Error in dim(x) <- length(x) : attempt to set an attribute on NULL

[njtierney]

Got a little mysterious bug here, it only comes up sometimes, but just logging it here so I can refer back to this.

>   lkj_checked <- check_samples(
+     x = x,
+     iid_function = iid,
+     sampler = hmc(),
+     one_by_one = TRUE
+   )
running 4 chains simultaneously on up to 8 CPU cores

    warmup                                      0/1000 | eta:  ?s          Error in dim(x) <- length(x) : attempt to set an attribute on NULL
Called from: as.array.default(batch_results$all_states)
Browse[1]> traceback()
No traceback available 
Browse[1]> n
> traceback()
14: as.array.default(batch_results$all_states)
13: as.array(batch_results$all_states) at inference_class.R#747
12: self$run_burst(n_samples = burst_lengths[burst]) at inference_class.R#401
11: sampler$run_chain(n_samples = n_samples, thin = thin, warmup = warmup, 
        verbose = verbose, pb_update = pb_update, one_by_one = one_by_one, 
     ... at inference.R#381
10: dispatch(sampler$run_chain(n_samples = n_samples, thin = thin, 
        warmup = warmup, verbose = verbose, pb_update = pb_update, 
     ... at inference.R#388
9: run_samplers(samplers = samplers, n_samples = n_samples, thin = thin, 
       warmup = warmup, verbose = verbose, pb_update = pb_update, 
    ... at inference.R#265
8: force(expr)
7: tryCatchList(expr, classes, parentenv, handlers)
6: tryCatch(force(expr), finally = {
       data$`__exit__`(NULL, NULL, NULL)
    ...
5: with.python.builtin.object(tf$device(compute_options), {
       message_if_using_gpu(compute_options)
    ...
4: with(tf$device(compute_options), {
       message_if_using_gpu(compute_options)
    ... at inference.R#204
3: mcmc(model, sampler = sampler, verbose = verbose, one_by_one = one_by_one) at helpers.R#739
2: get_enough_draws(model = m, sampler = sampler, n_effective = n_effective, 
       verbose = TRUE, one_by_one = one_by_one, time_limit = time_limit) at helpers.R#854
1: check_samples(x = x, iid_function = iid, sampler = hmc(), one_by_one = TRUE)