Not accepting assignment statements

[Jonathan Harris]

In a relatively simple autoregressive model I declare in the data block:

int N;

vector[N] x;

vector[N] y;

...

and in the parameters block:

real alpha;

real beta;

...

and in the model block:

vector[N] yhat;

alpha ~ cauchy(abar,asig);

beta ~ cauchy(betabar,betasig);

yhat=beta*x+alpha;

 and get error:

Illegal statement beginning with non-void expression parsed as

  yhat

Not a legal assignment, sampling, or function statement.

I also get a similar error if I write this in loop format. Am I not allowed to create a vector in the model block? Is there something else going on?

[Daniel Lee]

Currently, all variable declarations must come at the start of any block. If there are any statements between the declaration of yhat and the start of the block, "{", you'll see a message like the one you're getting. 

Daniel

P.S. It really helps to post the whole Stan program. I'm just guessing as to what's wrong. 

P.P.S. It also helps to have the whole error message. 

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

[Jonathan Harris]

I even tried the MA(Q) example on page 93 of the manual and get a similar error. I am using R 3.2.3 and rstan:

rstan (Version 2.9.0-3, packaged: 2016-02-11 15:54:41 UTC, GitRev: 05c3d0058b6a)

[Jonathan Harris]

Thanks Daniel: But this sample program from the reference manual also fails:

Full program stanalt.stan:

data {

int<lower=0> Q; // num previous noise terms

int<lower=3> T; // num observations

vector[T] y; // observation at time t

}

parameters {

real mu; // mean

real<lower=0> sigma; // error scale

vector[Q] theta; // error coeff, lag -t

}

transformed parameters {

vector[T] epsilon; // error term at time t

for (t in 1:T) {

epsilon[t] = y[t] - mu;

for (q in 1:min(t - 1, Q))

epsilon[t] = epsilon[t] - theta[q] * epsilon[t - q];

}

}

model {

vector[T] eta;

mu ~ cauchy(0, 2.5);

theta ~ cauchy(0, 2.5);

sigma ~ cauchy(0, 2.5);

for (t in 1:T) {

eta[t] = mu;

for (q in 1:min(t - 1, Q))

eta[t] = eta[t] + theta[q] * epsilon[t - q];

}

y ~ normal(eta, sigma);

}

Error message and session info:

> fit1=stan(file="C:/Users/jgharris/RProjects/stanalt.stan",data=standataalt,chains=2,iter=2000)             

SYNTAX ERROR, MESSAGE(S) FROM PARSER:

Illegal statement beginning with non-void expression parsed as

  epsilon[t]

Not a legal assignment, sampling, or function statement.  Note that

  * Assignment statements only allow variables (with optional indexes) on the left;

    if you see an outer function logical_lt (<) with negated (-) second argument,

    it indicates an assignment statement A <- B with illegal left

    side A parsed as expression (A < (-B)).

  * Sampling statements allow arbitrary value-denoting expressions on the left.

  * Functions used as statements must be declared to have void returns

ERROR at line 14

 12:    vector[T] epsilon; // error term at time t

 13:    for (t in 1:T) {

 14:    epsilon[t] = y[t] - mu;

        ^

 15:    for (q in 1:min(t - 1, Q))

PARSER EXPECTED: "}"

Error in stanc(file = file, model_code = model_code, model_name = model_name,  : 

  failed to parse Stan model 'stanalt' due to the above error.

In addition: Warning messages:

1: package ‘rstan’ was built under R version 3.2.5 

2: package ‘ggplot2’ was built under R version 3.2.5 

> sessionInfo()

R version 3.2.3 (2015-12-10)

Platform: x86_64-w64-mingw32/x64 (64-bit)

Running under: Windows 7 x64 (build 7601) Service Pack 1

locale:

[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252   

[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C                          

[5] LC_TIME=English_United States.1252    

attached base packages:

[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:

[1] reshape2_1.4.1 dplyr_0.4.3    rstan_2.9.0-3  ggplot2_2.1.0 

loaded via a namespace (and not attached):

 [1] Rcpp_0.12.5      assertthat_0.1   grid_3.2.3       plyr_1.8.4       R6_2.1.2         gtable_0.2.0    

 [7] DBI_0.4-1        magrittr_1.5     stats4_3.2.3     scales_0.4.0     stringi_1.1.1    tools_3.2.3     

[13] stringr_1.0.0    munsell_0.4.3    parallel_3.2.3   inline_0.3.14    colorspace_1.2-6 gridExtra_2.2.1 

> 

[Jonathan Harris]

My variable declarations are at the start of the block (my original program was encapsulated in a string, the sample from the manual, pasted into a file).

Also the full original program is:

stan1<- "

data {

int N;

vector[N] x;

vector[N] y;

real abar;

real asig;

real betabar;

real betasig;

real sigmabar;

real sigmasig;

real rhobar;

real rhosig;

}

parameters {

real alpha;

real beta;

real<lower=0> sigma;

real rho;

}

model

{

vector[N] resid;

vector[N] eps;

vector[N] yhat;

alpha ~ cauchy(abar,asig);

beta ~ cauchy(betabar,betasig);

sigma ~ lognormal(sigmabar,sigmasig);

rho ~ cauchy(rhobar,rhosig);

yhat=beta*x+alpha;

for(n in 1:N){

    yhat[n]=beta*x[n]+alpha;

    resid[n]=y[n]-yhat[n];

}

eps[1]=resid[1];

for(n in 2:N) {

    eps[n]=resid[n]-rho*resid[n-1];

}

eps ~ normal (0,sigma);

}

"            

 

On Saturday, June 25, 2016 at 8:04:38 AM UTC-4, Daniel Lee wrote:

[Jonathan Harris]

And this arrangement of the model block also fails:

model

{

vector[N] yhat;

yhat=beta*x+alpha;

vector[N] resid;

for(n in 1:N){

    resid[n]=y[n]-yhat[n];

}

vector[N] eps;

eps[1]=resid[1];

for(n in 2:N) {

    eps[n]=resid[n]-rho*resid[n-1];

}

alpha ~ cauchy(abar,asig);

beta ~ cauchy(betabar,betasig);

sigma ~ lognormal(sigmabar,sigmasig);

rho ~ cauchy(rhobar,rhosig);

eps ~ normal (0,sigma);

[Eric Innocents EBOULET]

Hi,

You can try replacing in your model "=" with "<-"

I hope that it will run.

Bests

Eric

[Jonathan Harris]

THat worked. I am puzzled by why the examples in the documentation show "="

[Daniel Lee]

Ah. Yes. The "=" syntax is available starting with v2.10.0. You're running rstan 2.9.0. rstan 2.10 isn't available yet. 

Daniel

[Bob Carpenter]

We should probably not add the new manual until
all of the interfaces are working with it. We'll
be backwards compatible with the old one.

- Bob

[Daniel Lee]

Does the rstan distribution include the reference guide?

[Ben Goodrich]

No

[Daniel Lee]

Jonathan, sorry about the confusion.... I've updated the documentation page so there's a link to v2.9.0 of the manual.

Ben / Bob: if you think we should remove the v2.10.0 version link, please do so or let me know and I'll take care of it.

Daniel

On Sat, Jun 25, 2016 at 5:51 PM, Ben Goodrich <goodri...@gmail.com> wrote:

No

[Ben Goodrich]

On Monday, June 27, 2016 at 9:49:53 AM UTC-4, Daniel Lee wrote:

Jonathan, sorry about the confusion.... I've updated the documentation page so there's a link to v2.9.0 of the manual.

Ben / Bob: if you think we should remove the v2.10.0 version link, please do so or let me know and I'll take care of it.

We can leave it; rstan 2.10.1 is fine now.

Ben