The Discovery Platform makes working with simulators as easy as filling out a simple form. Its Neural Network engine provides the acceleration you need to explore the entire design space.
Simplify
Easily manage complex collaborative projects across large teams and millions of simulations. Run simulations in the cloud or on any of your own resources with the click of a button.
Predict
Create high-accuracy Neural Network emulators with minimal simulation data using the platform’s intelligent data sampling system. Predict new behaviour in a fraction of a second.
Optimise
Choose your target for optimisation and allow the platform to automatically select the best algorithm and parameters. Leverage our emulation technology to explore the entire space with orders-of-magnitude acceleration.
Simply describe the problem and let the platform handle finding the solution.
Adding your simulator to the platform is a painless task — you only need to describe how to run your simulation once. Armed with this information, the platform can:
Optimise designs for a custom objective
Obtain robust sensitivities and error bars with statistical sampling
Generate an emulator to explore the entire design space
Track the results across jobs and resources with live updating, customisable metrics, and visualisations.
One platform. Unlimited possibilities.
The Discovery Platform is already impacting research and industry.
Semiconductors
Ever-increasing design complexity is a growing verification challenge across the semiconductor industry. The Discovery Platform is addressing this challenge by:
Interfacing with popular SPICE simulators to automate intelligent data collection
Training neural network emulators that can explore the circuit design space in minutes
Providing rapid verification, analysis of out-of-spec conditions, and powerful alternatives to traditional datasheets
Engineering
Today’s engineers are pushing simulation technology to its limits. The Discovery Platform is enabling them to design and optimise game-changing systems by:
Building ultra-fast surrogates compatible with existing simulation tools
