please confirm buggy behaviour of ifft_real

[Klaas Hofstra]

The following code exhibits buggy behavior:
int N = 256;

cvec f;
vec t;

fft_real(randn(N), f);

cvec tmp = f;

ifft_real(f, t);

cvec d = tmp - f;
cout << d << endl;

For N < 256 the vector 'd' is the zero vector as expected. However, for N >= 256 the first half of the vector 'd' is non-zero.
I'm linking with FFTW 3.

I've tested this on two different machines. Could you please confirm the behaviour of this code so I can make a bug report?

Cheers,

Klaas Hofstra

[Adam Piątyszek]

Hi Klaas,

I will try to look into this problem in a few days, but can't promise because my laptop was stolen this morning. So I need some time to set up my working environment...

BR,
/ediap

[Adam Piątyszek]

OK. I have tested this on my installation with FFTW3 and it seems to be a bug. The content of `f' should not be modified when using ifft(const vec &f, vec &t). Please submit a bug report so we won't forget about this issue.

Thanks in advance!
/ediap