Loss of significance is an undesirable effect in calculations using floating-point arithmetic. It occurs when an operation on two numbers increases relative error substantially more than it increases absolute error, for example in subtracting two nearly equal numbers (known as catastrophic cancellation). The effect is that the number of accurate (significant) digits in the result is reduced unacceptably. Ways to avoid this effect are studied in numerical analysis.
Read more about Loss Of Significance: Demonstration of The Problem, Workarounds, Loss of Significant Bits, Instability of The Quadratic Equation, A Better Algorithm
Famous quotes containing the words loss of, loss and/or significance:
“... imprisonment itself, entailing loss of liberty, loss of citizenship, separation from family and loved ones, is punishment enough for most individuals, no matter how favorable the circumstances under which the time is passed.”
—Mary B. Harris (1874–1957)
“Every farewell combines loss and new freedom.”
—Mason Cooley (b. 1927)
“Politics is not an end, but a means. It is not a product, but a process. It is the art of government. Like other values it has its counterfeits. So much emphasis has been placed upon the false that the significance of the true has been obscured and politics has come to convey the meaning of crafty and cunning selfishness, instead of candid and sincere service.”
—Calvin Coolidge (1872–1933)