Skip to content

Releases

v1.0.0: Initial Production Release and PyPI Distribution

Choose a tag to compare

@iago-fernandez iago-fernandez released this 15 Nov 13:15
92b22a2

This release marks the first stable milestone of the Pyomo Conic Engine. It establishes a lightweight, high-performance extension for native Semidefinite Programming (SDP) and Second-Order Cone Programming (SOCP) modeling in Pyomo, coupled with automated Conic Benchmark Format (CBF) solver export and official distribution on PyPI.

Conic Modeling Engine

• Semidefinite Programming (SDP): Implementation of Linear Matrix Inequality (LMI) abstractions supporting affine matrix decompositions into constant ($D$) and variable ($H_t$) coefficient matrices without auxiliary slack variables.
• Second-Order Cone Programming (SOCP): Native modeling of Lorentz cones ($\mathcal{Q}^d$) in both component-wise norm inequality format and vector membership representations.
• Linear-Time Symbolic Traversal: Integration with Pyomo's generate_standard_repn to parse affine expressions in linear time with $O(N)$ symbolic complexity and zero duplicate AST allocations.

Solver Integration

• Conic Benchmark Format (CBF v4): Automatic generation of synchronized CBF files supporting PSDCON, HCOORD, DCOORD, CON ($Q$), ACOORD, and BCOORD sections compatible with RAPOSa, MOSEK, and Clarabel.
• Synchronized Problem Triplet: Coordinated generation of AMPL .nl bytecode, .col column variable index maps, and .cbf problem files, ensuring seamless variable coordinate alignment across solvers.
• Selective File Serialization: Exporter configuration flags (export_nl, export_col, tolerance, verbose) for customized solver pipeline targets and headless benchmarking.

Packaging and CI/CD

• Official PyPI Distribution: Configuration for deployment to the Python Package Index as pyomo-conic with PEP 621 metadata and standard setuptools.build_meta.
• Automated Multi-Version CI: Continuous integration matrix validating builds and test suites across Python 3.12, Python 3.13, and Python 3.14 on ubuntu-latest.
• OIDC Trusted Publishing: Cryptographically verified deployment workflow using OpenID Connect attestation directly from GitHub Actions without long-lived API tokens.

This release represents the stable package published to PyPI and verified for research and production environments.