GridUtilities.jl

Utility functions for discrete fields on grids
Author JuliaIBPM
Popularity
4 Stars
Updated Last
2 Months Ago
Started In
June 2020

GridUtilities.jl

Utility functions for discrete fields defined on grids

Build Status
Build Status Coverage

About the package

This package provides various utility functions for working with fields defined on Cartesian grids or on points immersed in these grids. This includes functions for

  • Storing histories of these fields
  • Creating interpolatable versions of the fields that can be interrogated in function-like manner
  • Storing data, writing to file, and loading from file

This is a registered package in Julia, so it can be installed in the usual way,

] add GridUtilities

Then type

using GridUtilities

Data sampling

Here, we will sample and store data from a simple update model

u0 = ones(5,5)
t = 0.0
u = deepcopy(u0)

Now we set up a StorePlan to provide details on what we wish to store and how often. We assign names to each variable.

t_sample = 0.05
S = StorePlan(t_sample,"state" => u0, "time" => t)
data_history = initialize_storage(S)

When we advance a simple dynamical model, we store the data at the sampling interval with store_data!

store_data!(data_history,t,S,"state"=>deepcopy(u),"time"=>t) # initial state
for i in 1:100
    u .+= u0 # simple model, just for making it dynamic
    t += 0.01
    store_data!(data_history,t,S,"state"=>u,"time"=>t)
end

The data is stored in a Dict structure that is easy to inspect:

data_history["state"][3]
data_history["time"][3]

The package uses JLD to save and load data. To save the full set of stored data, e.g.,

save("stuff.jld","data",data_history)

and to load it back in

d = load("stuff.jld")

and, e.g., type

d["data"]["state"]

to get the history of the state variable.

Writing data to file

The package also has capabilities for writing data periodically to file (i.e., writing a restart file). Let's see how this would work

u0 = ones(5,5)
t = 0.0
u = deepcopy(u0)

We set up a WritePlan for this periodic storage. Here, we specify writing the data every 0.1 time increments.

filen = "restart.jld"
restart_Δt = 0.1
R = WritePlan(filen,restart_Δt,"state" => u,"t" => t)

Now, in the loop, we use an extended version of the JLD.save function.

for i in 1:100
    u .+= u0 # simple model, just for making it dynamic
    t += 0.01
    save(t,R,"state" => u,"t" => t)
end

We can restart this by loading from the file

restart = load(R)
u = restart["state"]
t = restart["t"]

and keep running

for i in 1:100
    u .+= u0 # simple model, just for making it dynamic
    t += 0.01
    save(t,R,"state" => u,"t" => t)
end

Used By Packages

No packages found.