[julia-users] Re: weird 2D fft error: returns NaN error
Aren't you just dividing by zero? I don't know what it means for both arrays to contain (0,0) but if there is an i,j such that X[i,j] == 0 Y[i,j] == 0, then sqrt(X.^2+ Y.^2) will have a zero at i,j. On Thursday, January 1, 2015 2:30:55 AM UTC-5, Zahirul ALAM wrote: There is a NaN item at the very centre. How to deal with it? On Thursday, 1 January 2015 02:25:34 UTC-5, Zahirul ALAM wrote: any help please On Thursday, 1 January 2015 01:07:33 UTC-5, Zahirul ALAM wrote: may be not On Thursday, 1 January 2015 01:05:02 UTC-5, Zahirul ALAM wrote: I found the reason because both arrays contains (0, 0). Is there way around it? On Thursday, 1 January 2015 00:50:56 UTC-5, Zahirul ALAM wrote: Happy new year!! I am encountering a very odd fft error. I am trying fft NxN data. The data is produced using a mathematical equation. After debugging I have found that the following expression is gives NaN error: fft(besselj(1, sqrt(X.^2+Y.^2))./sqrt(X.^2+Y.^2)) where X and Y are Array{Float64,2}. if I write fft(besselj(1, sqrt(X.^2+Y.^2)) I get write answer. Seems like the division with the array is the issue. is issue. any idea why? is it a bug?
[julia-users] Re: weird 2D fft error: returns NaN error
it is divide by zero error. I am sorry. I was way to drunk. Sorry everyone. maybe we need to think about a better way of handling such error. On Thursday, 1 January 2015 03:16:22 UTC-5, Gustavo Goretkin wrote: Aren't you just dividing by zero? I don't know what it means for both arrays to contain (0,0) but if there is an i,j such that X[i,j] == 0 Y[i,j] == 0, then sqrt(X.^2+ Y.^2) will have a zero at i,j. On Thursday, January 1, 2015 2:30:55 AM UTC-5, Zahirul ALAM wrote: There is a NaN item at the very centre. How to deal with it? On Thursday, 1 January 2015 02:25:34 UTC-5, Zahirul ALAM wrote: any help please On Thursday, 1 January 2015 01:07:33 UTC-5, Zahirul ALAM wrote: may be not On Thursday, 1 January 2015 01:05:02 UTC-5, Zahirul ALAM wrote: I found the reason because both arrays contains (0, 0). Is there way around it? On Thursday, 1 January 2015 00:50:56 UTC-5, Zahirul ALAM wrote: Happy new year!! I am encountering a very odd fft error. I am trying fft NxN data. The data is produced using a mathematical equation. After debugging I have found that the following expression is gives NaN error: fft(besselj(1, sqrt(X.^2+Y.^2))./sqrt(X.^2+Y.^2)) where X and Y are Array{Float64,2}. if I write fft(besselj(1, sqrt(X.^2+Y.^2)) I get write answer. Seems like the division with the array is the issue. is issue. any idea why? is it a bug?
[julia-users] Re: weird 2D fft error: returns NaN error
You had 0/0. Maybe you want an exception to be raised, but floating point standard has that NaN is the result. Numpy has an optional warning message, if I remember, but I don't know if there is a performance penalty for checking every division floating point computation.
[julia-users] Re: weird 2D fft error: returns NaN error
I found the reason because both arrays contains (0, 0). Is there way around it? On Thursday, 1 January 2015 00:50:56 UTC-5, Zahirul ALAM wrote: Happy new year!! I am encountering a very odd fft error. I am trying fft NxN data. The data is produced using a mathematical equation. After debugging I have found that the following expression is gives NaN error: fft(besselj(1, sqrt(X.^2+Y.^2))./sqrt(X.^2+Y.^2)) where X and Y are Array{Float64,2}. if I write fft(besselj(1, sqrt(X.^2+Y.^2)) I get write answer. Seems like the division with the array is the issue. is issue. any idea why? is it a bug?
[julia-users] Re: weird 2D fft error: returns NaN error
may be not On Thursday, 1 January 2015 01:05:02 UTC-5, Zahirul ALAM wrote: I found the reason because both arrays contains (0, 0). Is there way around it? On Thursday, 1 January 2015 00:50:56 UTC-5, Zahirul ALAM wrote: Happy new year!! I am encountering a very odd fft error. I am trying fft NxN data. The data is produced using a mathematical equation. After debugging I have found that the following expression is gives NaN error: fft(besselj(1, sqrt(X.^2+Y.^2))./sqrt(X.^2+Y.^2)) where X and Y are Array{Float64,2}. if I write fft(besselj(1, sqrt(X.^2+Y.^2)) I get write answer. Seems like the division with the array is the issue. is issue. any idea why? is it a bug?
[julia-users] Re: weird 2D fft error: returns NaN error
any help please On Thursday, 1 January 2015 01:07:33 UTC-5, Zahirul ALAM wrote: may be not On Thursday, 1 January 2015 01:05:02 UTC-5, Zahirul ALAM wrote: I found the reason because both arrays contains (0, 0). Is there way around it? On Thursday, 1 January 2015 00:50:56 UTC-5, Zahirul ALAM wrote: Happy new year!! I am encountering a very odd fft error. I am trying fft NxN data. The data is produced using a mathematical equation. After debugging I have found that the following expression is gives NaN error: fft(besselj(1, sqrt(X.^2+Y.^2))./sqrt(X.^2+Y.^2)) where X and Y are Array{Float64,2}. if I write fft(besselj(1, sqrt(X.^2+Y.^2)) I get write answer. Seems like the division with the array is the issue. is issue. any idea why? is it a bug?
[julia-users] Re: weird 2D fft error: returns NaN error
There is a NaN item at the very centre. How to deal with it? On Thursday, 1 January 2015 02:25:34 UTC-5, Zahirul ALAM wrote: any help please On Thursday, 1 January 2015 01:07:33 UTC-5, Zahirul ALAM wrote: may be not On Thursday, 1 January 2015 01:05:02 UTC-5, Zahirul ALAM wrote: I found the reason because both arrays contains (0, 0). Is there way around it? On Thursday, 1 January 2015 00:50:56 UTC-5, Zahirul ALAM wrote: Happy new year!! I am encountering a very odd fft error. I am trying fft NxN data. The data is produced using a mathematical equation. After debugging I have found that the following expression is gives NaN error: fft(besselj(1, sqrt(X.^2+Y.^2))./sqrt(X.^2+Y.^2)) where X and Y are Array{Float64,2}. if I write fft(besselj(1, sqrt(X.^2+Y.^2)) I get write answer. Seems like the division with the array is the issue. is issue. any idea why? is it a bug?