A bug when running runWarpLDAModels() with exactly one topic

[Van1yu3]

Hi!

I would like to report a bug when running runWarpLDAModels() with exactly one topic.

The error message is

Error in alpha/t : non-numeric argument to binary operator

After checking the R code, I finally located the error, which is in line 644 of RunModels.R. The t in this line should be topic, I think.

Best
Van1yu3

[Van1yu3]

Also, the t should be topic in line 638.

[cbravo93]

Hi @Van1yu3 !

Thanks a lot for noticing, it is corrected now! The bug was affecting 644 and 647.

Cheers,

C

[Van1yu3]

Hi @cbravo93 !

Thank you for your reply. Sadly I still got some problems when the topic is exactly one.

Indeed, there are two issues when I run runWarpLDAModels() with exacly one topic:

1. I got the following error message:

Error: $ operator is invalid for atomic vectors
I guess this error is from line 651 , where the y is not what topic_sums is expected to be taken from.

2. The random seed seemed not to function since I found I got 2 different results with the same random seed. This happens probably because you forgot to set.seed() in the runWarpLDAModels() somewhere. I only see set.seed() function in line 160 in the runCGSModels().

[cbravo93]

Hi @Van1yu3 !

Sorry, now it is solved (and tested). Normally we run more than a model.

Also, for the model selection use: cisTopicObject <- selectModel(cisTopicObject, type='maximum')

Hope it works well, thanks for reporting back!

C

[Van1yu3]

The problem is now solved. Thank you for correcting it!