Release 1.1.0¶
MicroQuantum 1.1.0 is the first minor release after General Availability. It is strictly additive over 1.0.0: no public API was renamed or removed, and every 1.0.0 program keeps working unchanged.
Status¶
Version: 1.1.0.
Runtime: Pure Python + NumPy. Python 3.10 - 3.13.
Platforms: designed for Windows, Linux, macOS and BSD; CI-verified on GitHub-hosted Ubuntu Linux, Windows and macOS. See Platform Support for the per-platform status.
Dependency: NumPy is the only hard runtime dependency.
Documentation: https://ajit-ai.github.io/microquantum/
License: MIT.
What 1.1.0 adds over 1.0.0¶
Core engine expansion —
microquantum.coregrows 17 focused subpackages (gates, operators, observables, states, channels, measurements, parameters, tensor, information, execution, architecture, resources, gradients, serialization, transpiler, plus circuit/pauli/registers packages): gate hierarchies, first-class observables/states/channels, a backend-independent execution model, hardware-independent architecture descriptions and a pass-based transpiler framework. See Core Expansion.SDK extension surface (Phase 121) — contracts (
ConstrainedOptimizationProblem,TimeEvolutionProblem,ExcitedStateProblem,AdapterRegistry,Budget), IR control nodes (Loop/Switch) with cost models, execution scheduling and caching, quantum counting, PUBO support, a data-reuploading classifier withfit, the Steane code, Jordan-Wigner transforms, CDR mitigation, provider polling/proxy support, mirror benchmarking, experiment checkpoints and stdlib factories. See Phase 121 — SDK Extension Surface.Correctness hardening (Phase 122) — fixed amplitude-estimation QPE accuracy, single-sourced SDK version reporting, strict search typing, offline provider replay fixtures, scheduled/budgeted batch execution, tensor-network depth coverage and hardware-aware transpiler passes (SWAP routing, noise-aware layout, commutation-aware cancellation). See Phase 122 — Correctness Hardening & Hardware Readiness.
Stability¶
The 1.0.0 SemVer commitments from General Availability (v1.0.0) still hold: backward-compatible additions land in minor releases, serialized schemas stay versioned with additive-only fields, and the 368-name public gateway is covered by contract tests.
Upgrading from 1.0.0¶
pip install --upgrade microquantum — no code changes required.
One intentional behavior note: optimizer RNG streams were never part
of the compatibility contract and may differ where new
resampling/blocking options are enabled (defaults preserve prior
behavior).