Showing 1–2 of 2 results for author: Smith, R P
-
JAXFit: Trust Region Method for Nonlinear Least-Squares Curve Fitting on the GPU
Authors:
Lucas R. Hofer,
Milan Krstajić,
Robert P. Smith
Abstract:
We implement a trust region method on the GPU for nonlinear least squares curve fitting problems using a new deep learning Python library called JAX. Our open source package, JAXFit, works for both unconstrained and constrained curve fitting problems and allows the fit functions to be defined in Python alone -- without any specialized knowledge of either the GPU or CUDA programming. Since JAXFit r…
▽ More
We implement a trust region method on the GPU for nonlinear least squares curve fitting problems using a new deep learning Python library called JAX. Our open source package, JAXFit, works for both unconstrained and constrained curve fitting problems and allows the fit functions to be defined in Python alone -- without any specialized knowledge of either the GPU or CUDA programming. Since JAXFit runs on the GPU, it is much faster than CPU based libraries and even other GPU based libraries, despite being very easy to use. Additionally, due to JAX's deep learning foundations, the Jacobian in JAXFit's trust region algorithm is calculated with automatic differentiation, rather than than using derivative approximations or requiring the user to define the fit function's partial derivatives.
△ Less
Submitted 25 August, 2022;
originally announced August 2022.
-
An Experimental Information Gathering and Utilization Systems (IGUS) Robot to Demonstrate the Physics of Now
Authors:
Ronald P. Gruber,
Ryan P. Smith
Abstract:
The past, present and future are not fundamental properties of Minkowski spacetime. It has been suggested that they are properties of a class of information gathering and utilizing systems (IGUSs).The past, present and future are psychologically created phenomena not actually properties of spacetime. A human is a model IGUS robot. We develop a way to establish that the past, present, and future do…
▽ More
The past, present and future are not fundamental properties of Minkowski spacetime. It has been suggested that they are properties of a class of information gathering and utilizing systems (IGUSs).The past, present and future are psychologically created phenomena not actually properties of spacetime. A human is a model IGUS robot. We develop a way to establish that the past, present, and future do not follow from the laws of physics by constructing robots that process information differently and therefore experience different nows (presents). We construct a customized virtual reality (VR) system which allows an observer to switch between present and past. This robot (human with VR system) can experience immersion in the immediate past ad libitum. Being able to actually construct an IGUS that has the same present at two different coordinates along the worldline lends support to the IGUS hypothesis.
△ Less
Submitted 10 December, 2018;
originally announced December 2018.
