python - Comparing dict values and popping the keys -

i have big dictionary simplified looks this:

my_dict = {'a': [-33.27, -2.12, 5.23], 'b': [-57.11, 9.82, -26.13], ...}

so keys strings , values lists of floats.

what want reduce it's size running criterion finds , removes redundant key: value pairs.

the criterion in pseudo-code is:

for every key i, find whether different key j exists in dict such that:

• value_of_key_i[0] > value_of_key_j[0] and
• value_of_key_i[1] > value_of_key_j[1] and
• abs(value_of_key_i[2]) < abs(value_of_key_j[2])

what wrote task this:

to_remove = [] ilcs, iloads in running_load.items():     jlcs, jloads in running_load.items():         if iloads[0] > jloads[0] , iloads[1] > jloads[1] , abs(iloads[2]) < abs(jloads[2]):             # print(iloads, jloads)             to_remove.append(ilcs)             break  in to_remove:     running_load.pop(i) 

which work, explicitly traversing dict twice , necessary additional for-loop popping feels inefficient..

is there better way? more efficient generators and, let's say, any()?

p.s: problem approach cannot test equality since @ point values going tested against (and yes 1 check , continue but...)

instead of using list of elements remove, create new dictionary , add in elements kept.

i don't see way reduce o(n^2) complexity in initial phase (compare every element against every other element), mentioned above sorting some.

also, think generate 2 different lists identical content items in dictionary (one list in each for). shouldn't count compared "compare every item every other item", still helps.