Matrix Indexing Being Slow

I'm running a function, let's say f(x), that depends on one variable (which may take any variable in the interval [0, 1]). This function gets called a lot of times (100,000+)

In my script, I can save a lot of calls of this function by storing f(0.5) for example, and then retrieving the stored value for f(0.5). I've tried to do this by creating a matrix A:

Row 1 = x

Row 2 = f(x)

However, my matrix indexing is taking a long time.

e.g. the below is taking 109 seconds with 11,000 calls

value = A(2, A(1, :) == 0.5);

Is there a more efficient way to do this? (Either the matrix indexing or the complete use of the function).

[Roger Stafford]

I think it is not the indexing that is slow. If you have accumulated a large number of values of the function in A, it is the scanning through all of these that takes the time. Writing A(1,:)==0.5 causes a scan of all different values which have been obtained.

Answer 1

[Adam]

Try using containers.Map.

You can set your values like 0.5 as keys and then use them to directly index into the map to retrieve the value

doc containers.Map

I haven't tried using it in a very dynamic way such as this or with such a large set of values so it may not be faster at all, but in theory a map which, by definition, has unique keys by which to look up a value should be what you want to do this. I have used maps for many things, but generally with < 50 elements in the map and those generally set at the start rather than being constantly dynamically added. Again, in theory, dynamically adding to a map should not cause large memory re-assignments and the like, but you would have to try it to see how effective it is.

[Groat]

Thank you for the suggestion Adam! I'd not heard of the containers.Map setup, but it's definitely quicker.