pmap | Process Map Visualization of event analysis in R | Data Visualization library

The problem is that in your lambda function, you are not using the argument names for rnbinom. The default order for the arguments if you don't name them is n, size, prob, mu, so you are passing 2000 to n, 25 to size and 0.9 to prob. Just name the arguments explicitly as you did in your first example, and your code will work.

The problem is that the function you are trying to apply ('%*%') is a function which only takes two arguments and not three.

For your problem I would recommend Reduce (or purrr::reduce): in this case Reduce starts with two matrices and multiplies them. Then it multiplies the result with the third matrix. Theoretically you could go on like this.

I solved it with a simple refactoring of your code, automatizing thought a loop. The more data you add, the better your time spent. I believe this fct[[paste0(names(df[i]),"_v")]] can be refactored in an small function to look even better

One option may look like so:

1. I have put the parameters for your curves in a data.frame
2. Making use of a function factory and pmap to loop over the params df to create a list of your icc functions.

The rest is pretty straighforward.

1. Loop over the list of functions to get the derivatives.

2. Use map to add the stat_function layers.

The issue is that df_pt is a data.frame and it needs to be used as input in each of the looped element. So, wrap it with list so that it gets recycled as a single unit. When we loop over the data.frame, the column is a unit and this triggers the error Erro: Element 3 of .l must have length 1 or 30, not 3 as the number of columns is 3.

Thanks to the help from issue of gtExtras:

One option to get rid of the message would be to set the names inside your function like so:

Create a list of calls and splice them in:

No, there is no real analog to CUDA threadid in JAX. Details about GPU thread assignment are handled at a lower level by the XLA compiler, and I don't know of any straightforward API to plumb this information back to JAX's Python runtime.

One case where JAX does offer higher-level handling of device assignment is when using pmap; in this case you can explicitly pass a set of device IDs to the pmapped function if you want logic that depends on the device on which the mapped code is being executed. For example, I ran the following on an 8-device system: