SolvJump Configurator: Examples
Run journeys 01→14 in order. Each page below is a full walkthrough: what the scenario teaches, which APIs matter, the important code pattern, act-by-act narration, and commands to run locally.
Examples source: Journey scripts, study YAMLs, and model modules ship in the SolvJump release bundle under
python/examples/. Each journey has a guided tour (narrated script) and a production CLI command — seePROGRESSION.mdin the bundle. SolvJump never ships onboard — only proven SolvSRK configs from golden runs go to deployment. Beta release. Products in this catalog other than the SolvSRK family and SolvScout / SolvTune are beta — suitable for trials and evaluation; APIs and packaging may change before GA. Do not deploy beta builds in production programs without a signed agreement with Resonix. Activate a trial license before running examples — see Install and Licensing.
What the journeys teach
Simulation engineers run thousands of parameter combinations — battery configs, chemistry rate constants, comm window geometries — before freezing an onboard config. Full time-stepped SolvSRK is the truth oracle but expensive at scale.
SolvJump is the ground-side accelerator. It calibrates a coarse jump propagator from short SolvSRK runs and sweeps parameters when physics gates pass. Speed versus integrating every grid point with SolvSRK is scenario- and gate-dependent: jump, adaptive MFBO, or a full-SRK screen are all valid completions. Ranked finalists are proven with a golden SolvSRK run. SolvJump never ships onboard — only the proven SolvSRK config goes to SolvSRK-Edge.
The python/examples/ bundle is the product tutorial: models, study YAMLs, and guided
journeys 01–14 covering license, YAML, Session, refusal, golden proof, rank verification,
and satellite-scale HPC gating.
Journey index
| # | Journey | What you learn |
|---|---|---|
| 01 | License and first contact | Confirm the wheel, native libsolvjump, machine code, and seat status before running any study. Production wheels are lic… |
| 02 | Scaffold, check-model, dry-run | Scaffold a runnable study with solvjump init, validate the model contract with check-model, and preview grid size wi… |
| 03 | First YAML study | Run a complete parameter sweep from YAML: calibration, jump propagation when gates pass, honest SolvSRK fallback when th… |
| 04 | Python Session and LHS | Drive the same study from Python with Session.from_model, grid sweep(), and method="lhs" for Latin hypercube — the… |
| 05 | Sessions and live what-if | Inspect prior study sessions, read sweep metadata, and call propagate() for fast what-if on a calibrated jump operator… |
| 06 | Proprietary chemistry | Sweep stiff Robertson kinetics with cache control: trajectory-only models (no terminal_metrics) still complete but pro… |
| 07 | Ranked mission envelope | Ranked mission envelope: terminal_metrics turns each trajectory into mission_ok gates and a sortable finalists table… |
| 08 | Adaptive search and evidence | Adaptive search (--strategy adaptive) when jump is untrusted: budgeted full SolvSRK integrations plus evidence artifac… |
| 09 | SolvTune objective | Pass SolvTune objectives (balanced, precision, survival, efficiency) into solvjump run --objective so SRK veri… |
| 10 | When jump refuses | When jump physics gates refuse the operator, SolvJump still completes the study — auto-routing to adaptive MFBO when `te… |
| 11 | Golden proof | Golden proof run: re-integrate the top-ranked design point at full SolvSRK fidelity and seal the receipt — the only conf… |
| 12 | Rank verification | solvjump verify-ranking checks that finalists ordering holds under tightened SRK thresholds — without re-sweeping the … |
| 13 | Scale and HPC gate | Satellite comm window study at scale: quick / standard / full grid tiers, 5-minute ops cadence (not 50 s micro-steps), r… |
| 14 | Advanced Python surface | Advanced Python surface: inspect_model, EvidenceStore, RankSurrogate, MFBOConfig, and mfbo_search for custom o… |
Start with Journey 01 if this is your first install.
Set SOLVSRK_QUIET=1 only when you want silent CLI runs (no narration).
Example bundle
Every journey is a domain scenario with narrated acts — run the guided tour first to see what each number means, then copy the production one-liner into your pipeline.
| Resource | Purpose |
|---|---|
PROGRESSION.md | Core runbook — journeys 01→06 with dual commands |
SKU_PROGRESSION.md | Vertical SKU runbook — Grid, Pricing, Control, Climate, Thermal, Specialty |
COVERAGE.md | Capability matrix — which APIs each journey exercises |
APPLICATIONS.md | Where the product applies in real programs |
run_examples.py | Interactive menu for core journeys |
run_sku_examples.py | Interactive menu for vertical SKU walkthroughs |
Guided journeys
| # | Script | Scenario |
|---|---|---|
| 01 | 01_license_and_status.py | Machine code, seat file, version — before any sweep |
| 02 | 02_init_check_dry_run.py | A runnable model.py + study.yaml without integrating yet |
| 03 | 03_first_study.py | Brusselator example |
| 04 | 04_python_session.py | Same Brusselator physics |
| 05 | 05_sessions_and_whatif.py | Reload a calibrated operator — no new fit |
| 06 | 06_proprietary_chemistry.py | Robertson (S3) |
| 07 | 07_ranked_mission.py | 256-point drone envelope |
| 08 | 08_adaptive_and_evidence.py | Variant: battery |
| 09 | 09_solvtune_objective.py | balanced |
| 10 | 10_when_jump_refuses.py | Brusselator long chain |
| 11 | 11_golden_proof.py | Explore, then prove one finalist with full-horizon SolvSRK |
| 12 | 12_rank_verification.py | Prove jump order vs full SolvSRK — do not re-run the grid |
| 13 | 13_scale_and_hpc.py | Tier: quick |
| 14 | 14_advanced_python.py | inspect_model · EvidenceStore · RankSurrogate · MFBOConfig |
Run from the python/ directory after install and license activation.
Set SOLVJUMP_QUIET=1 only when you want silent CLI runs (no narration).