Closed williamty closed 2 years ago
Well I can see why 6.9.0 would behave this way but not 7.1.0; there must be something I don't understand, since the cause of that behavior had been removed from the code. It doesn't behave that way on my mac (I just checked). But I have a Windows 10 machine right now that I had borrowed from the IT department. I'm returning it tomorrow, but maybe before I do I can check this behavior on it, see if I can reproduce this problem.
I used RFCI algorithm and mixed data as you know. I will download 7.1.0 again and try.
It happens again after running the Jar file(tetrad-gui-7.1.0-launch.jar) about 2-3 minutes. :-(
Can I run Tetrad jar file through command line? I'm not familiar with java.
Wait, are you running out of memory? That could slow it down and cause it to hang. Try this at the command line:
java -Xmx[#]g -jar [name of jar].jar
where # is the number of gigabytes of RAM you're willing to allow Java to use to allocate objects.
I don't think so. The RAM of my laptop is 24GB, and I never see it's used to more than 20GB.
It still happens after setting the RAM to 12GB.
D:\program>java -Xmx12g -jar tetrad-gui-7.1.0-launch.jar
config file: /config/prodConfig.xml
Infer demiliter for file: w.csv
File index = 0 has been loaded successfully
SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See http://www.slf4j.org/codes.html#StaticLoggerBinder for further details.
Searching at depth 0.
You are using conditional Gaussian? Could you use degenerate Gaussian instead? I am worried about your discrete variable with 31 categories; for CG that can imply sizable data structures for storing the information; DG separates out the discrete categories into separate continuous-valued variables and so can in principle avoid this problem.
I used DG instead of CG. cause I've read the papers, and I think maybe DG is better.
Oh it's using testwise deletion because of the missing values. How many missing values are there?
Also, can you impute the missing values?
yes, I have a dataset with imputation data. I forgot the number... maybe there are about 3000 to 5000 missing data.
The same. It's no use to change to imputated data. I found Tetrad always called system interrupt and I don't know why.
Wait your imputed data has not missing data?
And it still objects?
It's very difficult to debug this without seeing the data. I often have this problem though.
How about I give you a mixed dataset that I know loads in Tetrad?
OK, no problem. But I think the problem is not caused by data. I can use the mixed dataset loaded in Tetrad. I think maybe something is wrong between the system and java.
Just windows 10 family. not occur in windows 10 enterprise.
After running the Tetrad 6.9.0 or 7.1.0 jar file on Windows 10, I minimized its window. But the minimized Tetrad program would continue trying to set itself to be the first running process since then, and attempting to pop up, and I can do nothing because of losing control my keyboard and mouse, which was occupied by Tetrad. Then I set the priority of java to low in Windows task-manager, but it doesn't work. How to deal with that?