How to understand amortized analysis potential?

Since you gave kind of a vague question. I'll just go with the basics. Amortized analysis in computer science is basically the study of worst case run times regarding a sequence of operations.

When looking at potential, it is the physicist's method.

phi (initial state) = 0 and every state after is larger than 0.

It keeps track of time but relies on states to know where it is.

The equation C + phi (state') - phi (state) is the main equation. C is the time for an operation, "state" is before and "state'" is after.

There are sets of equations that dictate average run time with this.

ex.

phi (H) = 2n-m. n=number of elements, m=size of array.

This equation is used to calculate the time to double the size.