Value costing: insertCoin, unionValue, scaleValue #7435
Conversation
InsertCoin
f$t - time |
UnionValueLinear model in sum of total sizes: not very good.
Quadratic model in square of sum of total sizes: pretty good!
|
ana-pantilie
changed the title
ScaleValueWith randomized quantities and scalars and randomized total size of value
|
ana-pantilie
force-pushed
the
ana/value-costing-final
branch
2 times, most recently
from
a892843 to
1e44958
Compare
ana-pantilie
force-pushed
the
ana/value-costing-final
branch
from
1e44958 to
abdc6ed
Compare
| | QuadraticInY OneVariableQuadraticFunction | ||
| | QuadraticInZ OneVariableQuadraticFunction | ||
| | QuadraticInXAndY TwoVariableQuadraticFunction | ||
| | SquareOfSum SquareOfTwoVariableSumFunction |
There was a problem hiding this comment.
This most likely won't be needed - a + bx + cy should be good enough.
There was a problem hiding this comment.
The SquareOfSum stuff is from before the recent experimentation for unionValue, so it's outdated.
a + bx + cy should be good enough.
I would assume a, b and c are the coefficients we are trying to determine while x and y are the sizes of the inputs, right? Then, as far as I understand, this translates to running R's linear regression algorithm for the t ~ x + y formula. @kwxm am I understanding what @zliu41 is suggesting correctly?
There was a problem hiding this comment.
Btw, just to be sure, I(x + y) means "find a coefficient for (x + y)" while x + y means "find a coefficient for x and a coefficient for y", right?
| ) | ||
| | pId <- policyIds | ||
| ] | ||
| amt = mkQuantity 10000000 |
There was a problem hiding this comment.
I don't think we need to use smaller integers at all - just always use maxBound / 2.
There was a problem hiding this comment.
I keep getting errors when using integers close to the maxBound, not sure why yet. This is the same issue I ran into when working on the scaleValue generator. These errors appear despite taking precautions to not allow the additions/multiplications to overflow. Since for unionValue the size of the integers doesn't matter I have postponed the investigation for a bit later, but it's on my TODO list.
Fixes https://github.com/IntersectMBO/plutus-private/issues/1899
Fixes https://github.com/IntersectMBO/plutus-private/issues/1900