Hi,
I am very happy about the new RDataFrame and am using it frequently.
I am using RDataFrame with PyROOT in version 6.16.00.
I want to apply some Filters to the RDataFrame and then visualize all filter nodes with the ROOT.RDF.SaveGraph() function.
From the documentation I get, that for this I have to call the function on the head node. But by doing this I get only an empty .dot file. Calling the function on a daughter node on the other hand produces the expected (sub)graph.
I am using the following code to get this result:
import ROOT
import numpy as np
def fill_tree(treeName, fileName):
ROOT.gRandom.SetSeed(1)
tdf = ROOT.ROOT.RDataFrame(100)
tdf.Define("b1", "gRandom->Gaus()")\
.Define("b2", "gRandom->Gaus()")\
.Define("b3", "gRandom->Gaus()")\
.Define("b4", "gRandom->Gaus()")\
.Snapshot(treeName, fileName)
def fslice(frame, binedges, varname):
if binedges == ():
return ()
left, right = np.array_split(binedges, 2)
if len(left) == 0:
return ()
splitval = left[-1]
if len(left) == 1:
return (
frame.Filter(
"{1} < {0}".format(splitval, varname),
"{1} < {0}".format(splitval, varname)),
frame.Filter(
"{1} > {0}".format(splitval, varname),
"{1} > {0}".format(splitval, varname))
)
else:
return fslice(
frame.Filter(
"{1} < {0}".format(splitval, varname),
"{1} < {0}".format(splitval, varname)),
left[:-1],
varname
) + fslice(
frame.Filter(
"{1} > {0}".format(splitval, varname),
"{1} > {0}".format(splitval, varname)),
right,
varname)
treeName = 'tree'
fileName = treeName+".root"
fill_tree(treeName, treeName+".root")
f = ROOT.ROOT.RDataFrame(treeName, fileName)
ROOT.RDF.SaveGraph(f, "graph.dot")
frames = fslice(f, np.linspace(-1,1,6), "b1")
ROOT.RDF.SaveGraph(frames[-1], "graph1.dot")
ROOT.RDF.SaveGraph(f, "graph2.dot")
Even with a very simple
f = ROOT.ROOT.RDataFrame(treeName, fileName)
f2 = f.Filter("b2 > 2.")
ROOT.RDF.SaveGraph(f, "graph.dot")
I get only a empty graph.dot file with:
digraph {
}
Can somebody help me getting the complete graphs?
Cheers,
Christian