A method is provided for determining soil properties for an area of land, from soil spectrum data. In an embodiment, the method includes receiving soil spectrum data records from hyperspectral sensors that represent a mean soil spectrum of a specific geo-location of the area of land and removing interference signals from the spectrum data records to create soil spectral bands. The method also includes predicting a plurality of soil property datasets based on a partial least-square regression and the soil spectral bands and selecting specific soil property datasets from the plurality of soil property datasets to represent soil properties of the specific geo-location, wherein the specific soil property datasets include property data and spectral band data for spectral bands used to determine the property data. The specific soil property datasets may then be used to generate a crop prescription of recommended hybrid seeds or population densities for the specific geo-location.

Systems and methods for reducing computer resources in determining a set of solutions for multi-objective optimization based on competing metrics. The methods and system include: receiving data related to each metric; optimizing each metric, irrespective of the other metric to provide endpoints of a Pareto front. A third optimization is made between the end points, thereby providing three optimization points to generate a Pareto front which represents a set of solutions for the competing metrics.

A system expands an existing library based on simultaneous optimization of a circuit design being built and the library cells being used. The system receives a library of cells and a circuit design and performs synthesis and optimization of the circuit design. The system evaluates the circuit design to identify portions that may be candidates for new library cells. The system analyzes the library to determine whether there is an existing library cell that can be used, whether the new libcell should be added to the library, or whether the new libcell should replace an existing libcell. The system performs modeling for the new libcell to measure the improvement obtained by use of the new libcell. The system recommends the new libcell for addition to the library based on the performance modeling.

To create high-performance metasurface devices (110) in an inverse design process over a large design space (100), a deep neural network may be used as a surrogate model in lieu of a full physics simulation to more efficiently predict figures of merit for given input metasurface topologies during the iterative topology optimization. The neural network may also serve to efficiently compute, via backpropagation, gradients of the figures of merit with respect to design parameters, as are used to update the topology in each iteration.

A computing device receives user input for a design of an experiment. The user input indicates respective factor identities for factors in the design of the experiment, respective response identities for responses to options for the factors in the design, and a user-defined objective for the one or more responses. Additionally, based on the user input, the computing device displays a subset of a set of multiple model types and a user-definable amount of design runs for the design. Each design run presents settings according to the design for each of the factors. The computing device also receives settings indicating a user-selected model type from the subset and a user-defined amount of design runs, and based on the settings, selects one or more design construction criteria for generating the design. The computing device then generates the design according to the design construction criteria.

A method and computer program for designing straightening aligners are disclosed. The method for designing straightening aligners according to an exemplary embodiment of the present invention includes generating current alignment information representing a patient's current tooth alignment state, and generating orthodontic alignment information for multiple orthodontic steps that are reached sequentially by correcting the current alignment information, wherein generating the orthodontic alignment information sets the multiple orthodontic steps such that the total sum of orthodontic forces applied to the patient's teeth per one orthodontic step is less than or equal to a preset total sum of orthodontic forces in order to reach a tooth alignment state corresponding to the orthodontic alignment information of each orthodontic step.

A method of modelling system behaviour of a physical system, the method including, in one or more electronic processing devices obtaining quantified system data measured for the physical system, the quantified system data being at least partially indicative of the system behaviour for at least a time period, forming at least one population of model units, each model unit including model parameters and at least part of a model, the model parameters being at least partially based on the quantified system data, each model including one or more mathematical equations for modelling system behaviour, for each model unit calculating at least one solution trajectory for at least part of the at least one time period; determining a fitness value based at least in part on the at least one solution trajectory; and, selecting a combination of model units using the fitness values of each model unit, the combination of model units representing a collective model that models the system behaviour.

An architecture search system evaluates a search space of neural network and hardware architectures with a plurality of candidate controllers. Each controller attempts to identify an optimized architecture using a different optimization algorithm. To identify a controller for the search space, the architecture search system samples subspaces of the search space having a portion of the neural network search space and a portion of the hardware search space. For each subspace, candidate controllers are scored with respect to the optimized design determined by the respective candidate controllers. Using the scores for the various candidate controllers across the sampled subspaces, a controller is selected to optimize the overall network architecture search space.