Hello all

Working on a deep dive of Bulletproofs+ today

Some parts seem to switch up notation for no good reason :(

it might be better to contact them first with a list of general questions before doing a deep dive, since the efficiency gains don't seem huge

sarang: have you seen the Tari Labs stuff on amortization of the bulletproofs inner-product proof?

it's funny how the inner-product proof is becoming the focus area

The IPP is where the magic happens

Size benefits, delaying verification to a single round

for those that haven't seen it

Looks like the Bulletproofs+ construction should be ever-so-slightly faster than Bulletproofs overall

It's not clear why the authors saw a slight performance hit compared to the dalek BP library

And, as has been pointed out, it saves 96 bytes per proof, regardless of the number of aggregated outputs contained therein

Also interesting is that the prover is more efficient as well

so you think the verification should be more efficient too?

Negligibly, but yes

RIP matrix apparently

Maybe the dalek library is better optimized

Most likely

Would depend on precisely how they implemented BP+