Information Technology Reference
In-Depth Information
■
Terminology from This Chapter
analog technology
ASCII
binary number
bit
bit map
bits per inch (bpi)
byte
character set
(character
encoding system)
color table
data averaging
data compression
data sampling
digital technology
dots per inch
EBCDIC
floatingpoint
number
font table
GIF
gray scale
Huffman algorithm
JPEG or JPG
PostScript
representational
error
resolution
roundoff error
shape table
Unicode
word
■
Discussion Questions
1.
The decimal number 1/10 translates in binary to the infinite
value 0.000110001100011 . . ., so the decimal 1/10 cannot
be stored exactly in either the fixedsize or variablesize ap
proach for floatingpoint numbers. Similarly, the decimal
number 1/100 cannot be stored exactly and must involve
some representational error.
a.
A bank computer could store account balances in either
dollars or cents. (Thus, $1.23 could be stored either as the
floatingpoint number 1.23 or as the integer 123.) Discuss
the relative merits of each approach, and find at least one
advantage of each approach.
b.
When interest is computed for a bank account, the result
will rarely translate to an exact number of pennies. Rather,
the interest will have to be translated to a penny amount. If
the interest is always truncated or rounded down to the
nearest cent, discuss what should be done with the rest.
(This will be a trivial amount for each account, but could
add up if a bank has thousands or millions of accounts.) If
the interest is rounded to the nearest cent, then sometimes
the bank would pay a fraction too much interest to an ac
count. Again, this is trivial for a single account, but the
possibility arises that the bank could pay this overage thou
sands or millions of times. Discuss the relative merits of
rounding down (truncating) or rounding to the nearest cent
from the standpoint of both the bank and the customer.
