For-loop between global tables

Hi all!

I want to update a specific global table (Demand_x) each time a replication in the Experimenter starts. The update comes from other global tables that are defined in the model (Demand_1, Demand_2,...).

In the "Start of Replication"-Trigger of the Advanced Tab in the Experimenter I put:

/**INPUT PARAMETER*/
double replication = parval(1);
double scenario = parval(2);
for(replication=1;replication<=10;replication++)
{
	for (int period=1; period<=gettablerows("Demand_1"); period++)
	{
	for (int product=1; product<=gettablecols("Demand_1"); product++)
	{
	settablenum("Demand_x",period,product, gettablenum("Demand_1",period,product));
	}
	}
}

This code updates me my global table "Demand_x" with the input data from "Demand_1". But I have ten other global tables (Demand_2,..., Demand_10) that should be used for updating "Demand_x" each time a replication starts.

How can I realize this? I thought of entering another 'for-loop' that loops between the global tables but I don't know how.

Thank you very much for your advice!
Simon

Do you want to aggregate the 10 demand profiles into one table?

If so, what do you want to do with the replications? If all the replications are reading from Demand_x, why not do that on start of experiment?

If you just want to agregate them based on the table name you could use this:

int numdemandtables=10;
clearglobaltable("Demand_x");
for (int demand=1;demand<=numdemandtables;demand++) {
	treenode demandtable=reftable(concat("Demand_",numtostring(demand)));
	for (int period=1; period<=gettablerows(demandtable); period++)
	{
		for (int product=1; product<=gettablecols(demandtable); product++)
		{
			inc(gettablecell("Demand_x",period,product), gettablenum(demandtable,period,product));
		}
	}
}

Thank you Jason!

In my model I have different scenarios and I want to test these scenarios under different demand patterns. That's why I want to replicate my model several times.

I came myself to a solution (it is not very nice and maybe nasty but at least it seems that it works out):

/**INPUT PARAMETER*/
double replication = parval(1);
double scenario = parval(2);
string Tablename;
	Tablename = concat("Demand_",numtostring(replication));
		for (int period=1; period<=gettablerows(Tablename); period++)
		{
			for (int product=1; product<=gettablecols(Tablename); product++)
			{
				settablenum("Demand_x",period,product,gettablenum(Tablename,period,product));
			}
		}

I am using the replication pointer that counts each time a replication is performed and change it to a string. Not very nice but doable.

But this is only a workaround of my original problem. I wanted my original global table "Demand_x" to update ITSELF before a replication is done. I don't want to replace the data by other data from another global table. This is done by my workaround now.
I want the Experimenter to generate, each time a repliacation starts, a new demand profile (e.g. generally distributed, 15 products, 5000 periods) by a random number generator. Where do I have to code this? Again in the "Start of replication" trigger? Are there any examples, where I can get an idea of how this code might look?

Thank you very much!

Simon

Again, I found a solution to my above stated problem:

For each replication in the Experimenter I wanted to generate a new demand pattern for my model. Here is the code:

/**INPUT PARAMETER*/
double replication = parval(1);
double scenario = parval(2);
treenode NumberofReplication = node("/Tools/Experimenter/NumberOfRunsPerScenario", model());
int R = getnodenum(NumberofReplication);
int i = replication;
for(i=1; i<=R; i++)
{	
	for (int period=1; period<=gettablerows("Demand"); period++)
	{
		for (int product=1; product<=gettablecols("Demand"); product++)
		{
			settablenum("Demand",period,product,fabs(round(normal(5,2,1))));
		}
	}	
}

Maybe somebody is in need of something similar...

Have fun!
Simon

Hi

I was trying to do likewise, but I want to update the pattern by the scenario.

It seems the value of scenario is not functioning. Any ideas?

Currently I used a Global Variable as variable and assign a value to it. Though it works, but just to check anything wrong with the scenario variables at the Start of Run of Experiment.