Exporting samples.csv in RStan

[Peter Li]

Is there a way to export the samples.csv from RStan? Also, is there a place where all the functions in R are listed?


Thanks,
Peter

[Ben Goodrich]

Not sure what you mean by export. The doc for the sample_file in the stan() function says

A character string of file name for specifying where to write samples for all parameters and other saved quantities. If not provided, files are not created. When the folder specified is not writable, tempdir() is used. When there are multiple chains, an underscore and chain number are appended to the file name

There are thousands of functions in R even before you get into the optional R packages. If you just mean all the functions in rstan, you can do

> ls("package:rstan")
 [1] "constrain_pars"      "extract"             "get_adaptation_info" "get_cppcode"        
 [5] "get_cppo"            "get_cppo_mode"       "get_cxxflags"        "get_inits"          
 [9] "get_logposterior"    "get_num_upars"       "get_posterior_mean"  "get_sampler_params"
[13] "get_seed"            "get_seeds"           "get_stancode"        "get_stanmodel"      
[17] "grad_log_prob"       "log_prob"            "makeconf_path"       "monitor"            
[21] "optimizing"          "plot"                "read_rdump"          "read_stan_csv"      
[25] "rstan_options"       "sampling"            "set_cppo"            "sflist2stanfit"    
[29] "show"                "stan"                "stanc"               "stan_demo"          
[33] "stan_model"          "stan_rdump"          "stan_version"        "summary"            
[37] "traceplot"           "unconstrain_pars"  

Ben

[Peter Li]

Ah, got it. Thanks. Also, do you know how I can specify inits in RStan? I tried setting inits equal to a list of all of the initial values that I wanted to use:

  radon_group_chr.sf1 <- sampling(radon_group_chr.sm, dataList.3, chains=1,init=dataList.4,sample_file=paste('radon_group_cp_',count,'.csv',sep=''))

 but I get the error:

Error in config_argss(chains = chains, iter = iter, warmup = warmup, thin = thin,  : 

  initial value list mismatchs number of chains

error specification of arguments; sampling not done

[Bob Carpenter]

You need to pass in the inits as a function. So pass in
init_fun, where you define:

init_fun = function() {
return(list(alpha=alpha_init,
beta=beta_init));
}

- Bob

On 10/5/13 8:40 PM, Peter Li wrote:
> Ah, got it. Thanks. Also, do you know how I can specify inits in RStan? I tried setting inits equal to a list of all of
> the initial values that I wanted to use:
>

> radon_group_chr.sf1 <- sampling(radon_group_chr.sm <http://radon_group_chr.sm>, dataList.3,

> --
> You received this message because you are subscribed to the Google Groups "stan development mailing list" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to stan-dev+u...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.

[Jiqiang Guo]

On Sat, Oct 5, 2013 at 8:40 PM, Peter Li <li.pe...@gmail.com> wrote:

Also, do you know how I can specify inits in RStan? I tried setting inits equal to a list of all of the initial values that I wanted to use:

  radon_group_chr.sf1 <- sampling(radon_group_chr.sm, dataList.3, chains=1,init=dataList.4,sample_file=paste('radon_group_cp_',count,'.csv',sep=''))

 but I get the error:

Error in config_argss(chains = chains, iter = iter, warmup = warmup, thin = thin,  : 

  initial value list mismatchs number of chains

error specification of arguments; sampling not done

It needs to be a list, each element of which needs to be a list of all the initial values.  Suppose you have parameter alpha in your model, then the inits argument can be fed say for 4 chains as follows:

inits = list(c1=list(alpha=1), c2=list(alpha=2), c3=list(alpha=3), c4=list(alpha=4)).

So I think in your example, you should do init = list(dataList.4).

--

Jiqiang 

[Peter Li]

Ah alright. Thanks a lot Bob and Jiqiang. 

--