This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision | ||
difx:difxdft [2011/07/01 16:30] johnmorgan |
difx:difxdft [2015/10/21 10:08] (current) |
||
---|---|---|---|
Line 1: | Line 1: | ||
+ | |||
====== DFT and FFT implementation in DiFX ====== | ====== DFT and FFT implementation in DiFX ====== | ||
Line 51: | Line 52: | ||
===Accuracy=== | ===Accuracy=== | ||
- | A basic test of accuracy was carried out by comparing the output visibilities from datasets generated different FFT sizes after averaging each down to 64 channels within mpifxcorr (only the central 32 were used for the analysis). After averaging down to the same number of channels, the plots of comparing the phase for 512 vs. 1024 channels are comparable with those for 512 vs. 64 channels. Amplitude is not shown here but there is no significant difference there either. | + | A basic test of accuracy was carried out by comparing the output visibilities from datasets generated different FFT sizes after averaging each down to 64 channels within mpifxcorr (only the central 32 were used for the analysis). After averaging down to the same number of channels, the plots of comparing the phase for 512 vs. 1024 channels are comparable with those for 512 vs. 640 channels. Scatter seems to increase somewhat. Amplitude is not shown here but they agree to ~1%. |
^ nFFTChan | ^ nFFTChan |