matplotlib-users

[Matplotlib-users] NaN's and infs

[Darren D. <dd...@co...>]

Would somebody kindly direct me to some information on how to deal with 
dividing by zero?  I am getting ValueError: math domain error, is there 
anything I can do to return an inf instead?

Thanks,

Darren

Re: [Matplotlib-users] NaN's and infs

[Darren D. <dd...@co...>]

On Monday 04 October 2004 10:38 am, you wrote:
> On Monday 04 October 2004 00:58, Darren Dale wrote:
> > Would somebody kindly direct me to some information on how to deal with
> > dividing by zero?  I am getting ValueError: math domain error, is there
> > anything I can do to return an inf instead?
> >
> > Thanks,
> >
> > Darren
>
> If you can't avoid the division by zero, you can handle the exception:
>
> try:
>  a = x/0
> except ValueError:
>  pass # or some other outcome (a = 'inf')
>
> as far as I know python does not have a representation for infinity. (if I
> am wrong, somebody please correct me)


There are some special representations that can be imported, for example

from numarray.ieeespecial import inf

However, My simulations require HEAVY array mathematics, so I dont have an 
opportunity to test for exceptions.

Darren

Re: [Matplotlib-users] NaN's and infs

[Darren D. <dd...@co...>]

>
> from the numpy list:
>  > numarray allows one to customize how errors are handled. You can
>  > choose:
>  >
>  > 1) to silently ignore all errors.
>  > 2) print a warning message (default)
>  > 3) raise an exception.
>  >
>  > One may separately set one of these three behaviors for each of
>  > the 4 ieee categories of floating point errors, namely
>  >
>  > 1) invalid results (i.e., NaNs)
>  > 2) divide by zeros (Infs)
>  > 3) overflows
>  > 4) underflows
>
> now try this:
> from numarray import *
> 1./arange(10)
> Warning: Encountered divide by zero(s)  in divide
> array([              inf,   1.00000000e+000,   5.00000000e-001,
>          3.33333333e-001,   2.50000000e-001,   2.00000000e-001,
>          1.66666667e-001,   1.42857143e-001,   1.25000000e-001,
>          1.11111111e-001])
>
>
> there is the inf!!
>

Thank you Flavio,

I had tried the test you just suggested, and only got the warning message. I 
incorrectly assumed that the result had not been returned. Now I see that it 
was returned:

from numarray import *
a=1./arange(10) #displays error
print a              # displays a with the inf

Thanks again,

Darren