float64 print digits

[Ulrich Dorda]

I need a pytho nscript to read numbers(with loads of digits) from a
file, do some basic math on it and write the result out to another file.

My problem: I don't get python to use more digits:

In order to try this I type:

The normal precision one:

>>from numpy import *
x=1.23456789123456789123456789
print "%35.25e" %x

1.2345678912345679000000000e+000
Now I try to use float64 to get more digits

>>z=zeros(3,float64)
z[0]

0.0

>>type(z[0])

<type 'numpy.float64'>

>>z[0]=1.23456789123456789123456789
type(z[0])

<type 'numpy.float64'>

>>print "%35.25e" %z[0]

1.2345678912345679000000000e+000

This cuts the digits just like the 32bit case.

Can anyone please help me get more digits?

Thank you very much in advance,

Ulrich

[Peter Otten]

Ulrich Dorda wrote:

[Warning: I'm no expert and don't have numpy installed]

I need a pytho nscript to read numbers(with loads of digits) from a
file, do some basic math on it and write the result out to another file.

My problem: I don't get python to use more digits:

In order to try this I type:

The normal precision one:

>>from numpy import *
>>x=1.23456789123456789123456789
>>print "%35.25e" %x

1.2345678912345679000000000e+000
Now I try to use float64 to get more digits

>>z=zeros(3,float64)
>>z[0]

0.0

>>type(z[0])

<type 'numpy.float64'>

>>z[0]=1.23456789123456789123456789

The right side of the assignment is a float, so the extra digits would
already be lost before you get a chance to convert to float64. But then
Python's float is a C double and should already use 64 bits...

>>type(z[0])

<type 'numpy.float64'>

>>print "%35.25e" %z[0]

1.2345678912345679000000000e+000

This cuts the digits just like the 32bit case.

Can anyone please help me get more digits?

gmpy?

Peter

[Grant Edwards]

On 2007-03-02, Ulrich Dorda <ul**********@cern.chwrote:

I need a pytho nscript to read numbers(with loads of digits) from a
file, do some basic math on it and write the result out to another file.

My problem: I don't get python to use more digits:

In order to try this I type:

The normal precision one:

>from numpy import *
x=1.23456789123456789123456789
print "%35.25e" %x

1.2345678912345679000000000e+000
Now I try to use float64 to get more digits

You're already using 64 bit floats (which have about 15
significant digits).

>z=zeros(3,float64)
z[0]

0.0

>type(z[0])

<type 'numpy.float64'>

>z[0]=1.23456789123456789123456789
type(z[0])

<type 'numpy.float64'>

>print "%35.25e" %z[0]

1.2345678912345679000000000e+000

This cuts the digits just like the 32bit case.

What 32-bit case? A 32-bit float only has 7-8 significant
digits.

Can anyone please help me get more digits?

Use the decimal module?

--
Grant Edwards grante Yow! What a
at COINCIDENCE! I'm an
visi.com authorized "SNOOTS OF THE
STARS" dealer!!