ) devices.

When to Use

• Executing quantum algorithms on real IBM Quantum backends.
• Characterizing hardware noise (T1, T2 relaxation times).
• Implementing Error Mitigation to improve result accuracy.
• Using Qiskit Pulse to define custom microwave pulses (OpenPulse).
• Optimizing circuits for specific hardware topologies (coupling maps).
• Benchmarking quantum advantage in real-world conditions.

Reference Documentation

• IBM Quantum Learning: https://learning.quantum.ibm.com/
• Qiskit Runtime Docs: https://docs.quantum.ibm.com/run
• Qiskit Pulse Guide: https://docs.quantum.ibm.com/build/pulse
• Search patterns: QiskitRuntimeService, Sampler, Estimator, transpile, InstructionScheduleMap

Core Principles

1. Qiskit Runtime (Primitives)

The modern way to interact with hardware. Instead of sending raw circuits, you use Primitives:

• Sampler: Returns quasi-probabilities (bitstrings).
• Estimator: Returns expectation values of observables (e.g., energy).

2. Transpilation (Hardware Mapping)

Physical backends only support a small set of "Basis Gates" (e.g., rz, x, sx, ecr). The Transpiler rewrites your abstract math into these specific instructions and maps virtual qubits to physical ones based on error rates.

3. Error Mitigation

Unlike Error Correction (which requires thousands of qubits), Mitigation uses statistical tricks (like TREX or ZNE) to "clean" the results after execution.

Quick Reference: Connecting to Hardware

Installation

pip install qiskit-ibm-runtime

Setup and Job Execution

from qiskit_ibm_runtime import QiskitRuntimeService, SamplerV2 as Sampler

# 1. Initialize service (requires API key from quantum.ibm.com)
# service = QiskitRuntimeService(channel="ibm_quantum", token="YOUR_TOKEN")
service = QiskitRuntimeService() # Uses saved credentials

# 2. Select backend
backend = service.least_busy(operational=True, simulator=False)

# 3. Run job using Primitives
sampler = Sampler(backend=backend)
job = sampler.run([my_circuit])
result = job.result()