Get indices of matching values (vectorized find function)

I want to find a function that does a "vectorized find" that would do the below:

>> a = rand(6, 1)

a =

    0.4387
    0.3816
    0.7655
    0.7952
    0.1869
    0.4898

>> b = randi(6, [6, 1])

b =

     3
     4
     5
     5
     2
     5

>> c = a(b)

c =

    0.7655
    0.7952
    0.1869
    0.1869
    0.3816
    0.1869

>> b_reconstructed = somefunction(a, c)  % What is somefunction?

I know I can do something like

b_reconstructed = zeros(size(a));
for i = 1:numel(c)
    b_reconstructed(i) = find(a == c(i), 1, 'first');
end

But that seems very inefficient if a, b, and c are very large (>10k rows)

Answer 1

[Dyuman Joshi]

Use ismembertol when dealing with floating point numbers, ismember when dealing with integers.

a = rand(6, 1)

a = 6×1

0.9147 0.9363 0.2163 0.6877 0.0031 0.8379

b = randi(6, [6, 1])

b = 6×1

4 1 3 1 5 4

c = a(b)

c = 6×1

0.6877 0.9147 0.2163 0.9147 0.0031 0.6877

B = zeros(size(a));
for i = 1:numel(c)
    B(i) = find(a == c(i), 1, 'first');
end
B

B = 6×1

4 1 3 1 5 4

%Indices of elements in 'a' for each element in 'c' that is member of 'a'
[~,idx] = ismembertol(c,a)

idx = 6×1

4 1 3 1 5 4

[Matt J]

However, tolerances shouldn't be necessary if it is known, as in the example, that c is an exact subset of a.

Answer 2

[Matt J]

[~,b_reconstructed]=ismember(c,a)