Select/Delete with Sublist elements?

Question

Probably easy and short question, I still didn't fully figure out how to easily select/delete sublists from a list. Example:

tt = {{2, 4}, {4, 8}};

I want to delete/select all the elements where the difference between them (Abs) is greater than four.

Answer 1

I would use either

Cases[data, {x_, y_} /; Abs[x - y] > 4]

or

With[{diff = Abs[data[[All, 1]] - data[[All, 2]]] - 4}, 
   Pick[data, UnitStep[diff]*Unitize[diff], 1]
]

The first clearly demonstrates what you are trying to do, the second is much faster...

data = RandomInteger[{0, 100}, {10^6, 2}];

(m1 =  Cases[data, {x_, y_} /; Abs[x - y] > 4]); // AbsoluteTiming

==> {2.8393092, Null}

(m2 = 
    With[{diff = Abs[data[[All, 1]] - data[[All, 2]]] - 4}, 
     Pick[data,UnitStep[diff]*Unitize[diff], 1]]); // AbsoluteTiming

==> {0.1248024, Null}

m1 == m2

==> True

Assuming you are working with random data and don't expect equality, the second method can be simplified to...

Pick[data, UnitStep[Abs[data[[All, 1]] - data[[All, 2]]] - 4], 1]

Edit:

Of course if you want to delete elements where Abs[x-y] > 4 you can modify the Cases definition as

Cases[data, {x_, y_} /; Abs[x - y] <= 4]

And the Pick method by swapping out the 1 in the last argument with 0.

Answer 2

You can select all sublists the elements of which differ by more than 3 as follows:

Select[tt, Abs[#[[1]] - #[[2]]] > 3 &]
(*{{4, 8}}*)

or using conditional patterns:

Cases[tt, x_ /; Abs[x[[1]] - x[[2]]] > 3]

(and you can delete them by either selecting those Not satisfying the condition, or using DeleteCases).

You could also do

Scan[If[Abs[#[[1]] - #[[2]]] > 3, Sow[#]] &, tt] // Reap // 
  Last // Last

or even

MapThread[If[Abs[#1 - #2] > 3, Sow[{#1, #2}]] &, Transpose@tt]; // 
   Reap // Last // Last

Finally using Compile and Internal`Bag, as described by Andy here, you can do

cs = Compile[{{lst, _Integer, 2}},
   Module[{bag = Internal`Bag[], l = Length@lst},
    Do[
     If[
      Abs[lst[[i, 1]] - lst[[i, 2]]] > 3.,
      Internal`StuffBag[
       bag,
       lst[[i]], 1
       ]
      ],
     {i, l}
     ];
    Partition[Internal`BagPart[bag, All], 2]],
   CompilationTarget -> "C",
   RuntimeOptions -> "Speed"
   ];

which is roughly as fast as the Pick approach described in another answer (neither unpacks).

Answer 3

Cases[tt, {a_, b_} /; Abs[a - b] >= 4]

 {{4, 8}}

or

DeleteCases[tt, {a_, b_} /; Abs[a - b] < 4]

 {{4, 8}}

The latter approach could be better if pairs Abs[a-b] > 4 are generic.

Answer 4

I would also choose Pick for speed, but I would formulate it a little differently:

Pick[
  tt,
  UnitStep[Abs[# - #2] & @@ Transpose[tt] - 5],
  1
]

To pick the complement change the third argument from 1 to 0.

Answer 5

Although not as fast as Pick, another possibility is to use GatherBy which allows selecting/deleting subsets of data based on a condition:

  data = {{2, 4}, {4, 8}, {5, 9}, {1, 3}};
  {deleted,selected}= GatherBy[data, UnitStep[Abs[#[[1]] - #[[2]]] - 4] &]

returns:

   {{{2, 4}, {1, 3}}, {{4, 8}, {5, 9}}}