Saving residuals after xtmixed

[Kirstin Vach]

Dear all,
I can't find out how to store the estimated correlation after xtmixed.

I run:

xtmixed y cons1 cons4 if (Phase==1 & (cons1==1 |cons4==1 )) , nocons || Pat_Nr: cons1 cons4 , nocons reml cov(unstructured) ///
|| Woche: ,nocons res(unstructured,t(cons1) )

Obtaining starting values by EM:

Performing gradient-based optimization:

Iteration 0: log restricted-likelihood = 58.969166
Iteration 1: log restricted-likelihood = 62.175238
Iteration 2: log restricted-likelihood = 62.332964
Iteration 3: log restricted-likelihood = 62.339482
Iteration 4: log restricted-likelihood = 62.339706
Iteration 5: log restricted-likelihood = 62.339707

Computing standard errors:

Mixed-effects REML regression Number of obs = 66

-------------------------------------------------------------
| No. of Observations per Group
Group Variable | Groups Minimum Average Maximum
----------------+--------------------------------------------
Pat_Nr | 11 6 6.0 6
Woche | 33 2 2.0 2
-------------------------------------------------------------

Wald chi2(2) = 31.62
Log restricted-likelihood = 62.339707 Prob > chi2 = 0.0000

------------------------------------------------------------------------------
y | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
cons1 | .0954103 .0193423 4.93 0.000 .0575001 .1333204
cons4 | .0414372 .0120296 3.44 0.001 .0178596 .0650148
------------------------------------------------------------------------------

------------------------------------------------------------------------------
Random-effects Parameters | Estimate Std. Err. [95% Conf. Interval]
-----------------------------+------------------------------------------------
Pat_Nr: Unstructured |
sd(cons1) | .0124668 .0832817 2.57e-08 6053.555
sd(cons4) | .0017214 .0020688 .0001632 .0181509
corr(cons1,cons4) | .999999 .0063395 -1 1
-----------------------------+------------------------------------------------
Woche: (empty) |
-----------------------------+------------------------------------------------
Residual: Unstructured |
sd(e0) | .0690406 .0086346 .0540317 .0882187
sd(e1) | .1089946 .0161692 .0814949 .1457739
corr(e0,e1) | .1526103 .1737057 -.1923375 .4639924
------------------------------------------------------------------------------
LR test vs. linear model: chi2(5) = 7.39 Prob > chi2 = 0.1935

Note: LR test is conservative and provided only for reference.

.
It's clear to me how to store corr(cons1,con4)=0.99999 (local corr=tanh(_b[atr1_1_1_2:_cons])).
But I can't find a solution to store corr(e0,e1), namely 0.1526. Where do I find this?
Regards
Kirstin