Originally posted by: SystemAdmin

[UDOPS said:]

I have been toying with some samples to learn OPL capabilities. The following sample will run. In the comments is an incomplete statement that I would want to produce a set of the lowest prices. Is there a way to compose this in a single line statement without creating additional named sets or arrays?

tuple domestic {
string name;
float price;
int source;
}

tuple import {
string name;
float price;
}

tuple retail {
string name;
float saleprice;
}

{domestic} skus = {<", 2.65, 1>, <", 1.75, 2>, <", 0.5, 2>, <", 1.5, 3>, <", 1.5, 2>, <", 1.6, 1>};

{import} europe = {<", 1.25>,<", 4.55>, <", 6.77>};

//merge sets (lists) on two columns
{retail} sale_all = {<f,x> | <f, x, src> in skus} union europe;

//intersection of sets, only members in both sets
{retail} sale_intersection = {<f,x> | <f, x, src> in skus, <f, p> in europe};

//set of all fruits, lowest price
{retail} sale_low = {<f.name, 1> | f in skus} union {<f.name, 2> | f in europe};

//incomplete attempt {retail} sale_low = {<f.name, min()> | f in sale_all};


//set of all fruits, highest price


//set of all fruits, average price

execute{ //display constructed sets, add carriage return between results
writeln("Intersection "+sale_intersection)
writeln("\nAll "+sale_all)
writeln("\nLow "+sale_low)
};
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

[afleischer said:]

Hi

what about

{retail} sale_low= {<f.name, min(<f.name,p> in sale_all) p> | f in sale_all};
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

[UDOPS said:]

Thanks, that works. Now how about the average?

tuple domestic {
string name;
float price;
string source;
}

tuple import {
string name;
float price;
}

tuple retail {
string name;
float saleprice;
}

{domestic} skus = {<", 2.65, "OH">, <", 1.75, "CA">, <", 0.5, "CA">, <", 1.5, "TX">, <", 1.5, "CA">, <", 1.6, "OH">};

{import} europe = {<", 1.25>,<", 4.55>, <", 6.77>};

//merge sets (lists) on three columns
{domestic} sale_all = skus union {<n,p,"ER">| <n,p> in europe};

//intersection of sets, only members in both sets
{retail} sale_intersection = {<f,x> | <f, x, src> in skus, <f, p> in europe};

//set of all fruits, lowest price
{retail} sale_low= {<f.name, min(<f.name,p,src> in sale_all) p> | f in sale_all};

//set of all fruits, highest price
{retail} sale_high= {<f.name, max(<f.name,p,src> in sale_all) p> | f in sale_all};


//set of all fruits, average price
{retail} sale_avg= {<f.name, sum(<f.name,p,src> in sale_all) p> | f in sale_all};

//card(A in sale_all: A.name==f.name) does not work

execute{ //display constructed sets, add carriage return between results
writeln("Intersection "+sale_intersection)
writeln("\nAll "+sale_all)
writeln("\nLow "+sale_low)
writeln("\nHigh "+sale_high)
writeln("\nAverage "+sale_avg)
};
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

[afleischer said:]

Hi

That could be expressed like this:

{retail} sale_avg= {<f.name, sum(<f.name,p,src> in sale_all) p/sum(<f.name,p2,src2> in sale_all) 1> | f in sale_all};

Alex

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

[UDOPS said:]

That is clever, it never would have occurred to me to sum 1's over the implied iteration of the sum() function.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

[UDOPS said:]

Here is another complex set construction that is not working, but I think you can see what I am attempting:


{trackexpand} TrackPlan = {<p.block, time, p.capacity, p.time_minimum, maxl(p.no_wait, &#91;o.no_wait | o in Orders: time>=o.time_from && time<=o.time_to])> | p in Profile, time in Ts..Te};

I want the "wait" element to be the maximum of the current slice and a matching slice of another set, Orders.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

[UDOPS said:]

I thought about this, and although I intend there only be one matching record in the set Orders, it occurred to me that there might be multiple records in a faulty data set, so I tried this:

{trackexpand} TrackPlan = {<p.block, time, p.capacity, p.time_minimum, maxl(p.no_wait, max(o in Orders: time>=o.time_from && time<=o.time_to) o.no_wait)> | p in Profile, time in Ts..Te};

and it worked. I guess this is the best method. I thought about "sum", but this is a binary value and I don't want it creeping out of range.
#DecisionOptimization
#OPLusingCPLEXOptimizer