# omega matrix

**URL:** <https://forum.certara.com/t/omega-matrix/3618>\
**Category:** Modelling and Simulation\
**Created:** [June 22, 2022, 8:38am UTC](https://forum.certara.com/t/omega-matrix/3618 "2022-06-22T08:38:28Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![crystal](https://avatars.discourse-cdn.com/v4/letter/c/2bfe46/32.png) [@crystal](https://forum.certara.com/u/crystal)\
**Post date:** [June 22, 2022, 8:38am UTC](https://forum.certara.com/t/omega-matrix/3618/1 "2022-06-22T08:38:28Z")

</div>

Dear all,  
I am confused about how to determine whether to use diagonal omega matrix or a full block omega matrix when attempting to converge a population PK model. which plot or Table in the output panel can I use to  
determine?  
I cannot find the answer in the User’s Guide. Does anyone know it?  
Thank you very much.

---

<div class="post-metadata">

**Author:** ![Samer\_Mouksassi](https://avatars.discourse-cdn.com/v4/letter/s/ce7236/32.png) [@Samer\_Mouksassi](https://forum.certara.com/u/Samer_Mouksassi)\
**Post date:** [June 22, 2022, 8:50am UTC](https://forum.certara.com/t/omega-matrix/3618/2 "2022-06-22T08:50:13Z")

</div>

if you data support fitting full omega go for it also use newer engines like QRPEM

you can then look at the omega and its standard error and also look at the computed correlation and shrinkages

then you can simplify your omega matrix if you see some off diagonals to be very small and not needed.

you can do a likelihood ratio test for off diagonal elements of omega

for variance components have a look at :  
[https://www.researchgate.net/publication/322048811\_Likelihood\_ratio\_test\_for\_variance\_components\_in\_nonlinear\_mixed\_effects\_models](https://www.researchgate.net/publication/322048811_Likelihood_ratio_test_for_variance_components_in_nonlinear_mixed_effects_models)

---

<div class="post-metadata">

**Author:** ![crystal](https://avatars.discourse-cdn.com/v4/letter/c/2bfe46/32.png) [@crystal](https://forum.certara.com/u/crystal)\
**Post date:** [June 23, 2022, 12:57am UTC](https://forum.certara.com/t/omega-matrix/3618/3 "2022-06-23T00:57:37Z")

</div>

> [@smouksassi1](#):
>
> if you data support fitting full omega go for it also use newer engines like QRPEM
> 
> you can then look at the omega and its standard error and also look at the computed correlation and shrinkages
> 
> then you can simplify your omega matrix if you see some off diagonals to be very small and not needed.
> 
> you can do a likelihood ratio test for off diagonal elements of omega
> 
> for variance components have a look at :  
> [https://www.researchgate.net/publication/322048811\_Likelihood\_ratio\_test\_for\_variance\_components\_in\_nonlinear\_mixed\_effects\_models](https://www.researchgate.net/publication/322048811_Likelihood_ratio_test_for_variance_components_in_nonlinear_mixed_effects_models)

Thanks a lot for your help, smouksassi1,I really appreciate it.

I still have a question regarding how to determine"some off diagonals to be very small and not needed",is there a criteria?

Thank you~~