MathGroup Archive 1995

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Base 10 to Base 2 conversion error ?

  • To: mathgroup at christensen.cybernetics.net
  • Subject: [mg1785] Re: Base 10 to Base 2 conversion error ?
  • From: Joe Gwinn <gwinn at sud2.ed.ray.com>
  • Date: Sun, 30 Jul 1995 21:37:18 -0400
  • Organization: Raytheon

So, it's solved by the first case tried -> 0.01 doesn't work.  Sounds like 
the enumeration search worked quite well to answer the stated question.

It's certainly true that one can use lots and lots of digits, but it's 
expensive, and what's the point?

In financial markets, interest rates are stated in "basis points", which 
are hundredths of a percent, so in practice the precision of computation 
cannot exceed one part in ten thousand.

The classic non-religious reason for a bank to do things to the exact 
penny, and to use a carefully-chosen roundoff method for fitting the 
results of an interest calculation into the result field, is to help the 
detection of a classic fraud, where the less-than-a-penny roundoffs are 
deposited in an account which is periodically emptied by a conspiritor.  
(The other method of detection is to look for unbalanced accounts, where 
outgo does not equal income plus interest on balance.)

The other reason is that retail customers complain if things aren't done to 
the penny.   

But is anyone using Mathematica to prepare retail bank statements?  I think 
not.  So, the question remains: why isn't one part in a billion sufficient 
precision?

Joe Gwinn




  • Prev by Date: Re: 3D Axes Label & bounding box of graphics [help]
  • Next by Date: Re: Printing
  • Previous by thread: Re: Base 10 to Base 2 conversion error ?
  • Next by thread: Re: Base 10 to Base 2 conversion error ?