Using the wiener_lpdf() function in the cmdstanr package to implement a diffusion model, the construction of a Markov chain was unsuccessful

[luckyjlb]

I encountered the following problem when running the code related to the seven parameter diffusion model using the latest version 2.35 of the cmdstanr package. After repeated modifications, it still persists. I really can't find the reason. I hope to receive everyone's help.Thank you.
The code I am running mainly comes from https://valentinpratz.de/posts/2022-12-29-stan-wiener_full-level-1/ And I made corresponding adjustments. The current code does not report any errors, but building a Markov chain was unsuccessful.
##example
I used a new version of the function wiener_lpdf (rt [i] | a, t0, w, v [cnd [i]], sv, sw, st0)
I mainly learned from the following link.
#2822
example.zip

#code:
library(cmdstanr)
load("L1dat.RData")
data <- Level1dat
summary(data)
et_cmdstan_path(path=NULL)
chains = 4

initial values

init.stan = function(){
L = list()
for (i in 1:chains) {
L[[i]] = list(
sv = 1 + runif(1, -0.1, 0.1),
sw = 0.1 + runif(1, -0.05, 0.05),
st0 = 0.2 + runif(1, -0.05, 0.05)
)
}
return(L)
}

Compiling model

#mc <- cmdstan_model("Level1.stan")
#mc<-cmdstan_model(file)
mc <- cmdstan_model("D:\seven\level1\level1.stan")
stan.data = list(
N = nrow(data),
Ncnds = 3,
rt = data$RT,
stim = as.numeric(data$Stim),
resp = as.numeric(data$Resp),
cnd = as.numeric(data$Cond),
a = 1,
t0 = 0.3,
w = 0.45,
v = c(3.5, 2.5, 1.5)
)
outdir = "fits"
dir.create(outdir, recursive = T, showWarnings = FALSE)
m <- mc$sample(data=stan.data,
init = init.stan(),
refresh = 25,
iter_sampling = 250,
iter_warmup = 250,
chains = chains,
parallel_chains = 4,
output_dir = outdir,
output_basename = "fit",
save_warmup = TRUE
)
##issues
unning MCMC with 4 parallel chains...

Chain 2 Assertion failed: index >= 0 && index < size(), file stan/lib/stan_math/lib/eigen_3.4.0/Eigen/src/Core/DenseCoeffsBase.h, line 410Warning: Chain 2 finished unexpectedly!

Chain 1 Assertion failed: index >= 0 && index < size(), file stan/lib/stan_math/lib/eigen_3.4.0/Eigen/src/Core/DenseCoeffsBase.h, line 410Warning: Chain 1 finished unexpectedly!

Chain 3 Informational Message: The current Metropolis proposal is about to be rejected because of the following issue:

Chain 3 Exception: wiener_lpdf: Inter-trial variability in A-priori bias = 0.999729, but must be smaller than 2*(A-priori bias) = 0.9 (in 'C:/Users/13201/AppData/Local/Temp/RtmpYrSjfs/model-6fc8368c44a1.stan', line 29, column 8 to column 72)

Chain 3 If this warning occurs sporadically, such as for highly constrained variable types like covariance matrices, then the sampler is fine,

Chain 3 but if this warning occurs often then your model may be either severely ill-conditioned or misspecified.

Chain 3 Chain 4 Informational Message: The current Metropolis proposal is about to be rejected because of the following issue:

Chain 4 Exception: wiener_lpdf: Inter-trial variability in A-priori bias = 0.999978, but must be smaller than 2*(A-priori bias) = 0.9 (in 'C:/Users/13201/AppData/Local/Temp/RtmpYrSjfs/model-6fc8368c44a1.stan', line 29, column 8 to column 72)

Chain 4 If this warning occurs sporadically, such as for highly constrained variable types like covariance matrices, then the sampler is fine,

Chain 4 but if this warning occurs often then your model may be either severely ill-conditioned or misspecified.

##Expected output
I hope to generate Markov chains normally.Hope to receive help.

####Current version:
cmdstanr packages 2.35

[SteveBronder]

This would probably be better posted on the Stan discourse forums

https://discourse.mc-stan.org/