classification 3D unable to decrease criterion along line search direction in mdscale

[jkfindeisen]

Trying to run the 3D demo with fewer particles (first 30 particles) and get

Warning: Padding initial configuration with zeros to achieve full dimension. 
> In mdscale (line 374)
In MDS_Clustering3D (line 21)
In Run_Classification3D (line 16)
In Demo_JCC_3D (line 74) 
Warning: Padding initial configuration with zeros to achieve full dimension. 
> In mdscale (line 374)
In MDS_Clustering3D (line 21)
In Run_Classification3D (line 16)
In Demo_JCC_3D (line 74) 
Error using mdscale>MDS (line 571)
Unable to decrease criterion along line search direction.

Error in mdscale (line 388)
            [Y,stress] = MDS(Y0,dissimilarities,weights,stressFun,metric,weighted,options);

Error in MDS_Clustering3D (line 21)
mds = mdscale(D,dimention,'Criterion','metricstress');

Error in Run_Classification3D (line 16)
        [clus,clusterfull]=MDS_Clustering3D(MatrixAfterAll2all_norm,nc,minClustSize,dimention);% First Classify

Error in Demo_JCC_3D (line 74)
Run_Classification3D

Is 3D averaging also possible with low number of particles?

[wexw]

Hi, You can try to change the parameter: dimension in the Demo Code dimention=30;%round(N/6)*10; If classification doesn't work, try different dimention values Now the dimension=30 which is too large for 30 particles. Maybe you can try dimention from 2 to 10

[jkfindeisen]

Setting dimention to 2 or 5 or 10 doesn't help.

[wexw]

Hi, we have uploaded a new [MDS_Clustering3D.m] document. (https://github.com/wexw/Joint-Registration-of-Multiple-Point-Clouds-for-Fast-Particle-Fusion-in-Localization-Microscopy/blob/main/3dcode/Functions_3D/MDS_Clustering3D.m). There you can read which parameters are changed to make it work. Hope this solves the issue.