I've been using Stan for a bit through Pystan and is really nice. Just recently however I tried to reduce the size of the output (particularly, the generated_quantities bit) using the same syntax than in STAN, by declaring save_warmup=False. However, when digging a bit into PyStan code I noticed that this is currently hard coded and not able to edit using the control argument (the dictionary that handles extra parameters in Pystan). I'm tempted to modify the source code directly, but just wondering if is something PyStan devs are currently considering.
Cheers and thanks for an amazing work! (both PyStan and Stan!)
Related
I am having a problem with my read_csv. I have used this function with no problem but the path/directory I am using is a little different than normal and I can't figure it out by myself.
This is the code I have been using:
X2022_03_08_habit_and_OCD_clinical <- read_csv("Box/OCD: Habit or Learning?/experiment/data/raw/survey-data/2022-03-08_habit-and-OCD_clinical.csv")
I have tired tweaking this by not using the first two arguments with no luck. Has anyone used Box in a path before (Box is in my finder, like desktop would be). I also tried updating r for that first error code but maybe it didn't take, I am not sure how to update again.
Here is the error code I have been receiving:Error Message
I would appreciate any help and I apologize if there is a simple answer I have been missing!
tl;dr on the solution: this error apparently gets flagged on Big Sur using XCode CL tools when you use CharacterVector.push_back() inappropriately. I commented out an offending line, and it promptly stopped throwing the error!
I'm currently working on using Rcpp to write a C++ function for R using Rcpp::sourceCpp("FilePathName"), and it's returning a weird error.
Library/FrameWorks/R.framework/Versions/4.1/Resources/library/Rcpp/include/Rcpp/Vector.h
Line 5: no matching constructor for initialization of 'std::string' (aka 'basic_string')
I've tried running the gamut on updating XCode, gfortran, R, & Rcpp, clearing .R/Environ & .R/MakeVars, and manually editing .R/MakeVars to avoid using clang, but I can't seem to get past this one weird error. I think it has something to do with the update to Big Sur, seeing as that seems to cause some weirdness from time to time.
Any insight as to how best to proceed? For now I'm just trawling through various articles and documentation pieces to find another instance of this error, but so far I've had no luck.
Thanks
I have got a memory image, that i can't find the source for and I want to get the code out of it again. What do i have to do to achieve that? I can obviously load the image, but then i'd need to guess the function names.
You may get "interesting" symbols with (apropos ""), and the function names with WITH-PACKAGE-ITERATOR and FBOUNDP. But the source code is (probably) lost: try DISASSEMBLE on functions and see the information which is there.
In addition to DISASSEMBLE, you might try EXT:UNCOMPILE. Note, however, that it will only work on functions compiled in an interactive session (i.e., from REPL), not on those loaded from a compiled .fas file.
So, the suggested procedure is:
LIST-ALL-PACKAGES - figure out which packages are interesting.
DO-EXTERNAL-SYMBOLS - figure out which symbols in the interesting packages are interesting.
DISASSEMBLE or EXT:UNCOMPILE on those interesting symbols.
However, the easiest way is to contact your vendor. Remember, CLISP is distributed under the GNU GPL.
I am attempting to automate the insertion of JPEG images into Powerpoint. I have a macro done for that already, except using R would be infinitely better for my purposes.
The package R2PPT should do this, I understand. However, I cannot use it. For example, when I try to use PPT.Open, I understand I can do it two different ways by calling method = "rcom" or method = "RDCOMClient". Using the latter, R will always crash, sending an error report to windows. Using the former, it tells me I need to install statconnDCOM , before giving the error:
Error in PPT.Open(x) : attempt to apply non-function.
I cannot install statconnDCOM freely, as I wouldn't call this work non-commercial. So if there isn't a way to get around this issue, are there at least some free alternatives to R2PPT so that I can save several hours of manual work with a simple R code? If there is a way for me to use R2PPT, that would be ideal.
Thanks!
Edit:
I'm using R version 2.15 and downloaded the most recent version of R2PPT. Powerpoint is 2007.
Do you have administrative privileges on this machine?
There is an issue with package RDCOMClient. It needs permissions to write file rdcom.err in the root of drive C:. If you don't have privileges to write to c:, there is a rather cumbersome workaround:
Close R
Create "c:\temp" folder if it doesn't exist.
Locate on your hard drive file rdcomclient.dll. It usually placed in \R\library\RDCOMClient\libs\i386\ and in \R\library\RDCOMClient\libs\x64\ (you need to patch file which corresponds your Windows version - 32 bit or 64 bit). It's recommended to make backup copy of this files before patching.
Open rdcomclient.dll in text editor (Notepad++, for example -http://notepad-plus-plus.org/)
Find in file string c:\rdcom.err - it occurs only once.
Go into overwrite mode (usually by pressing "Ins" key). It is very important that new path will have the same number of characters as original one. Type C:\temp\e.rr instead of c:\rdcom.err
Save the file.
Now all should work fine.
Arguably not an answer, but have you looked at using Sweave/knitr to render your presentations in LaTeX using something like Beamer? (As discussed on slide 17 here.)
Wouldn't help any with getting JPGs into a PowerPoint, but would certainly make putting R-output (numerical or graphical) into a presentation much easier!
Edit: if you want to use knitr (which I recommend), here's another reference.
I've been working on trying to automate the complicated process of building source code on a build machine and then transferring the compiled image files over to my embedded ARMv7 device to be flashed. Each step by itself is easy to automate with standard Linux Shell Script, but when trying to do everything in one giant script things get complicated. Thus far I've been using expect-lite to do the work, which is working except now I've run into a problem. When transferring the images over I have expect-lite code that looks like the following:
$imageDestination="/the/destination"
$imageSource="/the/source/"
>sftp $userName'#'$buildMachine
>$password
>get $imageSource'/'x-load_sdcard.bin.ift $imageDestination'/'MLO
>echo "Finished"
>bye
If you know a thing or two about expect-lite, then you'll know that the above variables will be read as "Shell" variables. The problem is that as far as I know SFTP doesn't allow the use of variables. Is there a way to tell expect-lite to use the predefined variables instead of trying to use "Shell" variables? Or, is there some cleaver way to get around this limitation without removing the variables?
All help is greatly appreciated.
Dreligor,
There is no scope issue. Expect-lite variables are all of global scope (as stated in the documentation). I think the problem is that you are using quotes which is making things more difficult. You should try:
$imageDestination=/the/destination
$imageSource=/the/source
>sftp $userName'#'$buildMachine
>$password
>get $imageSource/x-load_sdcard.bin.ift $imageDestination/MLO
>echo "Finished"
>bye
Craig Miller - author and maintainer of expect-lite
After some experimentation it turns out that this is a scope issue. The solution is to simply move the variable declarations down. They need to be declared after the script has connected to the remote machine via sftp. The fixed code is as follows:
>sftp $userName'#'$buildMachine
>$password
$imageDestination="/the/destination"
$imageSource="/the/source/"
>get $imageSource'/'x-load_sdcard.bin.ift $imageDestination'/'MLO
>echo "Finished"
>bye
Hopefully this will help others.