Why does .Net return infinity?

Question

This does not make any sense. In case of overflow or divide by zero, .NET should throw an exception or at least return an error code or bool (like TryParse). I think many would agree that it is difficult to anticipate that a framework would return infinity. This is because there is nothing really called infinity in computer science (How it will be represented in memory!). Furthermore, I cannot really do anything with it, I cannot use it as an input in another equation. This problem happened due to a bug that result in calling Math.Log(0).

What do I miss here?

https://learn.microsoft.com/en-us/dotnet/api/system.double.isinfinity?view=net-6.0

This is how IEEE floating-point math works: this is consistent across the majority of the world's programming languages. Various well-defined operations can result in positive/negative infinity or NaN, and the results of performing further operations on these values are also well-defined. This article is required reading for all computer scientists: https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html — canton7, Dec 21 '21 at 16:21
`This is because there is nothing really called infinity in computer science` Wrong. [IEEE 754](https://en.wikipedia.org/wiki/IEEE_754) was introduced in 1985 and contains not just infinites but Not-a-Number, which you'll find in data science languages and libraries, like R and Python. — Panagiotis Kanavos, Dec 21 '21 at 16:33

Dmitry Bychenko · Accepted Answer · 2021-12-21T17:01:12.537

You have 2 different cases:

When result is some kind of integer or decimal, DivideByZeroException exception will be thrown:

The exception that is thrown when there is an attempt to divide an integral or Decimal value by zero. (bold is mine)

// DivideByZeroException
long v = 123L / 0L;

// DivideByZeroException
decimal d = 456m / 0m;

When result is some kind of floating point, no exception will be thrown and there are 3 possible outcomes: +Inf, NaN, -Inf:

// Positive Infinity
double Positive = 123.0 / 0.0;

// NAN - Not A Number
double NotNumber = 0.0 / 0.0;

// Negative Infinity
double Negative = -456.78 / 0.0;

In case of Math.Log() we deal with floating point values (double) for both argument and result, that's why

 Math.Log(0.0) == double.NegativeInfinity

Note as well

 Math.Log(-1.0) == double.NaN

is quite natural implementation;

Why does .Net return infinity?

1 Answers1