Self-Dynamic Regulator with thermodynamic adaptation. Replaces PID with zero manual tuning. Now enhanced with ThalosGuard for enterprise stress management across cloud, trading, and IoT systems.
Traditional PID requires manual tuning and fails under changing conditions. Axiom adapts automatically.
Thermodynamic adaptation law automatically adjusts gain parameters. No manual tuning, no Ziegler-Nichols, no trial and error.
100/100 wins against PID under parameter uncertainty in randomized trials. Maintains performance when plant dynamics change.
Same interface as PID: error in, control signal out. Works with existing sensors, actuators, and PLCs. Python, C, and REST API.
Head-to-head comparison on thermal control benchmark (lower IAE is better)
| Test | PID (Tuned) | Axiom SDR | Improvement |
|---|---|---|---|
| Thermal Tracking (IAE) | 965 | 245 | 3.9x |
| Robustness (100 trials) | 0 wins | 100 wins | 100% |
| Noise Rejection (IAE) | 750 | 140 | 5.4x |
| Setpoint Change Response | 1.2s | 0.4s | 3.0x |
Drop-in replacement for PID in any language
# Python - 3 lines to replace PID from thalosforge import Axiom ctrl = Axiom(k_base=50) while True: temp = read_sensor() error = temp - setpoint output = ctrl.compute(error, dt=0.1) # That's it set_actuator(output)
Anywhere you use PID, Axiom performs better
Temperature control with varying loads, occupancy changes, and outdoor conditions. Reduces energy waste from overshoot.
Chemical reactors, distillation columns, and batch processes with changing dynamics. No re-tuning between batches.
Thermal management for EV batteries, data centers, and power electronics. Adapts to aging and environmental changes.
Motor control, position regulation, and force control. Handles payload changes and mechanical wear automatically.
Thermodynamic stress regulation for cloud, trading, and IoT at scale
Quantifies system strain from accumulated stress. Thermodynamic principles measure the "cost" of maintaining current state under load.
Real-time risk scoring based on current conditions. Adaptive thresholds respond to changing environments without manual recalibration.
Coordinate stress across fleets of controllers. Global optimization while respecting local constraints. Scale from 10 to 10,000 units.
Scale infrastructure based on actual stress, not just CPU%. Prevent cascading failures with coordinated fleet response.
Real-time position stress monitoring. Adaptive risk limits that respond to market volatility without manual intervention.
Coordinate thousands of edge devices. Balance load across the fleet while maintaining local autonomy.
Thermal and power stress management. Prevent hotspots and optimize cooling across racks and zones.
Run anywhere from microcontrollers to cloud
200 bytes RAM, <50μs latency. Runs on ESP32, STM32, Arduino. Header-only library, no dependencies.
Full-featured Python library with diagnostics, logging, and state persistence. Integrates with Modbus, BACnet, OPC-UA.
REST API for distributed control. 50,000+ requests/second. Managed state, multi-tenant, SOC 2 compliant.