Zipper.jl

Implementation of Zipper Entanglement Renormalization on Julia platform.

Plotting data

Zipper.jl uses Plots.jl as its plotting backend, and the plotting API can be accessed via Zipper.visualize.

Selecting backend for `Plots.jl`

One can specify the plotting backend for Plots.jl by a specified function and with the certain backend package installed, for example the plotting backend tested in Zipper.jl is PlotlyJS, and we have to setup the environment before plotting data using Zipper.visualize.

# Include the module to gain access to plotlyjs() method.
import Plots
# Specify PlotlyJS as the backend.
plotlyjs()

Before doing so one have to install PlotlyJS via Pkg.add("PlotlyJS"), by default PlotlyJS is included in the Zipper.jl package.

`Element` type that supports `Zipper.visualize`

You can use visualize directly with T<:FockMap, Subset{T<:Point}, RegionState{2}, CrystalSpectrum{1}, CrystalSpectrum{2}, SnappingResult and EigenSpectrum.

Modifying plots beyond `Zipper.visualize`

If you want to modify the plots after they have been shown, you can use Plots.plot! to adjust the figures. For example modifying the size,

state::RegionState{2} = ...
visualize(state)

# Change the plot size to 800x800.
plot!(size=(800, 800))

Noted that not all functions support this feature, the notable one is Zipper.visualize(::CrystalSpectrum{2}) which utilizes mesh3d from PlotlyJS directly, thus the plot cannot be modified using plot!.

Using multithreading to enhance performance

Zipper.jl by default disabled support for multithreading, by setting up multithreading environment by Zipper.setmaxthreads it will utilize all threads available to the Julia environment to enhance the performance of many functions related to the generation or operation of type CrystalFock and CrystalFockMap.

Parallel APIs

ParallelTasks defines a structure that stores the information of a set of parallel tasks.
paralleltasks is the main entry of using parallel computing in Zipper.jl, it creates an instance of Zipper.ParallelTasks.
parallel is to execute the set of parallel tasks encapsulated by Zipper.ParallelTasks.
setmaxthreads is to control how many threads Zipper will use during computations.
showtaskmeter is to control whether the progress bar will appear when parallel computing is being performed.

Progress bars are available by default for any methods that uses the parallel computing APIs so that user can track and estimate the time required to complete the operation.

Enable parallel computing in Visual Studio Code

If the Julia environment in Visual Studio Code does not provide multithreading capabilities, Zipper.jl cannot take advantage of parallel computing thus no performance enhancements will be made.

To allow Julia REPL in Visual Studio Code to use multiple threads, head to Julia settings in the settings.json, search for the setting julia.additionalArgs and add --threads=auto to the list if you want Julia to decide how many threads it will use in the REPL, which most likely being the number of logical processors your native computing environment provides, or --threads=N to manually set the N number of threads that Julia can use. Then adjust the number of threads Zipper can use by setmaxthreads(N).

Save/Load Zipper data objects

Zipper.jl supports saving/loading data objects from JSON files, so that data that requires long period of computation time can be loaded with ease. This API also support defining custom serialization rules for special data types. Noted that normal Julia types are also supported by fioload and fiosave.

Fio APIs

fiodir Set the current project directory of the session, this is the directory which all saving/loading data will be performed on.
fiolower Define a special lower function for the serialization of the object attributes, the lower function will be applied to the object before JSON serializes the object into a JSON string.
fioconstructor Define a special constructor function for the deserialization of the object, the constructor function will be called after all object attributes are deserialized.
fioparser Define a special parser function for the deserialization of the object attributes, the parser function will be applied to the attribute value after the object is deserialized into a Dict.
fiostoragetype Register a storage type for the given Julia type, the storage type will be serialized instead of the actual type when saving the object to a file. During deserialization the storage type will be deserialized and converted back to the actual Julia type. Please be noted that the storage type must be a subtype of Element and have the function convert(::Type{StorageType}, ::JuliaType) and convert(::Type{JuliaType}, ::StorageType) defined.
fiosave Save the given object to a file with the given name in the current project directory defined in fiodir(), the file name will be in the format of {name}.json.
fioload Load the object from the file with the given name from the current project directory defined in fiodir(), the file to be loaded will be in the path {project directory}/{name}.json. Depands on the type some object might require extra data file to be loaded.

Alphaharrius/Zipper.jl