GENETIS Loop Work: Fixing Consol Errors By Julie Rolla
1. Python programs are not running: I think either (1) the user isn't module loading python, or (2) they are loading the wrong version of Python (remember there's a discrepancy for Python2, and Python3).
Solution: I added a line "module load python/3.6-conda5.2" in our araenv.sh script to address the issue. All of the users are sourcing the araenv.sh script in their .bashrc when they login. It's in the directory: /fs/project/PAS0654/BiconeEvolutionOSC. I also added a line in the XF_Loop.sh script (at the top in the "Initialization of Variables" section) that sources araenv.sh just in case the user hasn't put that in their .bashrc. The line reads: source /fs/project/PAS0654/BiconeEvolutionOSC/araenv.sh
Error message fixed:
0
Data successfully read. Data successfully written.
Fitness scores successfully written.
Fitness function concluded.
rm: cannot remove 'runData.csv': No such file or directory
Unable to parse the pattern
^CTraceback (most recent call last):
File "LRPlot.py", line 99, in <module>
plotLR(generations, lengths, radii, g.numGens, g.destination)
File "LRPlot.py", line 45, in plotLR
plt.show()
File "/usr/lib64/python2.7/site-packages/matplotlib/pyplot.py", line 143, in show
_show(*args, **kw)
File "/usr/lib64/python2.7/site-packages/matplotlib/backend_bases.py", line 108, in __call__
self.mainloop()
File "/usr/lib64/python2.7/site-packages/matplotlib/backends/backend_gtk.py", line 83, in mainloop
gtk.main()
KeyboardInterrupt
2. rm runData.csv error: The following error was printed to consol
rm: cannot remove 'runData.csv': No such file or directory
I found these line in section (E) of the loop.
cd "$WorkingDir"
rm runData.csv
python gensData.py 0
cd Antenna_Performance_Metric
python LRPlot.py "$WorkingDir" "$WorkingDir"/Run_Outputs/$RunName 1 $NPOP
cd ..
# Note: gensData.py floats around in the main dir until it is moved to
I think this error is fine. This is how it works: (Step 1) removes runData.csv, if it exists. (Step 2) Runs gensData.py, which creates a new runData.csv. We wanted to delete it before the .py program was run at the start of the run, because gensData.py appends data to runData.csv, if it already exists. (alternatively if runData.csv doesn't exists it creates it-- which is what we want in this case). That's also why we don't delete it later in the "looping part" of XF_Scripts.sh (part G). We want it to append at this point, so we only delete it in the very beginning to make sure it doesn't append to an old run.
3. cp: cannot stat 'Antenna_Performance_Metric/1.uan': No such file or directory (similarly for 2.uan, 3.uan...etc): This error exists just as the fitness score is being calculated -- ie part E. Here is the print:
cp: cannot stat 'Antenna_Performance_Metric/1.uan': No such file or directory
cp: cannot stat 'Antenna_Performance_Metric/2.uan': No such file or directory
Congrats on getting a fitness score!
File "FScorePlot.py", line 1
-*- coding: utf-8 -*-
^
IndentationError: unexpected indent
Note that we have a few errors here. Right now I am directly addressing the error with the "cp" command. It looks like these files are actually being saved as gen_individual#.uan. For example 0_1.uan. The current line in part E says:
for i in `seq 1 $NPOP`
do
#Remove if plotting software doesnt need
#cp data/$i.uan ${i}uan.csv
cp Antenna_Performance_Metric/$i.uan "$WorkingDir"/Run_Outputs/$RunName/${i}.uan
done
However, this is not how these files are being saved. I corrected the first iteration of the loop (ie Part E for generation 0) by editing the line to say:
for i in `seq 1 $NPOP`
do
#Remove if plotting software doesnt need
#cp data/$i.uan ${i}uan.csv
cp Antenna_Performance_Metric/0_$i.uan "$WorkingDir"/Run_Outputs/$RunName/0_${i}.uan
done
And for the looping part of XF_Loop.sh (ie part G) it has been edited to say:
for i in `seq 1 $NPOP`
do
# Gens data used to create a .csv file for the uan file for gain plotting
# cp Antenna_Performance_Metric/$i.uan ${i}uan.csv
cp Antenna_Performance_Metric/${gen}_${i}.uan "$WorkingDir"/Run_Outputs/$RunName/${gen}_${i}.uan
done
4. Made Number of Neutrinos thrown (NNU) in AraSim's setup.txt file a variable: I've added the following line in part D -- as well as the looping section in part G.
These lines replace the number of neutrinos thrown in our setup.txt AraSim file with what ever number you assigned NNT at the top of XF_Loop.sh in the "variables" section. "setup_dummy.txt" is a copy of setup.txt that has NNU=num_nnu instead of a number. NNU is the number of neutrinos thrown (as seen in setup.txt). This "sed" line finds every instance of num_nnu in setup_dummy.txt and replaces it with $NNT (the number you assigned NNT in the variable section at the top of XF_Loop.sh). It then pipes this information into setup.txt (and overwrites the last setup.txt file allowing the number of neutrinos thrown to be as variable changed at the top of this script instead of manually changing it in setup.txt each time. Command works the following way: sed "s/oldword/newwordReplacingOldword/" path/to/filewiththisword.txt > path/to/fileWeAreOverwriting.txt. Both setup.txt and setup_dummy.txt can be found in the directory: /fs/project/PAS0654/BiconeEvolutionOSC/AraSim/
cd "$AraSimExec"
for i in `seq 1 $NPOP`
do
#This next line replaces the number of neutrinos thrown in our setup.txt AraSim file with what ever number you assigned NNT at the top of this program. setup_dummy.txt is a copy of setup.txt that has \
NNU=num_nnu (NNU is the number of neutrinos thrown. This line finds every instance of num_nnu in setup_dummy.txt and replaces it with $NNT (the number you assigned NNT above). It then pipes this infor\
mation into setup.txt (and overwrites the last setup.txt file allowing the number of neutrinos thrown to be as variable changed at the top of this script instead of manually changing it in setup.txt e\
ach time. Command works the following way: sed "s/oldword/newwordReplacingOldword/" path/to/filewiththisword.txt > path/to/fileWeAreOverwriting.txt
sed "s/num_nnu/$NNT/" /fs/project/PAS0654/BiconeEvolutionOSC/AraSim/setup_dummy.txt > /fs/project/PAS0654/BiconeEvolutionOSC/AraSim/setup.txt
#We will want to call a job here to do what this AraSim call is doing so it can run in parallel
cd $WorkingDir
qsub -v num=$i AraSimCall.sh
rm outputs/*.root
done
5. rm cannot remove 'simulation_PEC.xmacro': No such file or directory error appearing in generations after the first (ie not seen in gen 0, but seen in gen 1+).
Statistics initialized.
Tournament parents selected.
Tournament complete.
rm: cannot remove 'simulation_PEC.xmacro': No such file or directory
I found that it tries to remove it twice. See here:
######## Loop (G) ######################################################################################
# 1. Does steps A-F for each generation
###########################################################################################################
for gen in `seq 1 $TotalGens`
do
# used for fixed number of generations
#gen=0; while [ `cat highfive.txt` -eq 0 ]; do (( gen++ ))
# use for runs until convergence
#should the line below be double indented?
read -p "Starting generation ${gen}. Press any key to continue... " -n1 -s
#Part A
cd $XmacrosDir
rm simulation_PEC.xmacro
cd "$WorkingDir"
./roulette_algorithm.exe cont $NPOP
cp generationDNA.csv Run_Outputs/$RunName/${gen}_generationDNA.csv
##chmod -R 777 $WorkingDir
#we can make a function that does this with the flag $gen
#doing this in part A would mean setting $gen=0
#Part B
# First, remove the old .xmacro files
#when do that, we end up making the files only readable; we should just overwrite them
#alternatively, we can just set them as rwe when the script makes them
cd $XmacrosDir
#I'm commenting out the next two lines. They are written to below (around lines 154-170)
#If we keep them this way then the files have restricted permissions
rm output.xmacro
rm simulation_PEC.xmacro
I have commented the first removal of simulation_PEC.xmacro. If there are any issues with it running, we can go ahead and comment it back in; however, I suspect this is not a necessary line to have added in. A run has not been performed since these edits. So, once it has been tested, we can delete the comments out:
#Part A
#I think these next two lines can be deleted. Its repeated again just below...11/30/19 comment by Julie #cd $XmacrosDir #rm simulation_PEC.xmacro
cd "$WorkingDir"
./roulette_algorithm.exe cont $NPOP
cp generationDNA.csv Run_Outputs/$RunName/${gen}_generationDNA.csv
6. Python Error from FScorePlot.py: The error can be seen in the line below.
Congrats on getting a fitness score!
File "FScorePlot.py", line 1
-*- coding: utf-8 -*-
^
IndentationError: unexpected indent
Congrats on getting some nice plots!
I've commented out this strange line at the start of this python program: -*- coding: utf-8 -*-. This fixed the immediate error, but threw another one when I tried to run it outside of the loop with the command
python FScorePlot.py /fs/project/PAS0654/BiconeEvolutionOSC/BiconeEvolution/current_antenna_evo_build/XF_Loop/Evolutionary_Loop/Run_Outputs/Julie_11_30/ /fs/project/PAS0654/BiconeEvolutionOSC/BiconeEvolution/current_antenna_evo_build/XF_Loop/Evolutionary_Loop/Run_Outputs/Julie_11_30/ 2
The next error it threw can be seen below.
Traceback (most recent call last):
File "FScorePlot.py", line 79, in <module>
fScores.append(fScorei[2:])
AttributeError: 'numpy.ndarray' object has no attribute 'append'
This is Alex P. I edited the append() and changed it to np.append(fScores, fScorei[2:]) on line 79 in FScorePlot.py
7. rm: cannot remove 'outputs/*.root': No such file or directory: This error comes up right after the AraSim job is submitted in part D.
1011007.pitzer-batch.ten.osc.edu
rm: cannot remove 'outputs/*.root': No such file or directory
Waiting for AraSim jobs to finish...
It's definitely coming from the lines:
for i in `seq 1 $NPOP`
do #We will want to call a job here to do what this AraSim call is doing so it can run in parallel
cd $WorkingDir
qsub -v num=$i AraSimCall.sh rm outputs/*.rootdone
We are currently in $WorkingDir at this point-- which is: WorkingDir= /fs/project/PAS0654/BiconeEvolutionOSC/BiconeEvolution/current_antenna_evo_build/XF_Loop/Evolutionary_Loop. There is not directory called "outputs" in this location. This is why we are getting an error.Can someone explain what .root files we are trying to remove, and why? I cannot proceed unless someone makes this clear to me, as I do not know what files we are looking to remove (and where they exist/when they are created).
Immediately after that, I get the following error:
Waiting for AraSim jobs to finish...
rm: remove write-protected regular file 'AraSimFlags/1.txt'? y
rm: remove write-protected regular file 'AraSimFlags/2.txt'? y
mv: cannot stat '*.root': No such file or directory
It looks like it is coming from the following line:
cd "$WorkingDir"/Antenna_Performance_Metric
mv *.root "$WorkingDir/Run_Outputs/$RunName/RootFilesGen0/"
However-- again-- it looks like it's trying to move .root files from the directory: /fs/project/PAS0654/BiconeEvolutionOSC/BiconeEvolution/current_antenna_evo_build/XF_Loop/Evolutionary_Loop/Antenna_Performance_Metric and put them into: /fs/project/PAS0654/BiconeEvolutionOSC/BiconeEvolution/current_antenna_evo_build/XF_Loop/Evolutionary_Loop/Run_Outputs/$RunName/RootFilesGen0/
It can't move these, because they do not exist where it is trying to take them from. Similarly, the question in bold above would help me with this a lot. From what I can see, we are only utilizing an .txt file output from Arasim. Does Arasim output this to .txt, or are we somehow converting from a .root to a .txt?
This error has not yet been corrected!!!!
Cade's comment here:
The .root files live in BiconeEvolutionOSC/AraSim/outputs/ folder. Thus, we were looking at the wrong directory. They were being removed because it is a giant file (GB size) and are created everytime AraSim runs. It just overwrites itself everytime though so it doesnt add up too much. We can have them deleted later if needed. It's our choice on whether we take it out completely, or redirect to delete the .root files in the correct directory.
8. New errors found after running with these corrections: These need to be fixed!
cp: cannot stat ‘Antenna_Performance_Metric/0_1.uan’: No such file or directory
cp: cannot stat ‘Antenna_Performance_Metric/0_2.uan’: No such file or directory
Congrats on getting a fitness score!
Traceback (most recent call last):
File "FScorePlot.py", line 86, in <module>
Gen = np.zeros(fScores.shape[0]*fScores.shape[1])
I found the cp error. It looks like I was incorrect in #3 about this cp command for the .uan files. I moved all of the currently existing .uan files into a new directory in
/fs/project/PAS0654/BiconeEvolutionOSC/BiconeEvolution/current_antenna_evo_build/XF_Loop/Evolutionary_Loop/Antenna_Performance_Metric/UAN
With all of them out of the Antenna_Performance_Metric directory, I did a run from scratch to see what it creates. Here's the outcome: So on generation 1, for individual one its going
1_1.uan
1_2.uan
1_3.uan
ie-- > individual_frequencyStep.uan
So once it hits all 60 frequencies for individual 1, it does all 60 for individual 2 by saying:
2_1.uan
2_2.uan
2_3.uan
...etc.This isn't great because it never actually cares about generation number. In section E (so generation 0) we currently have it as:
for i in `seq 1 $NPOP`
do
#Remove if plotting software doesnt need
#cp data/$i.uan ${i}uan.csv
cp Antenna_Performance_Metric/0_$i.uan "$WorkingDir"/Run_Outputs/$RunName/0_${i}.uan
done
and in section G of the looping part (gens 1+) we have:
for i in `seq 1 $NPOP`
do
# Gens data used to create a .csv file for the uan file for gain plotting
# cp Antenna_Performance_Metric/$i.uan ${i}uan.csv
cp Antenna_Performance_Metric/${gen}_${i}.uan "$WorkingDir"/Run_Outputs/$RunName/${gen}_${i}.uan
done
This doesn't seem like it's being output from XF the way we thought it is. I think the easiest way to do this is to edit the xmacros to have it output like:
generation_individual_frequency.uan
and thus
cp Antenna_Performance_Metric/${gen}_${i}_${frequency}.uan
If so, we'd have to make a new variable in the bash script for $frequency to go from 1-60, and have it loop moving all of those. I will be looking into this with Cade tomorrow 12/3/19 (since he is the xmacro script expert). Stay tuned!
Phew....thanks for sticking with me on this long update. |