Question 1

What is Zero Overshoot / GSRF?

Accepted Answer

Gradient-Stabilized Recursive Filtering (GSRF Practical) is a deterministic log-space soft thermostat that kills mid-band oscillation and compresses peaks toward a declared normal x*. Not a general tracker. Commercial product from BoonMind with locked evidence and published limits.

Question 2

Does GSRF always beat EMA?

Accepted Answer

No. On locked packs GSRF often wins mid-band osc31 (~+35% phase6.2) and peak compression (~38%, 49/52). It loses tracking MAD (0/52), delayed command paths, and fixed-FA residual detection.

Question 3

What is the frequency sweet spot?

Accepted Answer

Roughly periods of 50–120 samples, with peak osc advantage near +51% at P=80. Very fast (~P=20) and very slow (~P=400–600) can favor EMA on osc31. Pure-sine AC gain plateaus near 0.37 for GSRF while EMA approaches 1.0 at long periods.

Question 4

Can I use GSRF on delayed plant feedback as a command smoother?

Accepted Answer

No for balanced GSRF Practical on locked packs. MAD-to-measured was about +59%, +97%, and +106% worse than EMA at 3, 6, and 10 minute delays.

Question 5

Is the Python snippet a production license?

Accepted Answer

No. The public snippet is evaluation only. Production use requires a commercial license (Pro / Enterprise / QEC path tiers).

Question 6

Is there an open-source production core?

Accepted Answer

No. The site does not promise an open-source production core. Evaluation materials are free; production is proprietary commercial.

Question 7

What is the quantum ladder?

Accepted Answer

A separate QEC/BP preprocessing research and licensing track (code-capacity DEM; rotated surface codes d=5–13 under NDA). Not mixed into industrial homepage evidence. See /quantum.html.

Question 8

Where do the numbers come from?

Accepted Answer

GSRF Audit Workbench locked characterization packs. Public pages restate the claim ladder only. Primary ledger: /evidence.html.

Question 9

Who built this?

Accepted Answer

Zero Overshoot is the product line from BoonMind Labs (research and product development). Product contact gsrf@boonmind.io; general BoonMind info@boonmind.io.

Question 10

Does GSRF win industrial actuators?

Accepted Answer

Not without domain and metric labels. Independent map: stiction osc ~76% (n=25), slow-tracking wear ~68% provisional (n=25), overall wear ~18%. DAMADICS multi-day wear better on 1/75 days — not a plant wear championship. Robot true command+measured continuous fleet (ABB/UR/Franka, n=298): wear 298/298, 0 losses; osc/rate surface-labeled only. Flagship guide: /gsrf-for-actuators.html. Not a controller.

Question 11

What is DAMADICS in Zero Overshoot evidence?

Accepted Answer

A public real-plant actuator benchmark (Lublin sugar factory / IAIR). Multi-day Workbench staging (75 runs): composite mostly loss; wear better on 1/75 days — published honestly, not spun as universal plant wins or inverted 88% wear claims.

Question 12

Do you have true command+measured robot evidence?

Accepted Answer

Yes. Frozen Phase 5 MAIN on public continuous series: ABB RMPD TCP, NIST UR5 joints, Franka DROID joints — n=298, wear 298/298, 0 losses. Oscillation and rate depend on surface (do not merge into one robot percentage). EMA-proxy packs are labeled separately. See /evidence.html#robot-truecmd.

Question 13

What is the best signal filter for mid-band oscillation before AI models?

Accepted Answer

On locked packs, GSRF Practical often reduces mid-band osc31 (~+35%) and sweet-band spectral power (~70%) vs best EMA when the job is ring kill near a known normal — not when minimum tracking MAD or delayed plant command smoothing is the objective.

Question 14

How is GSRF different from a Kalman filter?

Accepted Answer

GSRF Practical is a deterministic log-space attractor stabilizer (soft thermostat) with published oscillation/peak niche and red lights. Kalman is a probabilistic state estimator. Prefer Kalman when you need explicit process/measurement noise models and state fusion; prefer GSRF when mid-band ring and peak chase toward a normal are the costly failure. See /compare.html.

Question 15

What is peak compression vs tracking MAD?

Accepted Answer

Peak compression measures how much peaks are pulled toward a declared normal; MAD tracking measures absolute deviation from the raw series. GSRF is strong on peak compression (~49/52 wins) and weak on MAD tracking (0/52) by design of the k_return spring.

Frequently asked questions