# Using only positive frequencies in fourier domain, How will it affect the ifft?

I am going to do some kind of transformation and transform a data to another domain, and again back to the first domain. For this, I take a fourier transform of the data and separate the positive part and I use the positive frequencies only.

I know that for example in the frequency domain of a wavelet we have positive and negative frequencies. what will be the effect of using only positive frequencies?

• What do you do with the negative frequencies? Do you set their values to zero or do you reduce the original vector size? – Matt L. Jan 1 '15 at 17:46
• @MattL.I don't consider them. I reduce the original vector size – user3482383 Jan 2 '15 at 7:40
• The positive and negative frequency results combine to cancel out imaginary data components. So from positive frequencies only, you end up with complex data instead of real data when you transform back via a generic IFFT. – hotpaw2 Jan 4 '15 at 5:22
• Where are the "negative frequencies" costing you? They are just the complex conjugate of the positive ones. In addition, if you aren't too wedded to FFT then there is another transform that doesn't have negative frequencies. If your interested I will look it up; I have forgotten the name. – rrogers Jan 7 '15 at 14:38

Assuming you start with real time samples, then the FT is symmetric around $f=0$. If you discard the negative frequencies and then calculate the IFT, you won't get the same data samples back; in particular, they will be complex.