Studer Gabriel
authored
The TreeSolve algorithm requires to store a LOT of partial solutions, that have all been stored as single std vectors. a LOT can mean several millions, which gives a memory overhead in the range of several hundred MB for the std vectors alone, thats much more than the data they actually contain. The new implementation contains ONE std vector that contains all data and some indexing magic to access it. Please note, that the size of this vector has an upper bound largely controlled by the max_complexity parameter in the TreeSolve algorithm.