Open PaulBellette opened 8 years ago
The way that the code keeps track of the minimum modified cost is a little slow, the bottleneck when profiling is
function Base.getindex(cost::ModifiedCost, i, j) @inbounds return cost.m[i,j] - cost.row_offsets[i] - cost.column_offsets[j] end
This gets called very often in step_six! so it might be a better idea just to use broadcasting to keep updating the cost matrix instead
The way that the code keeps track of the minimum modified cost is a little slow, the bottleneck when profiling is
This gets called very often in step_six! so it might be a better idea just to use broadcasting to keep updating the cost matrix instead