Preferring a 64 byte prefix over the current 8-bit prefix for both internal and leaf nodes.

[harsh4786]

For the merkle tree, we currently have a prefix concatenated to the preimage as described below:

• 0x00: leaf node
• 0x01: branch node with 1 child
• 0x02: branch node with 2 children

As per our discussions here we should consider concatenating a 64 byte prefix to the preimage, as a measure to optimize since the internal block size is of 64 bytes and this would maintain alignment between the two blocks of the internal state per node.

[samkim-crypto]

Okay, sounds good. In terms of the proof though, it is probably best to hide these details that are implementation specific. Otherwise, the proof will become very involved and unreadable. We can either add it to the end of the doc or probably just not include the implementation details in latex and just add it to simd.

[harsh4786]

Yeah makes sense to keep it separate from the proof.

[harsh4786]

On a second note, i feel we shouldn't congest the simd with implementation details, we can just mention in this proof about the 64byte prefix since this proof formalizes the commitment scheme. What do you think? @samkim-crypto