Likelihoods with shared components

[smar-git]

Dear David and Alexandre

I have found a way to implement the mode where the data are not observations on a point ( as in your example) but averages over lines using the new A.local argument that Havard has added to the f() function.
We still do not manage to fit a joint model where we have two likelihoods with some shared components. This is the problem where you suggested to use the “group” option. But I still do not understand how to implement this. You saied that you were going to write some more documentation and I wanted to ask if you manage to do that.
Thanks a lot
Sara

[AlexandreBSimas]

Hi Sara,

Yes! Dont worry, I did not forget, I will do it. I wanted to have this done by now, however my kids are on vacation (requiring a lot of attention), and to make things worse my whole family was sick these past weeks, so did not manage to work much.

In any case, I will update here when I implement it, will try to do it quickly. Also, very good news that you managed to implement it with this A.local argument! I was not aware of that argument, will take a look. I guess you can then probably implement it in inlabru. If you have any problems, I can try to help, you can message here. It is really super convenient to have these implementations in inlabru.

Best,
Alexandre

[smar-git]

Hi Sorry, didn’t want to stress 😊 I’ll wait for your news then From: Alexandre B. Simas ***@***.***> Date: Thursday, 27 June 2024 at 17:54 To: davidbolin/MetricGraph ***@***.***> Cc: Sara Martino ***@***.***>, Author ***@***.***> Subject: Re: [davidbolin/MetricGraph] Likelihoods with shared components (Discussion #8) Hi Sara, Yes! Dont worry, I did not forget, I will do it. I wanted to have this done by now, however my kids are on vacation (requiring a lot of attention), and to make things worse my whole family was sick these past weeks, so did not manage to work much. In any case, I will update here when I implement it, will try to do it quickly. Also, very good news that you managed to implement it with this A.local argument! I was not aware of that argument, will take a look. I guess you can then probably implement it in inlabru. If you have any problems, I can try to help, you can message here. It is really super convenient to have these implementations in inlabru. Best, Alexandre — Reply to this email directly, view it on GitHub<#8 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ANIDL5IDPWIXBYTB36YFX2DZJQYT3AVCNFSM6AAAAABKAGHYLGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TQOJWGEYTI>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[AlexandreBSimas]

Ahh, nono. Dont worry, you did not stress at all!
Sorry for the delay. I also expected to have this finished a while ago.

[AlexandreBSimas]

Hi Sara,

I have finished the vignette with the multiple likelihoods, with the example being exactly that model you mentioned. I used the group variable to handle the multiple likelihoods as I have mentioned to you before. I have modified the graph_data_spde() function to make it easier to handle these cases.

I have included examples with both INLA and inlabru.

The vignette is https://davidbolin.github.io/MetricGraph/articles/multi_likelihood.html

One thing is that since we are using the group variable to also handle the likelihoods, one might be confused on how to use replicates in such case, so I have also added examples in INLA and inlabru with multiple likelihoods and replicates.

Hope this helps you and solves your problem.

Best,
Alexandre

[AlexandreBSimas]

Hi Karina,

Thank you for your message. It really was supposed to work in general, not only when having observations on the same locations.

I have pushed a bugfix.

Please, could you try again and confirm whether it fixed for you?

Also, for you to get good estimates on the measurement error in the modified vignette you sent (since you were dividing the number of observations by 2), you probably will need to increase obs_per_edge to 150 or 200.

Best,
Alexandre

[karinalilleborge]

Hi Alexandre,
I did the remotes::install_github()- which did update a few things, but I still get the same error when I run the code from the start. I am currently trying the last example where I have removed only one observation in the df_graph2:

df_graph <- rbind(df_graph1, df_graph2[-1,])        

Otherwise, my code is the same as the vignette.
Thank you for the very quick response!
Bests,
Karina

[AlexandreBSimas]

Weird, here it is working just fine.

Did you close R, and reopened it (to load the new version of metricgraph)?

[karinalilleborge]

Sorry, I just cleared the workspace, but didn't load the new package, my mistake.
Thank you for fixing it so quickly! It seems to work now. 😊
Bests,
Karina

[AlexandreBSimas]

Great that it is working! My pleasure!

If you run into any other issues, just let me know.

[smar-git]

Thanks a lot! I’ll have a look at the vignette sara From: Alexandre B. Simas ***@***.***> Date: Sunday, 7 July 2024 at 03:09 To: davidbolin/MetricGraph ***@***.***> Cc: Sara Martino ***@***.***>, Author ***@***.***> Subject: Re: [davidbolin/MetricGraph] Likelihoods with shared components (Discussion #8) Ahh, just one thing. You will need to install the most recent version to be able to reproduce the vignette. That is, you need to run: remotes::install_github("davidbolin/metricgraph", ref = "devel") — Reply to this email directly, view it on GitHub<#8 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ANIDL5JKV4FE45P7Y5KRKFLZLCILRAVCNFSM6AAAAABKAGHYLGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TSNZXGA3TQ>. You are receiving this because you authored the thread.Message ID: ***@***.***>