Unsupervised Learning of Shape-invariant Lie Group Transformer by Embedding Ordinary Differential Equation

Author(s):  
Takumi Takada ◽  
Yoshiyuki Ohmura ◽  
Yasuo Kuniyoshi
2016 ◽  
Vol 5 (2) ◽  
pp. 117
Author(s):  
Sami H. Altoum

This paper deals with a numerical comparison between Lie group method and RK4 for solving an nonlinear ordinary differential equation. The Lie group method will be introduced as a analytical method and then compared to RK4 as a numerical method. Some examples will be considered and the global error we be computed numerically.


2013 ◽  
Vol 53 (5) ◽  
pp. 438-443 ◽  
Author(s):  
Decio Levi ◽  
Pavel Winternitz

We briefly review two different methods of applying Lie group theory in the numerical solution of ordinary differential equations. On specific examples we show how the symmetry preserving discretization provides difference schemes for which the “first differential approximation” is invariant under the same Lie group as the original ordinary differential equation.


Author(s):  
Razvan Gabriel Iagar ◽  
Philippe Laurençot

A classification of the behaviour of the solutions f(·, a) to the ordinary differential equation (|f′|p-2f′)′ + f - |f′|p-1 = 0 in (0,∞) with initial condition f(0, a) = a and f′(0, a) = 0 is provided, according to the value of the parameter a > 0 when the exponent p takes values in (1, 2). There is a threshold value a* that separates different behaviours of f(·, a): if a > a*, then f(·, a) vanishes at least once in (0,∞) and takes negative values, while f(·, a) is positive in (0,∞) and decays algebraically to zero as r→∞ if a ∊ (0, a*). At the threshold value, f(·, a*) is also positive in (0,∞) but decays exponentially fast to zero as r→∞. The proof of these results relies on a transformation to a first-order ordinary differential equation and a monotonicity property with respect to a > 0. This classification is one step in the description of the dynamics near the extinction time of a diffusive Hamilton–Jacobi equation with critical gradient absorption and fast diffusion.


Sign in / Sign up

Export Citation Format

Share Document