I think it should depend on the underlying architecture, on how it stores
the floating data types
In case floats and double are implemented using IEEE 754, then floats have 8
bits for precision and double have 11 bits for precision. Normally the
exponents are biased, which means that for float it
On 12 April 2010 23:57, souri datta souri.isthe...@gmail.com wrote:
Why only median of median?
@Souri: It's because using order statistics (randomized select) , you have
an expected time complexity of theta(n) .
But the upper bound is n^2 even to find the smallest element. So to ensure
I think it should depend on the underlying architecture, on how it stores
the floating data types
In case floats and double are implemented using IEEE 754, then floats have 8
bits for precision and double have 11 bits for precision. Normally the
exponents are biased, which means that for float it
Do u have to use only C++ ,cant u use scripting languages like
Pythonwhere precision is very good in Python..esp wen u use Si-Py
On Tue, Apr 13, 2010 at 10:10 PM, Himanshu Aggarwal lkml.himan...@gmail.com
wrote:
I think it should depend on the underlying architecture, on how it stores
