Does retuning the gain on the sampling-interval law, instead of the fixed-period twin, remove the missed-sample penalty?
research/free-retunegain
Does retuning the gain on the sampling-interval law, instead of the fixed-period twin, remove the missed-sample penalty? Yes for this plant: the mean-square-optimal gain is stable at every miss probability (worst E g² 0.478 vs a twin-gain limit of 0.270), at the price of a slower nominal loop; a gain designed for q = 0.3 stays stable at every true q. Stylised scalar model.
What it found
Follow-up to free-jittertwin on the same unstable scalar plant x′ = 3x. The gain minimising E g2 over the miss law (period h0, or 2h0 with probability q) is closed form and mean-square stable for every q (worst E g2 = 0.478, versus a twin-gain limit of q = 0.270). The price is a slower nominal loop (0.63–0.82 per on-time sample). A gain designed for q = 0.3 stays mean-square stable at every true q. Stylised scalar model; one plant; iid misses.
Paper, code and results
More in Digital twins & sim-to-real
- What does a 1:N model of a cargo UAV tell you about the 1:1 aircraft?Research programme (proposal stage): per-quantity error budget for a Froude-scaled cargo aircraft model, with a stdlib scaling calculator; first-order result: at 1:10 the model needs 17 to 35% more power than Froude scaling predicts
- Does a physics-shaped twin of a real buck converter beat plain ARX? Only partly: all models sit at 0.5 to 0.9 V free-run error against a 0.06 V quantisation floor, and the winner flips with which half of the 20 ms record is fittedDoes a physics-shaped twin of a real buck converter beat plain ARX? Only partly: all models sit at 0.5 to 0.9 V free-run error against a 0.06 V quantisation floor, and the winner flips with which half of the 20 ms record is fitted. One 20 ms record; natural frequency fitted, no component values.
- How much feedback delay can a digital twin's controller tolerate, and does a fixed-step simulator get that margin right?Exact: a fixed-step twin under-reports a PD loop's delay margin by 18.6% at dt = 0.1 s and 1.0% at 5 ms, because the largest stable delay is the largest n with (n + 1/2) dt below the true margin (25 of 25 cases)
- How wrong is a simulator's pendulum period at its default timestep?How wrong is a simulator's pendulum period at its default timestep? Exact elliptic-integral period against Euler, semi-implicit Euler, Verlet and RK4: Euler fails (14% period error at 2 ms), the symplectic pair is 0.75 ppm, and releasing from rest hides the symplectic error.
- Which 95% upper bound on a failure rate actually covers 95% at simulator-rollout sample sizes?Which 95% upper bound on a failure rate really covers 95% at rollout sample sizes: exact coverage shows only Clopper-Pearson never undercovers (min 0.950), Wilson dips to 0.93, Jeffreys to 0.85 and Wald to 0.06–0.75; zero-failure certification of p ≤ 1% takes 299 rollouts (CP), 268 (Wilson) or 191 (Jeffreys).
- How many fixed steps per orbit does a twin need before an eccentric orbit stops drifting?How many fixed steps per orbit does a twin need before an eccentric orbit stops drifting? Against the exact Kepler solution RK4 needs 64 to 4096 steps per orbit (e = 0 to 0.9), the symplectic pair 8 to 16 times more, and at e = 0.9 every fixed step fails.
All 164 Digital twins & sim-to-real projects →
Other research topics
Proper scoring & elicitation28Effort & peer prediction9Verification games & slashing19Detection & tolerances14Markets as routers13Prediction-market design9Paying contributors8Decentralised training systems23Digital twins & sim-to-real164LLM chip design loops40Robot manufacturing loops53Ambient AI assistant46Self-replicating manufacturing34Simulation-first design64Engineered clothing45Task-specific AI accelerators40Algorithmic chip-design loops42Open-source purpose-built robots44The software-engineering loop43Adversarial human imitation41Open-source residential construction41Autonomous multi-angle capture34Sensing, imaging and 3D31Energy supply and dense energy storage39Robot societies among people31Agentic co-design of low-cost robots27Laundry-hanging robots25Machine learning on qualitative data23Robots on the building site27Civilizations & group psychology1History & human sciences1
Produced with AI assistance (Claude). Results are from the code in this repository and are simulated or stylised unless the abstract says otherwise; nothing here implies affiliation with or endorsement by any lab, researcher or company cited. MIT licensed.