Think of this as the type which the spider algorithm will operate on.
-At this point, we have an orbit of angles, kneading sequence, internal address, oriented hubbard tree. I think we include all of this in the type. This gives me the image of a 'snowball' approach, where each calculation add onto the pile of associated data.
-How much can be done in-place to avoid allocations? The legs will be mutated at each step, and if we do it in the correct order can we avoid making another big legs vector?
Think of this as the type which the spider algorithm will operate on. -At this point, we have an orbit of angles, kneading sequence, internal address, oriented hubbard tree. I think we include all of this in the type. This gives me the image of a 'snowball' approach, where each calculation add onto the pile of associated data. -How much can be done in-place to avoid allocations? The legs will be mutated at each step, and if we do it in the correct order can we avoid making another big legs vector?