Is there a function in Stan equivalent to R’s which() command, that basically gives me an index of elements of a vector meeting a certain condition. Eg: > a <- c(0, 4, 5, 7, 0, 2, 45, 1, 0) > which(a==0) [1] 1 5 9 So I tried to write a function in Stan to match zeros: functions{ vector which…

[image] jroon: Is there a function in Stan equivalent to R’s which() command, that basically gives me an index of elements of a vector meeting a certain condition. Not yet. And you can’t write a generic one because we don’t have higher order functions. In Stan, if you have a particular test…

I found it easier to do stuff like this in R and feed the results to RStan as data. Then you’d have an number of observed Y int N_Yobs and a position vector (of where thise observed Y’s are) vector[N_Yobs] Ypos (the result of R’s which). Do that for the “missing” values as well and then, as a parame…

Eh admittedly I didnt’ think of that, but a complication is I’m doing this rowise in a loop. If I go with the option of precomputing it in R I’ll be passing into Stan N vectors of varying length for Ypos and Ymis. Sounds very messy!

An inefficient way would be to write two functions. (1) to calculate the number of matches (2) to find the matches where (1) would allow you to specify the length of the matches vector. You do have to loop over de target vector twice (or only until your last match). Hence the inefficiency. It’s p…

So I may have to rethink it a bit. The target vector is data defined as an array: vector[NvarsY] y [N]; I had intended to which0() inside a for(n in 1:N) loop, because outside of the loop I don’t really care about the index. However I can see that precomputing it in R or a transformed data{} bloc…

How I would do it: In R: N <- 40 K <- 2 mus <- c(-1.5, 0.5) sigmas <- c(1.7, 2.4) rho <- 0.4 Omega <- diag(1, nrow = 2, ncol = 2) Omega[2,1] <- rho Omega[1,2] <- rho L_Omega <- chol(Omega) Z <- matrix(rnorm(N*K), nrow = N, ncol = K) log_Y <- t(mus + diag(sigmas) %*% L_Omega %*% t(Z)) apply(log_…

Thanks @Max_Mantei , I’ve been working on something similar, but you just gave me an idea to improve it - many thanks!

Here check out this thread I started way back: [image] Stan equivalent of R's %in% function RStan Hi all - I had a need to create a version of R’s %in% function for Stan, so I decided to post the code here in case anyone finds it useful. Essentially checks if an integer …

Thanks @saudiwin , I actually saw that before making this thread. It won’t work for what I wanted. I’ve come to the conclusion a Stan version of which() is impossible because the length of the which() output varies depending on the data, and that would require storing the result in a ragged array and…

Thanks Bob. I ended up doing the indexing in R and passing it in as data, but good to know this solution too!

Just came across this when answering a different post. Here are slightly modified versions of Bob’s suggestions above that I just tested when responding to How to find the location of a value in a vector . I changed it to work with a vector input (instead of an int array) and fixed a typo where match…

Stan equivalent to R's which()

Algorithms

saudiwin March 14, 2019, 3:28pm 9

Here check out this thread I started way back:

Topic		Replies	Views
Equivalent of R's which? General	5	210	August 8, 2024
How to find the location of a value in a vector General	3	1661	December 13, 2020
Find or Which command in Stan? General	8	2624	July 20, 2017
Stan equivalent of R's %in% function RStan	9	2405	April 17, 2018
Dealing with data subsetting in Stan Developers	18	3886	August 20, 2019

Stan equivalent to R's which()

Related topics