algorithm - Working with large amounts of data in java: speed -
i want work 10k-100k data points in form of 16-tuples (x_1,...,x_16). of elements of tuple floats in [0,1], along 1 string , ints.
i want able lightning fast (preferrably <10ms) math operations on selected points of data. example: compute average of x_15 points satisfy: x_3 in [0.3,0.4] , x_5 > x_2.
my naive approach create class each tuple , math on classes. storage i'd write tuples text file when program finished , load them there when program starts.
is feasible , approach lightning fast?
it faster load tuples 2 dimensional array of floats rather 1 dimentional array of class instances, appears want doing lot of comparison between individual tuples (so have access class properties 100k times+ per query doing 1d array way)
Comments
Post a Comment