I understand that EVM is a root mean square of error vectors. If I have transmitted symbols and received symbols (once equalized), taking a root mean square would give me EVM. I believe this is actual EVM (we are using actual tx symbols and mapping them to equalized received symbols and then computing EVM).
How does a signal analyzer (which doesn't know tx symbols) computes EVM?
Does it calculate the EVM of each equalized received symbol with a closest ideal symbol on the constellation and then average it out to give final EVM value?