multiply_real_fft_gaussian (
        array,
        linewidths,
        shape,
        )

gaussians_1d ( linewidths,  shape )

gaussian_convolution ( data,  ijk_linewidths )

TODO: Optimization and stopability. This routine could be made much more memory efficient by performing the FFTs along each dimension separately and apply the 1-d FFT in just a 2-d plane at a time, instead of the entire 3-d data set. Temporary memory use could be reduced to the memory needed for one 2-d plane. This is important for handling larger data sets. I observe mild thrashing on a 256^3 volume on my Linux laptop with 1 Gbyte of memory. Working a plane at a time would also allow a progress callback to be invoked frequently and user-requested stop to be handled.

gaussian_convolve (
        volume,
        linewidth,
        use_subregion=True,
        use_step=False,
        )

Performs Gaussian convolution on current Volume Viewer data set producing a new data set that is shown in Volume Viewer.

The linewidth is specified in xyz space units and spans -1 SD to +1 SD.

ChimeraExtension.py

gui.py