Numpy - Determine nearly equivalent floats before finding max value and index in 1D array-CodePudding

I have been stuck on this problem for hours.

Let's say we have a 1D array of floats: [0.12345678, 0.23456788, 0.23456789]

I just need to find the max value of the array and it's index. I know I can do that with np.max and np.argmax.

res = np.max(array)

index = np.argmax(array)

In this case, the answer would be [0.23456789, 2].

But what happens when I need to check for near equivalence first? Basically, 0.23456788 and 0.23456789 are nearly equivalent, so only the smallest instance of that value is used. Therefore, the answer would be [0.23456788, 1].

How exactly do I do this without using for or while loops? I figured I need to use np.isclose(), but where do I use it? I'm kind of new to numpy, so any help is appreciated.

Example 1:

Array: [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.9, 0.9, 0.9]
Expected Answer: [0.9, 7]
Explanation: 0.9 is the max value, and it first appears in index 7

Example 2:

Array: [0.11111111, 0.22222222, 0.33333333, 0.44444444, 0,44444445, 0.12345678]
Expected Answer: [0.44444444, 3]
Explanation: 0.44444445 is the max, but because 0.44444444 is nearly equivalent and we want the lowest value of that instance, the answer becomes [0.44444444, 3]

CodePudding user response：

Just round the values to the number of decimal places you care about. That number is application dependent and only you can determine what will work for "close enough".

> data = np.array([0.12345678, 0.23456788, 0.23456789])
> decimals = 3
> data.round(decimals)
array([0.123, 0.235, 0.235])

> np.max(data.round(decimals)), np.argmax(data.round(decimals))
(0.235, 1)