umx_make_TwinData: support DD (non-additive component)

[tbates]

Allow this:

tmp = umx_make_TwinData(AA= .4, DD= .4, C=0)
m1 = umxACE(selDVs="var", dzCr=.25, data = tmp)

	a1	d1	e1
var	0.4	0.4	0.2

Originally posted by @tbates in #76 (comment)

[tbates]

Done: Example simulating both D & C > 0

tmp = umx_make_TwinData(nMZ = 1000, AA = .4, DD = .2, CC = .2)

	AA	CC	DD	EE
var	0.4	0.2	0.2	0.2
path	.63	.45	.45	.45

Then running ACE and ADE on the resulting data: makes a nice tutorial

m1 = umxACE(selDVs="var", data = tmp)

ACE

	a1	c1	e1
var	0.86	0.24	0.45

m1 = umxACE(selDVs="var", data = tmp, dzCr=.25)

ADE

	a1	d1	e1
var	0.9	.	0.44

[mcneale]

Interesting. Without additional relative types there’s an identification problem. Makes me wonder: might they be readily output too, given A/C/D covariance coefficients? The wrench in the works, as always is assortative mating.

[tbates]

Could add option to deviate rA from .5 pretty easily, and allow simulating data from a trait with assortative mating.

You mean output the other relatives? That's a nice idea. What would be most useful? Probably will be a back burner project, as we don't yet have umx functions for extended twin design, but a teaching toy would be easy to mock up.

No identification problem on the data generation side, of course :-) I just:

ACDE   , hA+C+qD
hA+C+qD, ACDE