Nicholas Solem

I am trying to get phase from real and imaginary parts using atan2 instead of cartopol, as multiple resources said this is possible. However, the results are not identical to the use of cartopol. Not only that, but when the original signal is reconstructed with the use of atan2, it is not exactly the same as the original.

I have attached a max patch demonstrating this on a simple bit of data. In that example, the error is pretty small, but in my actual patch where I need this to work, the error is as great as involving what appears to be a phase reversal.

Am I doing something wrong, or is there some way to improve the performance of atan2?

The answer cannot simply be, 'if cartopol works, just use that instead of atan2'. I cannot do this because I am actually creating a javascript file to do a bunch of operations on a buffer, and javascript has atan2 but does not have cartopol.

I have attached a max patch demonstrating this on a simple bit of data. In that example, the error is pretty small, but in my actual patch where I need this to work, the error is as great as involving what appears to be a phase reversal. 

The answer cannot simply be, 'if cartopol works, just use that instead of atan2'. I cannot do this because I am actually creating a javascript file to do a bunch of operations on a buffer, and javascript has atan2 but does not have cartopol.


using-atan2-for-phase-is-not-perfectaccurate

that the result of a custom math code differs from some externals object´s result does not mean your code is less accurate, it can also be the opposite.

it can depend on 32bit/64bit resolution or on the order you are calculating something.

however, if you need things to be "in sync", dont mix different methods/objects in one patch.

however, if you need things to be "in sync",  dont mix different methods/objects in one patch.


atan2 is just fine. the patch has a simple right to left problem. watch out for order of execution - trigger is your friend.

Using atan2 for phase is not perfect/accurate