Automate common data analysis and visualization tasks using the CERN pyroot library in a jupyter notebook.
Requires prior installation of pyroot (ROOT package) and dask, dask.distributed if parallel filling of histograms using a dask cluster is desired. Simple example of jupyter notebooks using the jupyroot classes are included in the repository.
class treeview()Manages a user-defined collection of ROOT histograms in the form of user provided functions for declaring each histogram and filling it from the contents of a ROOT tree chain. Automatic means are provided for running the fill function over a chain of tree files, and caching the result in a local ROOT file for quick access the next time it is requested. All histograms declared to a single treeview object must fill from the same input tree chain.
def __init__(treechain, savetoroot)Constructor required arguments:
- treechain = ROOT TChain object that has been configured with the list of ROOT files containing the rows of a ROOT tree to be analyzed, given in no particular order
- saveto = name of a locally writable ROOT file where user histograms are cached after filling; if more than one treeview object shares the same file then append a directory to the filename, as in saveto="myview.root/view_1"
def declare_histograms(setname, initfunc, fillfunc)Declares a list of user-defined histogram to treeview, with arguments:
- setname = user-defined unique name for this set of histograms
- initfunc = user-defined function that creates an empty instance of these histograms with title and axis labels assigned
- fillfunc = user-defined function that accepts a row of a ROOT tree and the list of histograms to be filled as its arguments, and fills the histograms from the contents of the tree Return value is the count of histograms declared.
def list_histograms(cached=False)Outputs a list of currently declared histograms managed by this object, returns the count of histograms found.
def enable_dask_cluster(client)Enable parallel filling of histograms from a chain of input ROOT tree files, using a dask cluster session provided by the user.
def fill_histograms(chunksize=1)Scan over the full chain of input ROOT files and fill all histograms that need filling if any, otherwise return immediately. Return value is the number of histograms that were updated.
- chunksize - (int) number of input files to process in one dask process, only relevant if parallel dask algorithm is enabled
def setup_canvas(prefix="canvas", width=500, height=400)Create a new ROOT canvas for plotting in the output window of a cell in a jupyter notebook, returns the name.
def update_canvas(cname=None)Loop through the list of canvases with active plots in them and issue an update to refresh the display in jupyterhub.
def draw(histos, options="", width=500, height=400, titles=1, stats=1, fits=0)Emulates the TH1.Draw() method for an array of histograms.
- histos - name of histogram to draw, or multidimensional array of the names of histograms to be drawn; [..] if histos is a one-dimensional list then the histograms identified in the list are drawn in equal-size divisions across a single row, with each plot occupying width x height in the drawing area of the canvas; [[..]..] if histos is a two-dimensional list then the histograms are drawn on a rectangular grid of equal-size boxes, each of dimension width x height; [[[..]..]..] if histos is a three-dimensional list then it is treated like a two-dimensional array, with the histograms listed in the third dimension all drawn on the same plot;
- options - a string or array of strings with the same shape as histos, that is passed to TH1.Draw in the option argument;
- width - optional argument giving the pixel width of each plot;
- height - optional argument giving the pixel height of each plot;
- titles - 1/0 for whether to display histogram titles or not;
- stats - 1/0 for whether to display statistics boxes or not;
- fits - 1/0 for whether to display fit parameter boxes or not; Return value is the total number of histograms plotted.
def get(hname)Look for histogram named hname in the cache and return if found, otherwise look for an empty histogram in the defined histodefs with this name and return the empty copy, else report error and return None.
def put(hist)Save histogram hist in the cache root directory, and return True for success, False for failure.
def dask_dashboard_link()The default behavior of the dask client is to provide a local url for the dask dashboard that is only usable on the local jupyterhub host. This method returns a link that should work anywhere on the internet, assuming the jupyterhub host is not blocked by internet firewalls.
def dump_histodefs()Scans through the histodefs dictionary and lists all of the objects that are registered within the data structure. If any are TH1 objects then they are checked that they have not been deleted at some point leaving an invalid reference behind.
class hddmview()Manages a user-defined collection of ROOT histograms in the form of user provided functions for declaring each histogram and filling it from the contents of a hddm record. Automatic means are provided for running the fill function over the list of input hddm files, and caching the result in a local ROOT file for quick access the next time it is requested. All histograms declared to a single hddmview object must fill from the same list of input hddm files.
def __init__(inputfiles, hddmclass, savetoroot)Constructor required arguments:
- inputfiles = list of input hddm files, either as regular unix pathnames or urls of the form root://path/to/file.hddm or http[s]://path/to/file.hddm
- hddmclass = reference to the istream class provided by the hddm module that decodes the data in the inputfiles, currently either gluex.hddm_s or gluex.hddm_r
- saveto = path to a locally writable ROOT file where user histograms are cached after filling; if more than one hddmview object shares the same file then append a directory to the filename, as in saveto="myview.root/view_1"
def declare_histograms(setname, initfunc, fillfunc)Declares a list of user-defined histogram to hddmview, with arguments:
- setname = user-defined unique name for this set of histograms
- initfunc = user-defined function that creates an empty instance of these histograms with title and axis labels assigned.
- fillfunc = user-defined function that accepts a hddm record and the list of histograms to be filled as its arguments, and fills the histograms from the contents of the hddm record. Return value is the count of histograms declared.
def list_histograms(cached=False)Outputs a list of currently declared histograms managed by this object, returns the count of histograms found.
def enable_dask_cluster(client)Enable parallel filling of histograms from different input hddm files, using a dask cluster session provided by the user, no return value.
def fill_histograms(chunksize=1)Scan over the full list of input hddm files and fill all histograms that need filling if any, otherwise return immediately. Return value is the number of histograms that were updated.
- chunksize - (int) number of input files to process in one dask process, only relevant if parallel dask algorithm is enabled
def setup_canvas(prefix="canvas", width=500, height=400)Create a new ROOT canvas for plotting in the output window of a cell in a jupyter notebook, returns the name.
def update_canvas(cname=None)Loop through the list of canvases with active plots in them and issue an update to refresh the display in jupyterhub.
def draw(histos, options="", width=500, height=400, titles=1, stats=1, fits=0)Emulates the TH1.Draw() method for an array of histograms.
- histos - name of histogram to draw, or multidimensional array of the names of histograms to be drawn; [..] if histos is a one-dimensional list then the histograms identified in the list are drawn in equal-size divisions across a single row, with each plot occupying width x height in the drawing area of the canvas; [[..]..] if histos is a two-dimensional list then the histograms are drawn on a rectangular grid of equal-size boxes, each of dimension width x height; [[[..]..]..] if histos is a three-dimensional list then it is treated like a two-dimensional array, with the histograms listed in the third dimension all drawn on the same plot;
- options - a string or array of strings with the same shape as histos, that is passed to TH1.Draw in the option argument;
- width - optional argument giving the pixel width of each plot;
- height - optional argument giving the pixel height of each plot;
- titles - 1/0 for whether to display histogram titles or not;
- stats - 1/0 for whether to display statistics boxes or not;
- fits - 1/0 for whether to display fit parameter boxes or not; Return value is the total number of histograms plotted.
def get(hname)Look for histogram named hname in the cache and return if found, otherwise look for an empty histogram in the defined histodefs with this name and return the empty copy, else report error and return None.
def put(hist)Save histogram hist in the cache root directory, and return True for success, False for failure.
def dask_dashboard_link()The default behavior of the dask client is to provide a local url for the dask dashboard that is only usable on the local jupyterhub host. This method returns a link that should work anywhere on the internet, assuming the jupyterhub host is not blocked by internet firewalls.
def dump_histodefs()Scans through the histodefs dictionary and lists all of the objects that are registered within the data structure. If any are TH1 objects then they are checked that they have not been deleted at some point leaving an invalid reference behind.