Term
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
| largest # vs diff between #s |
|
|
Term
|
Definition
| n + 1, where n is the fractional part in rep |
|
|
Term
| Smallest normalized floating point number in +-1f*2^m |
|
Definition
|
|
Term
| largest floating point number |
|
Definition
m = [L,U]
p = precision (n+10
Smallest: 2^(U+1)(1-2^-P) |
|
|
Term
|
Definition
| 2^-n (gap between 1 and next #) |
|
|
Term
|
Definition
32 bits.
sign = 1 bit
exponent = 8 bit (m-127) (range = -126,127)
significand = 23 bit |
|
|
Term
|
Definition
|
|
Term
|
Definition
64 bits
1 bit sign
11 bits exponent (shift = -1023, m = [-1022, 1023])
52 bits fractional |
|
|
Term
|
Definition
exponent and functional are all 0
[image] |
|
|
Term
|
Definition
sign is plus minus inf
exponent all 1s
significand all 0s
[image] |
|
|
Term
|
Definition
Undefined results
exp = 1111..111
signifiand = anything but 0s
[image] |
|
|
Term
| Subnormal number representation |
|
Definition
exponent all 0s.
significand is 0.{} |
|
|
Term
| Range of integer numbers you can represent exactly using binary system |
|
Definition
|
|
Term
| Rounding gap between x+ and x- |
|
Definition
|
|
Term
|
Definition
|
|
Term
| Absolute error for rounding |
|
Definition
|
|
Term
| Single precision gives you around __ accurate digits |
|
Definition
|
|
Term
| double precision gives you around __ accurate digits |
|
Definition
|
|
Term
|
Definition
|
|
Term
|
Definition
FromDigits["100010011110011", 2] -> base 10 to 2
BaseForm[10231,2] |
|
|
