Python package¶

The ibm-qdmi distribution provides the ibm.qdmi namespace and bundles the native library, device catalogue, headers, and CMake configuration. Follow installation to install from a checkout. Importing the package does not load a device or contact IBM.

Installed paths¶

Use the exported pathlib.Path constants to locate installed files:

Constant	Installed resource
`IBM_QDMI_LIBRARY_PATH`	Native shared library
`IBM_QDMI_CATALOG_PATH`	QDMI device catalogue
`IBM_QDMI_INCLUDE_DIR`	Public headers
`IBM_QDMI_CMAKE_DIR`	CMake package configuration

The package also exports IBM_QDMI_DEVICE_ID, IBM_QDMI_PREFIX, and __version__. The generated Python API documents these exports and the optional framework adapters.

from ibm.qdmi import IBM_QDMI_CATALOG_PATH, IBM_QDMI_LIBRARY_PATH

print(IBM_QDMI_LIBRARY_PATH)
print(IBM_QDMI_CATALOG_PATH)

Command-line interface¶

Inspect the installed package without opening a session:

ibm-qdmi --version
ibm-qdmi --lib_path
ibm-qdmi --catalog_path
ibm-qdmi --include_dir
ibm-qdmi --cmake_dir

python -m ibm.qdmi accepts the same options. See device discovery for catalogue entries and MQT Core configuration.

Framework integrations¶

The optional Qiskit integration supports transpilation, sampling, and energy estimation. The PennyLane integration supports finite-shot QNodes and gradients. Both use MQT Core’s shared QDMI adapters. Start with the examples to run workloads on a local simulator before selecting IBM hardware.