fixed systems
SimWrap cleared the registered threshold. Adaptive RLS and Anderson remain serious comparators when commissioning fits the budget.
simulatedTrueLoop SimWrap
Your simulator.
Inside a tighter budget.
SimWrap is the TrueLoop runtime applied to a coupled black-box vector simulator. It is ready through the existing Python client: send a configuration and simulator response, then evaluate the next configuration it returns.
The product
The runtime, wrapped around your simulator.
Your simulator remains the source of truth. SimWrap keeps the iterative state and chooses the next configuration from the response already produced by the simulator. It does not replace or accelerate the simulator itself.
01Configureapply x
→02Simulateobserve vector y
→03SimWrapreturn next x
Controlled simulator study
Usable next configurations under a six-response budget.
At 64 coupled variables, after six charged simulator responses, SimWrap produced a next configuration inside the registered hidden-truth threshold in 10 of 10 fixed trials and 9 of 10 changing trials.
changing systems
SimWrap cleared the registered threshold. The correct comparison is a matched race against a competently commissioned incumbent.
simulatedstructured scale trials
Post-update configurations passed across six synthetic response classes from 64 to 4,096 variables.
simulatedThese are post-update scoring results. In a black-box deployment, run the returned configuration through the simulator once more to verify acceptance. A current exact model can solve its supported case in one evaluation, and a well-commissioned solver may be better when commissioning fits the budget.
Technical fit
Use SimWrap when evaluations are the scarce resource.
The strongest fit is a writable configuration vector, an associated vector response, useful component-level output, expensive calls, and a deadline too short for identification or repeated diagnostic acquisitions.
Inverse response matching
Find a configuration whose simulated vector output approaches a specified target.
Digital-twin reconciliation
Correct setpoints as a coupled twin or plant representation changes.
Hardware-in-the-loop
Keep the outer loop within a strict evaluation and wall-clock budget.
Co-simulation control
Coordinate vector setpoints across coupled model boundaries.
Reduced-output engineering models
Control CFD, plasma, structural, or multiphysics models through selected response outputs.
Vector-valued stochastic calibration
Use decoded response statistics when derivatives are unavailable or unreliable.
These are candidate application mappings defined by the interface and operating conditions. They are not claims of validation in every named domain.
Evaluation to production
Start hosted.
Move offline when latency matters.
The hosted endpoint uses the same client lifecycle and is intended for evaluation and simulator calls measured in seconds or longer. For sub-second loops, data residency, or air-gapped production, contact us for an offline build and commercial license.