MATLAB: “permat” and more efficient coding

repeated elements

Dear all,

I have this list of countries

    listl={'ATl' 'BEl' 'ESTl' 'FRl' 'DEl' 'GRl' 'IEl' 'ITl' 'NETHl' 'PTl' 'SLOVAKl' 'SLOVENl' 'ESPl'};

and the following command

ixsL=repmat( [1: 2], [3,1] );
lol= arrayfun( @(ii) sprintf( 'ATl%u',ii), ixsL (:), 'uni', false );

As you can see the ATl inside the expression changes every time we refer to a new country.

For instance, if we refer to 'BEl' we have

lol= arrayfun( @(ii) sprintf( 'BEl%u',ii), ixsL (:), 'uni', false );

My goal is to create an "if statement" such that

*if listl(1) *to obtain the output of

lol= arrayfun( @(ii) sprintf( 'ATll%u',ii), ixsL (:), 'uni', false );

if listl(2) to obtain the output of

lol= arrayfun( @(ii) sprintf( 'BEl%u',ii), ixsL (:), 'uni', false );

I was thinking something like

lol= arrayfun( @(ii) sprintf( 'k%u',ii), ixsL (:), 'uni', false );

where k=listl(1) or k=listl(2).

this approach will save me my time instead of typing BEl or ATl inside the arrayfun

Best Answer

for k = 1 : length(listl)
  lol{k} = cellstr( num2str( ixsL(:), [listl{k} '%u'] );
end

Related Solutions

MATLAB: Creating a repeated matrix

This is a more "clever" way:

   ixs    = repmat( [1:6], [3,1] );
   ktotal = arrayfun( @(ii) sprintf('S%u',ii), ixs(:), 'uni', false );

However, a for-loop is probably the "best" way:

    ktotal = cell( 6*3, 1 );
    for ii = 1:6
        ix = 3*(ii-1)+1;
        ktotal( ix : ix+2 ) = {sprintf('S%u',ii)};
    end

MATLAB: GPU optimization of looped vector operations

This modification uses mat2tiles from the File Exchange, to help divide the computation into bigger, vectorized chunks

https://www.mathworks.com/matlabcentral/fileexchange/35085-mat2tiles-divide-array-into-equal-sized-sub-arrays

It runs in about 2 seconds on my graphics card (GeForce GTX 1080 Ti). Aside from increased vectorization, the key is to eliminate all the indexing expressions x([1:i-1, i+1:end]). Those are costly.

    tic;
    gnu_bar = gpuArray(nu_bar);
    gK = gpuArray(K);
    gdK = gpuArray(dK);
    gdN_KK = gpuArray(dN_KK);
    
    
    chunksize=1000;
    
    vv=gnu_bar.^2;
    vvchunks=mat2tiles(  vv , [1,chunksize]);
    numer=gnu_bar.*gK;
    c=1;
    
    for k=1:numel(vvchunks)
        Q=numer(:)./(vv.'-vvchunks{k});
        Q(c:len+1:end)=0;
        c=c+size(Q,2);
        vvchunks{k}=sum(Q,1);
    end
    
    gdN_KK=[vvchunks{:}];
    gdN_KK = 2*gdN_KK + gK./(2*gnu_bar) + gdK;    
    
    wait(gd)
    toc %Elapsed time is 2.027665 seconds.

Related Question