I have a model with an ordinal outcome (sratio family) with two crossed group effects plus a monotonic predictor and a factor covariate. The number of levels of the group effects are 112 and 491. The model has good Rhat, ESS and no warnings. The model object size is about 72 MB. I run the model with a memory allocation of 32 GB without difficulty

loo(object) yields a R crash. This is repeatable. I assume that this is a memory problem.

```
bprior <- prior(student_t(3, 0, 2.5), class = 'b')
fit0.brmsfit <- brm(
formula = as.ordered(Interventions) ~
mo(Risk) + Race + (1 | CensusTract) + (1 | AnonymousProvider),
data = CuratedPONV.df %>%
filter(ProviderCasesCount >= 50 & CensusTractCasesCount >= 25 & Anesthesia != 'MAC'),
family = sratio(),
prior = bprior,
iter = 2000,
cores = 8,
chains = 8
)
```

The data is PHI and can’t be shared.

I tried running the model with fewer iterations and received ESS warnings.

I can obtain a larger memory allocation with a batch job. Can memory requirements be calculated? How much will it take?

Is the function loo_subsample.brmsfit relevant for producing a loo object?

Nathan

Please also provide the following information in addition to your question:

- Operating System: Centos07
- brms Version: 2.17.0
- R: 4.1.1

Model output:

Family: sratio

Links: mu = logit; disc = identity

Formula: as.ordered(Interventions) ~ mo(Risk) + Race + (1 | CensusTract) + (1 | AnonymousProvider)

Data: CuratedPONV.df %>% filter(ProviderCasesCount >= 50 (Number of observations: 33073)

Draws: 8 chains, each with iter = 2000; warmup = 1000; thin = 1;

total post-warmup draws = 8000

Group-Level Effects:

~AnonymousProvider (Number of levels: 112)

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS

sd(Intercept) 0.63 0.04 0.56 0.73 1.00 1079 2295

~CensusTract (Number of levels: 491)

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS

sd(Intercept) 0.22 0.01 0.19 0.24 1.00 3540 5614

Population-Level Effects:

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS

Intercept[1] -4.24 0.16 -4.54 -3.93 1.00 1515 3296

Intercept[2] -2.97 0.15 -3.27 -2.66 1.00 1496 3170

Intercept[3] -1.26 0.15 -1.56 -0.95 1.00 1468 3398

Intercept[4] 0.57 0.15 0.26 0.87 1.00 1501 3070

Intercept[5] 0.77 0.16 0.46 1.09 1.00 1658 3466

Intercept[6] 3.63 0.40 2.91 4.48 1.00 6304 4891

RaceUnknown -0.13 0.04 -0.20 -0.06 1.00 11642 6694

RaceWhite 0.22 0.03 0.16 0.27 1.00 10865 6723

moRisk -0.52 0.05 -0.63 -0.43 1.00 3378 3914

Simplex Parameters:

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS

moRisk1[1] 0.24 0.04 0.17 0.31 1.00 3061 4114

moRisk1[2] 0.15 0.02 0.11 0.19 1.00 4250 4384

moRisk1[3] 0.02 0.01 0.00 0.03 1.00 7957 3760

moRisk1[4] 0.03 0.01 0.02 0.04 1.00 7316 6098

moRisk1[5] 0.06 0.01 0.04 0.09 1.00 5498 5180

moRisk1[6] 0.32 0.04 0.25 0.39 1.00 3687 4151

moRisk1[7] 0.18 0.08 0.03 0.32 1.00 3281 2923

Family Specific Parameters:

Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS

disc 1.00 0.00 1.00 1.00 NA NA NA

Looking forward to your topic!