Gradient Control Laboratories

GCL exists to build the future of CAD, CAE, and CAM software applications by exploring user experience paradigms, procedural and implicit geometric modeling, interactive programming languages, and modern data science.

We foster agile and humane software teams who pursue value streams via best-in-class engineering practices.

Our partners, Blake Courter and Luke Church founded or helped lead over a dozen successful startups ranging from engineering software through humanitarian communications. 

Stacks

Team

Research topics

Unit Gradient Fields: While implicit modeling, especially the use of signed distance fields, has delivered ease-of-use and robustness in 3D graphics, they are either too loose or constraining for professional engineering applications.  We research algorithms that focus on fields with unit gradient magnitude to enable implicits in scalable, mainstream engineering applications.

Geometry as Code:  Unlike traditional CAD scripting, which separates algorithms and geometry, our code-based approach lets logic and geometry coexist, allowing intelligent models that both generate and document design intent.  With geometry as code, engineering knowledge becomes portable, reusable, and understandable by AI, which, as free and open software, places equity with creators.  

The CAD Isomorphism: By abstracting geometry-as-code to an annotated syntax tree, we produce an isomorphism between textual syntax, visual programming, and engineering geometry itself.  We explore the human-computing interfaces of this system to build interactive CAD applications.

Second Order Design Complexity: When implicit modeling for optimization applications, two second order phenomena complicate the language:  First, some variables need to be fully differentiable and optimizable, so one ends up with elevated variables or parameters.  Also, when optimizing, one typically both implements a function and uses it as an argument.  How do we design interactive applications that make these relationships clear without undesirable emergent behavior?

Exactness in Engineering: When we solve engineering problems via partial differentiable equations and integrate gradients to find configurations, engineers implicitly assume solution space is contractible.  However, constrained engineering problems typically reveal configuration space holes, disconnected feasible regions, and hard constraints that may fragment solution space into multiple components.  To what extent can we observe and navigate the topology of solution space by measuring where exactness breaks in its cohomology chain?

Current residencies

Recent residencies

Active advisees, fellowships, editorships, collectives, and lecturing

info@gradientcontrol.com