Cross-referencing dates without exporting?

225 views
Skip to first unread message

Erik Marsh

unread,
Dec 20, 2021, 2:57:19 PM12/20/21
to OxCal
Hi all,

I have two models – the first has multiple stratigraphic set ups. The second groups some of these same dates into phases based on ceramic style. If I cross-reference the dates, then their posteriors are affected by their relationships in both models (as far as I understand it). However, I do not want the phases of ceramic styles to affect the posteriors in the stratigraphic models.

I know one awkward way to do this: I export each modeled result from the stratigraphic model, one date at a time, and then build new models with these saved files.

Is there a better or easier way?

Thanks
Erik

RAY KIDD

unread,
Dec 20, 2021, 3:46:02 PM12/20/21
to ox...@googlegroups.com
Hi Erik,
Remembering I no longer have OxCal on my laptop. ....
Are you making phases based on your ceramic styles then a stratigraphic model within these phases?  I’m not sure why cross reference is coming into it.  Unless I’m having too much happy juice.
Best wishes 
Ray

Sent from my iPhone
--
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.
To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/c1cfcf7c-07f8-49ad-91ce-40838b37326fn%40googlegroups.com.

MILLARD, ANDREW R.

unread,
Dec 21, 2021, 3:18:30 AM12/21/21
to ox...@googlegroups.com

Hi Erik,

 

Do you mean you want to run a model with stratigraphic constraints, and then you also want to run a model with stratigraphic and ceramic style constraints? Or is there a stratigraphic model and a ceramic model which are grouping the same set of dates in different ways? If you mean the second then there is no need to cross-reference, just repeat the dates. If the first, then exporting more than one distribution from a model loses all the information about constraints and correlations that were in the output, and you are better off with a simple and complex model that you can compare.

 

Best wishes

Andrew

--

Dr. Andrew Millard

Associate Professor of Archaeology,

Durham University, UK

Email: A.R.M...@durham.ac.uk 

Personal page: https://www.dur.ac.uk/directory/profile/?id=160

Scottish Soldiers Project: https://www.dur.ac.uk/scottishsoldiers

Dunbar 1650 MOOC: https://www.futurelearn.com/courses/battle-of-dunbar-1650

 

 

From: ox...@googlegroups.com <ox...@googlegroups.com> On Behalf Of Erik Marsh
Sent: 20 December 2021 19:57
To: OxCal <ox...@googlegroups.com>
Subject: Cross-referencing dates without exporting?

 

[EXTERNAL EMAIL]

--

Erik Marsh

unread,
Dec 21, 2021, 7:20:08 AM12/21/21
to OxCal
Andrew, I meant the first: a model with stratigraphic constraints, and then a model with both stratigraphic and ceramic style constraints.
I should have included an example – here are three stratigraphic setups, and then a ceramic phase that uses the same dates.
I want the ceramic phase dates to be adjusted by both the stratigraphic and ceramic style constraints. But at the same time, I want to be sure the dates in the stratigraphic models are not affected by the ceramic style constraints. If I cross-reference, as in the example, it seems that they will be.
Hopefully my question is clearer!
Erik

Sequence("Stratigraphic 1")
{
R_Date("A");
R_Date("B");
R_Date("C");
};

Sequence("Stratigraphic 2")
{
R_Date("E");
R_Date("F");
R_Date("G");
};

Sequence("Stratigraphic 3")
{
R_Date("H");
R_Date("I");
R_Date("J");
};

Phase("Ceramic Style 1")
{
Date("=A");
Date("=E");
Date("=H");
};


Thomas S. Dye

unread,
Dec 21, 2021, 12:17:17 PM12/21/21
to ox...@googlegroups.com
Aloha Erik,

What ceramic style constraints do you anticipate? I don't see
any in your example.

All the best,
Tom

Erik Marsh <erik....@gmail.com> writes:

> Andrew, I meant the first: a model with stratigraphic
> constraints, and then
> a model with both stratigraphic and ceramic style constraints.
> I should have included an example – here are three stratigraphic
> setups,
> and then a ceramic phase that uses the same dates.
> I want the ceramic phase dates to be adjusted by both the
> stratigraphic and
> ceramic style constraints. But at the same time, I want to be
> sure the
> dates in the stratigraphic models are *not* affected by the
>> *From:* ox...@googlegroups.com <ox...@googlegroups.com> *On
>> Behalf Of *Erik
>> Marsh
>> *Sent:* 20 December 2021 19:57
>> *To:* OxCal <ox...@googlegroups.com>
>> *Subject:* Cross-referencing dates without exporting?
>>
>>
>>
>> *[EXTERNAL EMAIL]*
>> <https://groups.google.com/d/msgid/oxcal/c1cfcf7c-07f8-49ad-91ce-40838b37326fn%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>


--
Thomas S. Dye
https://tsdye.online/tsdye

Erik Marsh

unread,
Dec 22, 2021, 7:57:03 AM12/22/21
to OxCal
That's true Thomas – the ceramic phase should be: 

Sequence()
{
Boundary("Start");
Phase("Ceramic Style 1")
{
Date("=A");
Date("=E");
Date("=H");
};
Boundary("End");
};

Thomas S. Dye

unread,
Dec 22, 2021, 1:13:59 PM12/22/21
to ox...@googlegroups.com
Aloha Erik,

If you anticipate the effects of assuming a uniform span prior
introduced by the Boundary() command, and want to avoid them,
then, if I understand correctly, you should be able to set the
UniformSpanPrior option, which is TRUE by default, to FALSE.

If your actual model includes other calls to Boundary(), and you'd
like them to use the uniform span prior, then as an alternative to
setting the UniformSpanPrior option to FALSE, you might
investigate using the First() and Last() commands instead of the
Boundary() commands with Phase("Ceramic Style 1").

Hope this helps,

Christopher Ramsey

unread,
Dec 22, 2021, 1:34:30 PM12/22/21
to OxCal group
Ideally you use the Boundaries in the sequence which defined the groupings.

You don't wish to apply the uniform span more than once so if you wish to avoid this in the cross referenced sequences you can just use the "Date" function. You can do this either way round - so you might for example choose to set all of your Ceramic phases as separate groups as Erik suggests below (but with the original dates in). Then in the main model don't use any boundaries.

For ceramic phases a trapezium may be more appropriate than a uniform.

If you don't apply the uniform span prior at all though you will overestimate the length of all the groups because of the statistical scatter.

Best wishes

Christopher
> To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/87v8zgik2k.fsf%40tsdye.online.

Thomas S. Dye

unread,
Dec 22, 2021, 2:14:39 PM12/22/21
to ox...@googlegroups.com
Aloha Christopher,

So this?

Sequence("Stratigraphic 1")
{
Boundary("Start Stratigraphic 1")
R_Date("A");
R_Date("B");
R_Date("C");
Boundary("End Stratigraphic 1")
};

Sequence("Stratigraphic 2")
{
Boundary("Start Stratigraphic 2")
R_Date("E");
R_Date("F");
R_Date("G");
Boundary("End Stratigraphic 2")
};

Sequence("Stratigraphic 3")
{
Boundary("Start Stratigraphic 3")
R_Date("H");
R_Date("I");
R_Date("J");
Boundary("End Stratigraphic 3")
};

Sequence()
{
Date("Start Ceramic Style 1");
Phase("Ceramic Style 1")
{
Date("=A");
Date("=E");
Date("=H");
};
Date("End Ceramic Style 1");
};

All the best,
Tom

Christopher Ramsey

unread,
Dec 22, 2021, 4:48:00 PM12/22/21
to OxCal group
Yes - in theory - but Dates on either end of a sequence like the last one will not have anything to bracket them so will be undetermined. It would be fine for intermediate dates in a longer stratigraphy though.

Christopher
> To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/87tuf0ih9g.fsf%40tsdye.online.

Erik Marsh

unread,
Dec 23, 2021, 6:42:18 AM12/23/21
to OxCal
For ceramic phases a trapezium may be more appropriate than a uniform.
Yes, good idea.
 
If you don't apply the uniform span prior at all though you will overestimate the length of all the groups because of the statistical scatter.
I agree – in this case there's no need to turn this off.
 
Ideally you use the Boundaries in the sequence which defined the groupings.
Actually, I have boundaries in all the sequences, I just left them out of my example.
 
You don't wish to apply the uniform span more than once so if you wish to avoid this in the cross referenced sequences you can just use the "Date" function. You can do this either way round - so you might for example choose to set all of your Ceramic phases as separate groups as Erik suggests below (but with the original dates in). Then in the main model don't use any boundaries.

Would this work? Similar to Tom's suggestion, it's a Phase without boundaries. Since these dates all are within boundaries in the stratigraphic sequences, then they are constrained. And if I understand it right, this means they are are only constrained by the stratigraphic models and not by the ceramic phase – which is what I was looking for.

Wouldn't it be the same to remove these placeholder boundaries? Date("Start Ceramic Style 1");
and simplify to:

Phase("Ceramic Style 1")
{
First();
Date("=A");
Date("=E");
Date("=H");
Last();
};

Erik Marsh

unread,
Dec 23, 2021, 7:38:11 AM12/23/21
to OxCal
Follow-up question: To get a KDE of the ceramic style dates, should I use KDE_Model or KDE_Plot?

Christopher Ramsey

unread,
Dec 23, 2021, 9:19:47 AM12/23/21
to OxCal group
If you are cross-referencing them from another model certainly use KDE_Plot which does no modelling.

If just doing each style on its own without another model KDE_Model would be appropriate (though of course non-Bayesian).

Best wishes

Christopher
> --
> 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.
> To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/f022da0e-c56d-4c62-8dcf-f91a552c0060n%40googlegroups.com.

Erik Marsh

unread,
Dec 26, 2021, 8:24:16 AM12/26/21
to OxCal
So I haven't been able to get this to work. I've put the dates within boundaries in their stratigraphic sequences and then cross-referenced them in a KDE_Plot with no boundaries (see code below). When the model gets to the "First" query, it cannot resolve the order. I tried replacing KDE_Plot with Phase and KDE_Model, and grouping the stratigraphic and ceramic phase as separate Plots, but I get the same result. I also tried Tom's suggestion of nest the Phase within "dummy" boundaries, Date("Start Ceramic Style 1"), etc.

Maybe I need another approach? 
What I want is to group all the dates with the same ceramic style, but with _only_ the constraints in the stratigraphic sequences. It's essentially a query of the those dates that already have stratigraphic constraints without introducing new constraints.

Thanks for any suggestions!
Erik

Sequence("Stratigraphic 1")
{
Boundary("Start Stratigraphic 1");
R_Date("A");
R_Date("B");
R_Date("C");
Boundary("End Stratigraphic 1");
};
Sequence("Stratigraphic 2")
{
Boundary("Start Stratigraphic 2");
R_Date("E");
R_Date("F");
R_Date("G");
Boundary("End Stratigraphic 2");
};
Sequence("Stratigraphic 3")
{
Boundary("Start Stratigraphic 3");
R_Date("H");
R_Date("I");
R_Date("J");
Boundary("End Stratigraphic 3");
};
KDE_Plot("Ceramic Style 1")
{
First("First");
Last("Last");
Date("=A");
Date("=E");
Date("=H");
};

Reply all
Reply to author
Forward
0 new messages