Fork me on GitHub

CAEV arithmetics

In general, the CAEV operation ADD is NOT commutative, instead you must obey chronological order. Same for SUB, the latest CAEV added must be subbed first (LIFO), this is the right inverse of ADD, called ctl_caev_sub(); to get the left inverse of ADD, respectively, the FIFO-sub, use ctl_caev_sup().

The arithmetic we implement here favours additive operations over multiplicative. In our (raw) syntax we will write x+p<-q meaning add x to the price, then, if p is non-zero, multiply by p, then for non-zero q divide by q.

Add for additive operations is defined as:

x1+0<-0 + x2+0<-0 = (x1+x2)+0<-0

and hence the inverse of x+0<-0 is (-x)+0<-0.

Similarly, add for multiplicative operations is defined as:

0+p1<-q1 + 0+p2<-q2 = 0+(p1p2)<-(q1q2)

which makes 0+q<-p the inverse of 0+p<-q, with 0+1<-1 being the neutral element.

For mixed operations the precedence rule means:

x1+p1<-q1 + x2+p2<-q2 = (x1+x2q1/p1)+(p1p2)<-(q1q2)

So a dividend before a split becomes

d+0<-0 + 0+p<-q = d+p<-q

and a split before a dividend is summed as

0+p<-q + d+0<-0 = (dq/p)+p<-q

This arithmetic satisfies dilution properties just properly, a dividend of 4 paid before a 2:1 split is the same as a dividend of 2 paid after a 2:1 split:

10@20         10@20
20@10         10@16 + 10*4
20@8 + 20*2   20@8  + 10*4
=160 + 40     =160  + 40

And so DVCA/4 + SPLF/2:1 = SPLF/2:1 + DVCA/2 or in raw syntax -4+0<-0 + 0+1<-2 = -4+1<-2 vs 0+1<-2 + -2+0<-0 = -4+1<-2

From this the two inverses of + become apparent, since the additive part is always absolute in value, subtraction of additive operations from the left is simply an addition.

x1+p1<-q1 -l x2+p2<-q2 := (p2(x1 - x2)/q2)+(p1/p2)<-(q1/q2)

And since all multiplicative operations are aggregated into the multiplicative cell of a sum, subtraction of additive operations from the right is defined as:

x1+p1<-q1 -r x2+p2<-q2 = (x1 - (q1/q2)x2/(p1/p2)+(p1/p2)<-(q1/q2)

The actual inverse elements can be computed via left or right subtraction from the neutral operation 0+0<-0:

0+0<-0  -l  x+p<-q = (-px/q)+q<-p

and

0+0<-0  -r  x+p<-q = (-px/q)+q<-p

respectively, which we simply call inv or inverse.

Additionally, we provide a reverse operation defined by

rev(x+p<-q) := -x+q<-p

and along with

act(x+p<-q, P) := p(P + x)/q

which we call action of P under x+p<-q, or we say x+p<-q acts on P, we can define the inverse in terms of act and rev as

inv(x+p<-q) = rev(act(x+p<-q))

where act(x+p<-q) is the operator that maps x to px/q. */