The RPN input method is based upon working with numbers stored in the automatic memory stack, which consists of the x, y, z, and t registers. The stack lift and drop operations are performed automatically and are key to understanding its use. Imagine the registers stacked on top of each other with t (on the top) above z, which is above y, which is above x. Both the lift and drop operations cause values to be shifted to an adjacent register, but shift in opposite directions.
The stack lift operation is performed to make room for a new value in the x register: the value in the t register is thrown away, the t register gets the z value, the z register gets the y value, and the y register gets the x value. This allows the new x value to be stored while preserving the original x value (in the y register).
The stack drop operation is performed when an operation requiring two operands is executed: the t register remains unaffected, the z register gets the t value, the y register gets the z value, and the x register gets the y value. The result of the calculation is then stored in the x register, effectively combining (or consuming) the original x and y values. Notice that the value in the t register is propagated downward. This is useful for operations involving multiple uses of a constant value.
Calculator operations can be categorized as either lifting the stack, dropping the stack, or neither lifting nor dropping the stack. Also, some operations disable stack lift, meaning that if the next operation normally lifts the stack, then the stack won't be lifted. Or, if the next operation lifts the stack twice, the stack will only be lifted once.
Let's look at a simple example to calculate the square of the sum of two numbers,