Model Hangs

[Rayfo...@aol.com]

Hello Christopher,

 

I'm having some difficulty getting a model to run.  Can you give me some pointers please.

 

Firefox 16.0.2 OxCal v4.2

 

I have 31 R_Dates in a Phase  (part of a larger model of overlapping phases).  The model 'hangs' after running for a long time.  The output shows dots at one end of the Start or  End boundaries and at one end of the Span.  A large spike shows at the end of the boundary and span, and some curious characters at the top left hand corner under the single plot window. 

 

Start SC

-1923

...

95.4

96.1

 

 

 

I've captured some of the warnings and errors from the Firefox error console (attached)

 

I ran several tests by commenting out  dates from the middle, one then three then five etc up to fifteen.  I found that commenting out 9 or more dates, the model converged and did not have the spike, i.e. a model with 22 or less dates ran OK, but not with more.

 

As the number of dates increases, the convergence gets erratic and the dots and spikes appear until finally with larger numbers, it hangs.

 

If I moved all the dates to a different area of the calibration curve, say by 500 years or 1,000 years, then the model ran OK, so perhaps there is a curve wiggle factor in there?

 

Any thoughts?

 

regards

 

Ray Kidd

 

Below is the model with three R_Dates commented out giving the result shown above.

 

 Plot()
 {
  Phase()
  {
   Sequence()
   {
    Boundary("Start SC");
    Phase("2")
    {
     R_Date("A1", 3580, 60);
     R_Date("A2", 3550, 35);
     R_Date("A3", 3545, 35);
     R_Date("A4", 3530, 50);
     R_Date("A5", 3525, 35);
     R_Date("A6", 3525, 35);
     R_Date("A7", 3520, 45);
     R_Date("A8", 3515, 35);
     R_Date("A9", 3510, 50);
     R_Date("A10", 3510, 35);
     R_Date("A11", 3490, 40);
     R_Date("A12", 3480, 40);
     R_Date("A13", 3480, 40);
     R_Date("A14", 3470, 40);
     //R_Date("A15", 3470, 40);  //these three commented out so this model has 28 dates
     //R_Date("A16", 3465, 40);
     //R_Date("A17", 3455, 40);
     R_Date("A18", 3450, 35);
     R_Date("A19", 3450, 40);
     R_Date("A20", 3435, 45);
     R_Date("A21", 3435, 35);
     R_Date("A22", 3435, 40);
     R_Date("A23", 3430, 60);
     R_Date("A24", 3410, 45);
     R_Date("A25", 3400, 40);
     R_Date("A26", 3390, 40);
     R_Date("A27", 3375, 40);
     R_Date("A28", 3370, 60);
     R_Date("A29", 3370, 35);
     R_Date("A30", 3360, 40);
     R_Date("A31", 3360, 40);
     Span("SCU");
    };
    Boundary("End SC");
   };
  };
 };

[Christopher Ramsey]

Ray

Interesting one this.

What is happening here is that there is a possibility (although small) that the phase is very short. The MCMC algorithm with the default settings uses the first few passes to determine the range of likely outputs. In this case it narrows the possibilities down too much - then in subsequent runs the program gets stuck at the ends of these limits producing the sharp spikes you see.

The problem can be eliminated by making sure that the program does more iterations each pass. For example the following code:

Options()
{
kIterations=1000;
};

Plot()
{
Phase()
{
Sequence()
{
Boundary("Start SC");
Phase("2")
{
R_Date("A1", 3580, 60);
R_Date("A2", 3550, 35);
R_Date("A3", 3545, 35);
R_Date("A4", 3530, 50);
R_Date("A5", 3525, 35);
R_Date("A6", 3525, 35);
R_Date("A7", 3520, 45);
R_Date("A8", 3515, 35);
R_Date("A9", 3510, 50);
R_Date("A10", 3510, 35);
R_Date("A11", 3490, 40);
R_Date("A12", 3480, 40);
R_Date("A13", 3480, 40);
R_Date("A14", 3470, 40);

R_Date("A15", 3470, 40);
R_Date("A16", 3465, 40);

R_Date("A17", 3455, 40);
R_Date("A18", 3450, 35);
R_Date("A19", 3450, 40);
R_Date("A20", 3435, 45);
R_Date("A21", 3435, 35);
R_Date("A22", 3435, 40);
R_Date("A23", 3430, 60);
R_Date("A24", 3410, 45);
R_Date("A25", 3400, 40);
R_Date("A26", 3390, 40);
R_Date("A27", 3375, 40);
R_Date("A28", 3370, 60);
R_Date("A29", 3370, 35);
R_Date("A30", 3360, 40);
R_Date("A31", 3360, 40);
Span("SCU");
};
Boundary("End SC");
};
};
};

Seems to run fine to completion. I suspect, if running online you were just getting the server to timeout as it has a time limit of a day and then just deletes the process.

The errors you showed are just for the plotting routine. I'm not quite sure why this is but the plotting routine seems to get confused about the axes when you have these spiky distributions that are obviously truncated. I will look into that different problem next time I look at the user interface.

Best wishes

Christopher

On 8 Sep 2013, at 16:18, Rayfo...@aol.com wrote:

> Hello Christopher,
>
> I'm having some difficulty getting a model to run. Can you give me some pointers please.
>
> Firefox 16.0.2 OxCal v4.2
>
> I have 31 R_Dates in a Phase (part of a larger model of overlapping phases). The model 'hangs' after running for a long time. The output shows dots at one end of the Start or End boundaries and at one end of the Span. A large spike shows at the end of the boundary and span, and some curious characters at the top left hand corner under the single plot window.
>

> Start SC <Raw.gif> -1923 ... 95.4 96.1
>
> <Untitled.jpg>

> --
> You received this message because you are subscribed to the Google Groups "OxCal" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
> <Error console.doc>

[Rayfo...@aol.com]

Hello Christopher,

 

Funnily enough I had tried increasing the k iterations, then discarded the idea.  Only to 50k, so I was too timid by far.

 

A useful lesson learned.

 

Many thanks

 

Ray