a56fcbedc2
First working nonlinear reach artifact for the PWR model. TMJets
Taylor-model scheme on the full 10-state closed-loop (unsaturated
ctrl_heatup, ramp reference via augmented time state x[11]).
Status:
T=10s : SUCCESS. 10583 reach-sets. T_c envelope [274.45, 295] C,
n envelope [-5.2e-4, 5.01e-3]. Over-approximation visible
(n can't be negative physically) but tube is sound and
bounded.
T=60s+ : FAILED. Exhausts 50k step budget then hits NaN in
precursor-decay term.
Root cause: prompt-neutron stiffness. Lambda=1e-4s forces TMJets'
adaptive stepper to ~1ms steps to resolve fast dynamics. 10583 steps
for 10s of sim time means we get ~10s/50000 = 2s horizon max before
step budget exhausts — inadequate for heatup's 5-hour obligation.
Remedy (next session): singular-perturbation reduction of the
neutronics. Treat n as quasi-static algebraic function of (T, C, rho)
rather than a dynamic state. Replaces stiff dn/dt with algebraic
relation, removes fast timescale from reach problem. Standard in
reactor-kinetics reach literature.
What this does prove:
- Julia/TMJets framework works for this system (previous
scaling-issue failure is gone with @taylorize'd RHS).
- Bilinear n*rho term handled correctly by Taylor models.
- Ramp reference via augmented time state x[11] is a workable
pattern for time-varying controller references in reach.
What this does NOT prove:
- Anything about heatup safety — 10s horizon is nowhere near the
mode's 5-hour obligation.
Includes sim_heatup.jl, a Rodas5 baseline using the same @taylorize-
able RHS form, for cross-validation of the reach tube against a
nominal trajectory once longer horizons are reachable.
WALKTHROUGH.md updated with the finding.
Hacker-Split: got partway up the Julia reach ladder, identified the
physical bottleneck (stiffness), named the fix (reduced-order PKE).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
pwr-hybrid-3-demo
Preliminary example for the HAHACS thesis — a verified hybrid controller for a small modular PWR startup. Composes three layers into one demonstrable pipeline:
- Discrete layer (
fret-pipeline/): FRET natural-language requirements → LTL → synthesized AIGER controller → state-machine diagram. - Continuous layer (
plant-model/): 10-state point kinetic equation + thermal-hydraulics PWR model with bounded steam-generator heat removal as the disturbance input. - Research context (
thesis/): the HAHACS PhD proposal that motivates and formalizes the methodology.
Layout
pwr-hybrid-3-demo/
CLAUDE.md AI-facing context and architecture map
docs/
architecture.md How the discrete and continuous layers compose
figures/ Shared figures for thesis + talks
fret-pipeline/ FRET → ltlsynt → AIGER → state machine
plant-model/ PWR point kinetics + thermal-hydraulics
reachability/ Continuous-mode verification (linear-model tube + Lyapunov barrier attempt; see README)
julia-port/ Parallel plant-model port + ReachabilityAnalysis.jl scaffold
hardware/ Ovation HIL artifacts (TBD)
claude_memory/ Session notes by AI agents (distilled up into CLAUDE.md over time)
thesis/ [submodule] PhD proposal
presentations/
2026DICE/ [submodule] DICE 2026 abstract
Quickstart
Clone with submodules:
git clone --recurse-submodules <url>
cd pwr-hybrid-3-demo
Run the controller synthesis pipeline:
cd fret-pipeline
python3 scripts/fret_to_synth.py pwr_hybrid_3.json specs/synthesis_config_v3.json
bash scripts/synthesize.sh specs/synthesis_config_v3.json circuits
python3 scripts/trace_aiger.py circuits/PWR_HYBRID_3_DRC.aag diagrams
dot -Tpng diagrams/PWR_HYBRID_3_DRC_states.dot -o diagrams/PWR_HYBRID_3_DRC_states.png
Run the plant model (MATLAB in plant-model/ — Octave compatibility not tested since the LQR pieces landed):
main % original single-scenario demo (null vs operation)
main_mode_sweep % all five DRC modes back-to-back, writes to ../docs/figures/
test_linearize % Jacobian sanity check, saves linearization for reach
Run the reach artifacts (reachability/):
reach_operation % linear reach tube for operation-mode LQR
barrier_lyapunov % Lyapunov-ellipsoid barrier cert attempt (sweeps weights)
Soundness note: the current reach tube is the LINEAR model's tube;
it is not yet a sound over-approximation of the nonlinear plant. See
reachability/README.md § Soundness status.
Prerequisites
- Python 3.10+
- Spot for
ltlsynt(brew install spot) - Graphviz for
dot(brew install graphviz) - MATLAB or GNU Octave for the plant model
- LaTeX (via
latexmk) for the thesis submodule
Further reading
CLAUDE.md— orientation for AI agents working in this repodocs/architecture.md— how the layers composethesis/CLAUDE.md— the thesis project structurefret-pipeline/README.md— FRET naming conventions and pipeline detailsplant-model/README.md— scenario setup and model equations