Optimizing Controls with Collimator

No matter the industry, and no matter the complexity of your models, Collimator has you covered when it comes to optimization: Auto-tuning, Model-Predictive Control, Trained Neural Networks, and Automatic Differentiation.


Auto-tuning is a convenient tool that automates the process of fine-tuning your model parameters. This process involves a systematic search through a predefined parameter space, using a cost reduction algorithm to evaluate the performance of each parameter configuration. The goal is to identify the set of parameters that minimize cost of operation across a variety of situations.

Benefits of auto-tuning with Collimator:

Model-Predictive Control (MPC)

Model-Predictive Control (MPC) is an advanced control strategy that uses a model of the system to predict future outcomes and make informed control decisions. Collimator's implementation of MPC allows for the optimization of controls actions for plant models that can be approximated linearly as well as several nonlinear varieties.

Advantages of MPC:

Trained Neural Networks

Trained Neural Networks are another important tool in Collimator's controls toolkit. These synthetic controllers are designed to learn from data, and are especially good at managing complex systems that may be .

Key Features of Trained Neural Networks in Collimator:

Automatic Differentiation (AD)

Automatic Differentiation is a cornerstone technology in Collimator, enabling the efficient computation of derivatives of functions. This is crucial for all manner of optimization tasks, where gradients are needed to adjust parameters iteratively until an optimal solution is found. AD in Collimator is not only accurate but also significantly faster than traditional numerical differentiation methods.

Impacts of Automatic Differentiation in Collimator: