Why float not precise even when it is a small decimal number when multiplying with a large long number in Java?

Question

I am multiplying a ratio to a sample count. The ratio is represented as a float number (0.5f) and a long type number represent the sample count.

In Java, the multipli