Thanks for the answer. Then how do I set this initial value phi(x)=0 for x<0 phi(x)=1 for x>0
this an Heaviside function of course for the diffusive problem u_t=Du_xx Ragards 2008/8/8 Jonathan Guyer <[EMAIL PROTECTED]> > > > On Aug 8, 2008, at 1:05 PM, franck kalala wrote: > > I am trying to solve the diffusive equation (u)_t=D(u)_xx with this >> initial value >> phi(x)=infinity for x>0 and ph(x)=1 for x>0 >> >> how do i set infinity value in PYTHON?, fipy? >> > > The literal answer to your question is: > > >>> from fipy import numerix > >>> numerix.inf > > or > > >>> from fipy import * > >>> inf > > This value is defined in the underlying NumPy library, however you should > always use fipy.numerix, rather than numpy, when you are using FiPy. > > > *** HOWEVER *** > > 1. This may not work on Windows. I believe that Windows does not > consistently or reliably support IEEE NaN and Inf values, at least within > Python. > > 2. Even if you manage to set this initial condition, it is guaranteed not > to do anything meaningful. FiPy is a numerical PDE solver. While infinity > has a precise mathematical definition, and a value with many of the > appropriate numerical properties is defined in IEEE numerics, infinity is > not generally a useful or meaningful value in a numerical context. If it > doesn't fail even sooner, FiPy will calculate an infinite flux at x=0 as a > result of this infinite gradient. The resulting solution matrix will almost > certainly be singular or otherwise fail to solve. > > In short, "inf" is a sign that your solution has gone astray, not something > you should be striving to represent. > > > > -- *********************************************** *********************************************** Franck Kalala Mutombo [EMAIL PROTECTED] African Institut for Mathematical Sciences, Muizenberg Cape Town, South Africa [EMAIL PROTECTED] ************************************************ ************************************************
