Optimization as regulation
Minimize a compatible response-matching objective through the regulation interface.
Usable means the verified error is inside the published threshold.
TrueLoop supports optimization when success is defined as matching a measured vector response to a feasible target. The customer supplies the configuration, the measured response, and the target. The runtime returns the next complete configuration.
This covers response matching, calibration, tracking, inverse problems, and nonlinear least-squares applications that expose error component by component. It uses the same regulation interface and acquisition accounting as setpoint control.
The application fits when every relevant component is measured before drift makes the reading stale, the target is reachable and separated from response extrema, readout quality is sufficient, and coupling remains inside the tested operating boundary.
It does not perform general optimization of scalar-only, multimodal, or combinatorial objectives.
opt = SWCOptimizer(key, n=len(x0), mode="regulation", target=target)
x = opt.start(x0)
for _ in range(rounds):
x = opt.step(measure(x), target=target)
opt.end()Verified six-acquisition result
In controlled simulation, TrueLoop reached the independently verified RMS usability threshold in 20 of 20 fixed and 20 of 20 changing-response trials at 100,000 globally coupled variables. Median verified RMS was 0.0648.
Under the same six-acquisition deadline, with drift near 0.1 rad per acquisition and global coupling, adaptive RLS, Anderson, tuned SPSA, random search, finite differences, coordinate search, and compact CMA-ES produced zero usable solutions. This is a controlled response-matching result, not general scalar optimization.
Run the envelope, then race the runtime and your incumbent under the same measured budget.