You have to take into consideration that the output format must be X.X where X is a number.

Random tests often expects incorrectly formatted numbers, such as 032. -> 032.0 and .470590 -> 0.470590.

In fact, there are no considerion about integers and leading/trailing zeroes, so the results can barely be called a decimal number: we don't normally express them as something like 000005.500000 instead of 5.5. (Trailing .0 on integer values is debatable as well.) Or rather, the title should be changed: Double is a very different beast from decimal number.

It is not specified what is considered a valid decimal number. .52 and 6. are somehow considered valid.

Similarly, null handling is required but not specified.