Double Precision
64 binary bits per floating-point number: 52-bit significand and 11 bit exponent.
ϵmach=2−52≈2.2×10−16
- roughly gives about 16 decimal digits (not places)
- limits range of exponent between -1022 and 1023
- Values greater than 2×10308 result in overflow and an
Inf
- underflow results in a 0
NaN
result of undefined arithmetic operations
Relative julia
functions: sign()
, exponent()
, significand()
, bitstring()
julia
doesn’t allow indexing by type float; when force coercian of noninteger floats to integers, get anInexactValue
error.