verdverm / go-pge Goto Github PK

I was able to successfully build and run go-pge. However, I have a few simple questions:

1. We would like to include PGE in a comparative study but we need to be able to run experiments successfully. What we need is to be able to provide the algorithm training data, let it crunch the numbers and then retrieve the final model or a set of models along with their training and testing performance using RMSE, or at least retrieve the models in a form that can be parsed and reused. However, go-pge (or rather scripts/test.sh) produces a bunch of files but it is by far not clear what is exactly contained in each file and what do the numbers mean. I can guess some of it based on the file names but it's only a guess. Could you provide some pointers as to where can we obtain these outputs?
2. What is the relation of go-pge and https://github.com/verdverm/pypge regarding the algorithm? Is go-pge the same algorithm as pypge for explicit problems? For our study, we like this go version a bit more as it does not involve multiple docker containers and has much more straightforward usage but the outputs are much less readable (hence the first question). However, pypge's outputs are not easy do read too. If pypge is the way to go, we have started a similar issue there: verdverm/pypge#12

If you could provide an answer the questions we would be happy to include your algorithm to our study and cite your paper.

PGE seems really cool. Thank you for releasing the code.

I've compiled and built ok. (BTW there are hardcoded /usr/tony paths in levmar.go.) I'm able to run the binary go-pge. And I'm able to change the config files e.g. to make it load the Korns-1 data.

But I am struggling to get it to actually run. When I run ./go-pge --cfg config/main/main_default.cfg I see successful loading of config files and data, and then I see this error:

Unknown init method
panic: runtime error: invalid memory address or nil pointer dereference
[signal 0xb code=0x1 addr=0x20 pc=0x4093f44]
[etc]

I have this in my main_default.cfg:

ProblemCfg = prob/prob_bench_default.cfg

and this in my prob_bench_default.cfg:

TrainData = benchmark/Korns_01.trn 
TestData = benchmark/Korns_01.tst