- 19 Sep, 2020 2 commits
-
-
Raymond Toy authored
Most if the fixes are due to computing the lengths (and indices) for LENSAV. Apparently the lengths should be log2(N) instead of log(N). But the fix for c1fm1f.f fixes a genuine bug in the complex fft which caused wrong results for non-power of 2 FFT sizes. Also fixed a couple of typos and incorrect return ier codes.
-
Raymond Toy authored
-
- 18 Sep, 2020 8 commits
-
-
Raymond Toy authored
This adds the asd file and the simple test program, based on the single-precision version, but updated to compile and test the double-precision version.
-
Raymond Toy authored
Check that the length of the array is consistent with the given n (fft length). This isn't perfect. Real FFTs of size 4 and 5 both produce transformed arrays of length 3, and there's no way to tell which is correct.
-
Raymond Toy authored
The test also picked up an error in convert-inverse-rfft where it wasn't handling FFTs of even length.
-
Raymond Toy authored
-
Raymond Toy authored
Reorder the functions too so the rfft and inverse-rfft are together and the support routines come before.
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
- 16 Sep, 2020 1 commit
-
-
Raymond Toy authored
-
- 23 Feb, 2010 3 commits