AXB Product Introduction
Sparse linear-system analysis, solver recommendation, execution, and reporting in one workbench.
AXB is a desktop workbench for Ax=b engineering workloads built around matrix intake, structural analysis, bottleneck diagnosis, backend recommendation, execution control, and report export. The product helps teams move from a raw matrix file to an informed solver run instead of guessing a backend by hand.
It fits work where sparse linear systems need repeatable profiling, plan selection, controlled execution, and comparable reporting across candidate solver paths.
Matrix intake and structural profiling
Load the target linear system and inspect matrix structure before any backend is chosen.
- Matrix A import
- Optional right-hand side import
- Feature extraction for solver planning
Bottleneck diagnosis and plan recommendation
Turn matrix characteristics into candidate solving plans that reflect backend suitability and runtime constraints.
- Risk and constraint diagnosis
- Candidate backend comparison
- Recommended plan list
Controlled execution and monitoring
Launch the selected plan and keep run status visible while the solve is in progress.
- Stage status and elapsed time
- Iteration and residual tracking for supported paths
- Single-machine CPU delivery surface
Verification, export, and local AI calling
Review the run result, export the report package, and reuse the same local workbench through the packaged MCP bridge when needed.
AXB workflow surface
AXB organizes sparse linear-system work around one flow: import, profile, diagnose, recommend, run, verify, and export.
Load matrix A and the matching right-hand side when the case includes it.
Inspect matrix features, review bottlenecks, and read the recommended backend list.
Launch the chosen plan and monitor stage status, elapsed time, and supported solver progress indicators.
Review the solve result, export the report package, and keep the case available for the next run or local tool call.