e3value user manual, first release

5.7 Transaction choice ratios

Figure 5.6 shows that the Business Traveler has a choice at each of its input and output ports. For example, it may buy 5 Train trips and 5 meals from Railway company 1. However, the logic of the choices allows three other transactions, not all of which are intended.

For example, the business Traveler may buy Train trips from Railway company 1 and food from Railway company 2. This is not intended but the diagram allows it. To rule it out, we need a transaction table.

Figure 5.6:Cardinalities in a network with transaction choice and merge. The labels and numbers computed by the editor are shown in comments signs.
Table 5.1:Transaction choices of the business Traveler in figure 5.6.





Actor

Tx

Ratio

Tf

Direction











Business Traveler

tx1

1

tf1

to Railway company 1

tf2

from Railway company 1

tf3

to Railway company 1

tf4

from Railway company 1





tx2

1

tf5

to Railway company 2

tf6

from Railway company 2

tf7

to Railway company 2

tf8

from Railway company 2





tx3

0

tf1

to Railway company 1

tf2

from Railway company 1

tf7

to Railway company 2

tf8

from Railway company 2





tx4

0

tf3

to Railway company 2

tf4

from Railway company 2

tf5

to Railway company 1

tf6

from Railway company 1





In a transaction table , we indicate which transactions are intended and in which ratio they occur. The table is part of the e3value model. For example, table 5.1 shows which transactions are intended in figure 5.6 and in which ratio. The table shows that the intended transactions occur an equal number of times.

The numbers in a transaction table are ratios. Transaction txi is chosen

∑-ri-
  j rj

times. In figure 5.6, transaction 2 is chosen 50% of the time, so the boundary elements in Company 2 are executed 125 times.

Figure 5.6 contains a transaction merge at the interface of Railway company 1. At a transaction merge, the numbers of the merged transactions are added, so the interface of Railway company 1 is triggered 126 times. These numbers are computed by the e3value analysis software and are shown as comments in figure 5.6.

Transaction tables specify the cardinality of transactions, and hence of transfers, which then override any cardinality specified at a port. For example, the company offers single Train trips to student Travelers, and indicates this with an explicitly specified port cardinality. This is the cardinality used in transactions with the student. In transaction with the business Traveler, the ratios specified for the transactions with the business Traveler are used.

The e3value tool contains a transaction analyzer, which in case of an ambiguous diagram like that in figure 5.6, generates a table of possible transactions and allows the user to select the ones that he or she intends. We explain this in detail in the next section.

5.7.1 Transaction dependency

Transaction dependencies have not yet been implemented.

In figure 5.7, A4 performs transaction tx2 = {t3, t4, t5, t6}with A1 and A2. However, A1 and A2 each can make a choice between tx2 and another transaction. This creates a transaction dependency between the choices of A1 and A2. Table 5.2 clarifies the situation.

Figure 5.7:Dependencies enforced by an interface.
Table 5.2:Dependent transaction choices of A1 and A2 in figure 5.7.





Actor

Tx

Ratio

Tf

Direction











A1

tx1

8

t1

to A3

t2

from A3





tx2

2

t3

to A4

t4

from A4











A2

tx2

3

t5

to A4

t6

from A4





tx3

4

t7

to A5

t8

from A5





For the sake of the example, in table 5.2 A1 chooses between transactions tx1 and tx2 in an 8:2 ratio and A2 chooses between transactions tx2 and tx3 in a 3:4 ratio. The dependency between A1 and A2 is that A1 must choose tx2 in the contract period exactly the same number of times as A2 chooses tx2 in the same period.

Suppose A1 has 60 need occurrences. Then it executes tx1 48 times and tx2 12 times. Due to the transaction dependency, this means that A2 executes tx2 12 times too. Together with the choice ratio of A2 this implies that A2 has 28 need occurrences, split in absolute numbers over 12 times tx3 and 16 times tx4.

The single interface of A4 thus implies the transaction dependency that

  r1           r3
r1 +-r2 ∗N1 = r3 +-r4 ∗ N2.

where ri is the ratio of transaction txi and Nj is the number of needs of Aj.