Information Technology

Most Recent Inventions

Physics ‘Office Hours’ educational learning platform

A physics education researcher at the University of Wisconsin-Green Bay has designed a novel and interactive app-based study aid platform for students in STEM disciplines. The platform’s interface is built around education research into how students conceptualize problems they do not understand. It is a novel tool to help students see why they are struggling with a particular problem, and what might help them solve it, rather than solving the problem for them. The team’s first working prototype, the Physics Office Hours app, has been designed for use in introductory-level college physics. The app is designed to mimic a scenario students might face during ‘office hours’ with a professor: Rather than offering an answer, the instructor guides the students through problems via a series of questions. A user-friendly online interface allows app content to be easily updated and changed over time and as more problem sets become available. In addition, the app architecture can easily be adapted to problem sets in other STEM disciplines and therefore serves as a platform technology.

Matrix Processor with Localized Memory to Increase Data Throughput

To address this challenge, Li’s team has developed computer architecture combining local memory elements and processing elements on a single integrated circuit substrate. In this way, data stored in a given local memory resource normally associated with a given processing unit is shared among multiple processing units. The sharing may be in a pattern following the logical interrelationship of a matrix calculation (e.g., along rows and columns in one or more dimensions of the matrix).

Sharing reduces memory replication (the need to store a given value in multiple local memory locations), thus reducing the need for local memory and unnecessary transfers of data between local memory and external memory. This permits high speed processing on local memories (on-chip) and reduces energy consumption associated with a calculation.

New Isogeometric Analysis Software for Seamless Integration of Design and Analysis

UW–Madison researchers have developed a new method for creating a CAD-compatible mesh during an isogeometric analysis process. Unlike existing techniques, the method creates meshes without any approximation and delivers optimal convergence rates.

In essence, the researchers have developed a smoothing step that prevents inconsistencies from being introduced into the meshing process as a geometric map of the object is being refined.

Detecting Seismic S Waves with Unprecedented Accuracy

UW–Madison researchers have developed an automatic and extremely accurate method for detecting features of interest in seismic data, including S waves and P waves. Unlike currently available (and error-prone) phase detection methods, the new software identifies potential picks in a single pass through the data without needing to estimate parameters or build a model. Seismic features are identified based on their similarity to a reference set of examples.

The software utilizes a k-nearest neighbors approach. This approach is based on a nonparametric time series classification method.

More Efficient Laminate Analysis

UW–Madison researchers have developed a method for analyzing composite laminate structures that combines the generality of 3-D FEA and efficiency of 2-D FEA whenever it is applicable. The new method works by substituting the laminate layers with much simpler virtual material models having matching characteristics (e.g., overall material properties and relationship between stresses and strains). The updated model can then by analyzed via fully automated 3-D FEA.

The virtual models may be referred to as ABD-equivalent models, as they result in the same ABD stiffness matrices as the real laminate and can act as substitutes if plate-shell assumptions apply.

Most Recent Patents

Memory Processing Unit Boosts Performance, Cuts Energy Usage

UW–Madison researchers have developed a system to dramatically improve the benefits of 3-D die-stacking memory. Their system allows a host processor to efficiently offload entire pieces of computation for faster processing with reduced power consumption.

More specifically, memory processing unit cores are tightly coupled with sections of stacked memory layers, combined as memory ‘vaults’ in hardware. Application code is segmented into discrete partitions (‘shards’) in software for storage in the vaults. As a result, an application program is effectively broken up for execution among multiple processing cores in close proximity to memory.

Indoor Navigation System for Mobile Devices

UW–Madison researchers have developed an indoor navigation system using a mobile device equipped with two photodetectors. The system is able to determine the angular position of different light fixtures while avoiding the limitations in bandwidth and sensitivity associated with standard camera detectors. It is suitable for facilities with high ceilings more than three or four meters above the floor

Specifically, the new method can be illustrated in three steps: (i) identify multiple light source signals within the field-of-view according to known light source signatures; (ii) determine the angles of the multiple light sources; and (iii) identify the location of the mobile device based on the angle of the multiple light sources and a known mapping of light sources to locations.

Architecture for Speculative Parallel Execution Improves Performance, Simplifies Programming

UW–Madison researchers have developed a system that permits speculative execution of program tasks prior to determining data dependency. Before commitment of the tasks in a sequential execution order, data dependencies are resolved through a token system that tracks read and write access to data elements accessed by the program portions.

Eliminating the need to wait until late in the program execution to detect or resolve dependencies helps improve processor utilization. Advancing the execution of tasks that ultimately do not experience data dependency problems may have a ripple-through effect, reducing later data dependencies as well.