01 / PHYSICAL SYSTEMS
TrueLoop Runtime
Maintain an n-channel analog layer while its n²-scale transform stays in the device.
- One parallel acquisition
- Complete multichannel update
- Hosted or offline
Stability for hybrid computing
Hybrid computing.
Ready to scale.
Move the heavy arithmetic into physics. Keep control digital.
TrueLoop keeps component-observable analog hardware true without putting a growing calibration burden back on the digital system.
Free 90-day evaluation. No card.
6
measurements to usablecold start · controlled simulation
1
acquisition per cyclecomplete configuration
1
acquisition on restartnear a known state
Digital directs.Physics computes.TrueLoop keeps it true.Maintenance tested to 100 million channels.Controlled simulation.
Start with your system
What do you run?
The missing layer
Digital directs the computation.
Physics performs the transform.
TrueLoop keeps the analog state true.
One read.One update.No growing calibration pass.
See how the architecture fits together The architecture
A hybrid machine
that can compose.
DIGITAL HOSTOrchestration, semantics, routing, and control flow
PHYSICAL COPROCESSORDense transforms carried by resident analog state
STABILITY LAYEROne complete readout and actuation frame per cycle
NEW DESIGN SPACEOn-device AI, photonics, quantum, RF, and sensing
Two production surfaces
One runtime.
Applied where it actually happens.
02 / SIMULATORS
TrueLoop SimWrap
Apply the same runtime to expensive vector-response simulations.
- Same Python client
- One response per update
- Offline for tight deadlines
Integration
One response vector.
One complete update.
Five lines. Your loop stays yours.
PYTHON / CONTROL LOOPFIVE LINES
opt = SWCOptimizer(key, n=len(x0), mode="regulation", target=target)
x = opt.start(x0)
for _ in range(budget):
x = opt.step(measure(x), target=target)
opt.end()measure(x) is your existing readout.
Read every component in parallel.
Carry bounded local memory per channel.
Return the next complete configuration.
Real time
Control keeps up.
Your acquisition sets the pace.
per channel-cycle
digital work per cycle
state per channel
Measured controller arithmetic on one commodity core. The device readout and actuation set the physical loop rate.
commercial QPUs
Measured hardware.
lower RMS error
IBM Heron. 156 qubits.
channels held flat
Independent reference implementation.
Fit first
Component-resolved readout. Feasible target. Bounded drift and coupling.
Clear boundary
No scalar-only search. No end-to-end hybrid workload claimed yet.