Installation instructions + schedule test (#979)

* Installation instructions + schedule test

* bump version to 0.18

* bump versions of submodules

Author: milankl

CHANGELOG.md

@@ -2,6 +2,9 @@
 
 ## Unreleased
 
+## v0.18
+
+- Installation main instructions in docs + schedule tests [#979](https://github.com/SpeedyWeather/SpeedyWeather.jl/pull/979)
 - Bug fix: Cloud reflectivity on half level above not below [#978](https://github.dev/SpeedyWeather/SpeedyWeather.jl/pull/978)
 - Fix absorptivity_water_vapor and absorptivity_cloud_base default values [#974](https://github.com/SpeedyWeather/SpeedyWeather.jl/pull/974)
 - Particle advection GPU ready [#897](https://github.com/SpeedyWeather/SpeedyWeather.jl/pull/897)

LowerTriangularArrays/Project.toml

@@ -1,7 +1,7 @@
 name = "LowerTriangularArrays"
 uuid = "ed20b33f-a660-45d0-a353-9c8c20bb5814"
 authors = ["LowerTriangularArrays contributors"]
-version = "0.1.1"
+version = "0.1.2"
 
 [deps]
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"

RingGrids/Project.toml

@@ -1,7 +1,7 @@
 name = "RingGrids"
 uuid = "d1845624-ad4f-453b-8ff4-a8db365bf3a7"
 authors = ["RingGrids contributors"]
-version = "0.1.1"
+version = "0.1.2"
 
 [deps]
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"

SpeedyTransforms/Project.toml

@@ -1,7 +1,7 @@
 name = "SpeedyTransforms"
 uuid = "ed023a3f-7a3c-42dd-bab6-68c21c8c3105"
 authors = ["SpeedyTransforms contributors"]
-version = "0.1.1"
+version = "0.1.2"
 
 [deps]
 AbstractFFTs = "621f4979-c628-5d54-868e-fcf4e3e8185c"

SpeedyWeather/Project.toml

@@ -1,7 +1,7 @@
 name = "SpeedyWeather"
 uuid = "9e226e20-d153-4fed-8a5b-493def4f21a9"
 authors = ["SpeedyWeather contributors"]
-version = "0.17.4"
+version = "0.18"
 
 [deps]
 AbstractFFTs = "621f4979-c628-5d54-868e-fcf4e3e8185c"

SpeedyWeather/src/dynamics/clock.jl

@@ -34,14 +34,16 @@ Adapt.adapt_structure(to, ::Clock) = nothing
 function timestep!(clock::Clock, Δt; increase_counter::Bool = true)
     clock.time += Δt
     # the first timestep is a half-step and doesn't count
-    return clock.timestep_counter += increase_counter
+    clock.timestep_counter += increase_counter
+    return nothing
 end
 
 # pretty printing
 function Base.show(io::IO, C::Clock)
     println(io, "$(typeof(C))")
     keys = propertynames(C)
-    return print_fields(io, C, keys)
+    print_fields(io, C, keys)
+    return nothing
 end
 
 # copy!

SpeedyWeather/src/output/schedule.jl

@@ -11,16 +11,16 @@ start+p, but p is rounded to match the multiple of the model timestep.
 Periodic schedules and event schedules can be combined, executing at both.
 A Schedule is supposed to be added into callbacks as fields
 
-    Base.@kwdef struct MyCallback
+    @kwdef struct MyCallback
         schedule::Schedule = Schedule(every=Day(1))
         other_fields
     end
 
-see also `initialize!(::Schedule,::Clock)` and `isscheduled(::Schedule,::Clock)`.
+see also `initialize!(::Schedule, ::Clock)` and `isscheduled(::Schedule, ::Clock)`.
 
 Fields
 $(TYPEDFIELDS)"""
-Base.@kwdef mutable struct Schedule <: AbstractSchedule
+@kwdef mutable struct Schedule <: AbstractSchedule
     "[OPTION] Execute every time period, first timestep excluded. Default=never."
     every::Second = Second(typemax(Int))
 
@@ -37,19 +37,18 @@ Base.@kwdef mutable struct Schedule <: AbstractSchedule
     counter::Int = 0
 end
 
-"""
-$(TYPEDSIGNATURES)
+"""$(TYPEDSIGNATURES)
 A Schedule based on DateTime arguments, For two consecutive time steps i, i+1, an event is
 scheduled at i+1 when it occurs in (i,i+1]."""
 Schedule(times::DateTime...) = Schedule(times = DateTime[times...])
 
-"""
-$(TYPEDSIGNATURES)
+"""$(TYPEDSIGNATURES)
 Initialize a Schedule with a Clock (which is assumed to have been initialized).
 Takes both scheduler.every and scheduler.times into account, such that
 both periodic and events can be scheduled simultaneously. But execution will
 happen only once if they coincide on a given time step."""
 function initialize!(scheduler::Schedule, clock::Clock)
+    clock.Δt == Millisecond(0) && error("Clock needs to be initialized with a non-zero time step before initializing schedule.")
     schedule = falses(clock.n_timesteps)    # initialize schedule as BitVector
 
     # PERIODIC SCHEDULE, always AFTER scheduler.every time period has passed

SpeedyWeather/test/output/schedule.jl

@@ -16,5 +16,28 @@
         # adapted schedule time step should be within 20%
         @test schedule.every.value ≈ Second(hour).value rtol = 2.0e-1
         @test schedule.steps ≈ period / hour rtol = 2.0e-1
+
+        for _ in 1:clock.n_timesteps
+            SpeedyWeather.timestep!(clock, clock.Δt)
+            isscheduled(schedule, clock)
+        end
+
+        @test schedule.counter == sum(schedule.schedule) > 1
+
+        # reinitialize clock
+        clock = Clock()
+        period = Day(1)
+        initialize!(clock, time_stepping, period)
+
+        schedule = Schedule(clock.time + Hour(6))
+        initialize!(schedule, clock)
+        @test sum(schedule.schedule) == 1
+        @test schedule.counter == 0     # no execution of isscheduled yet
+
+        for _ in 1:clock.n_timesteps
+            SpeedyWeather.timestep!(clock, clock.Δt)
+            isscheduled(schedule, clock)
+        end
+        @test schedule.counter == 1
     end
 end

SpeedyWeatherInternals/Project.toml

@@ -1,7 +1,7 @@
 name = "SpeedyWeatherInternals"
 uuid = "34489162-d270-4603-8b96-37b04f830d73"
 authors = ["SpeedyWeatherInternals contributors"]
-version = "0.1.2"
+version = "0.1.3"
 
 [deps]
 ComponentArrays = "b0b7db55-cfe3-40fc-9ded-d10e2dbeff66"

docs/src/differentiability.md

@@ -2,12 +2,12 @@
 
 SpeedyWeather.jl is written with differentiability in mind. This means that our model is differentiable by automatic differentiation (AD). If you are interested in machine learning (ML), this means that you can integrate our model directly into your ML models without the need to first train your neural networks offline. For atmospheric modellers this means that you get an adjoint model for free which is always generated automatically, so that we don't need to maintain it separately. This allows you to calibrate SpeedyWeather.jl in a fully automatic and data-driven way.
 
-!!! warning Work in progress
+!!! warning "Work in progress"
     The differentiability of SpeedyWeather.jl is still work in progress and some parts of this documentation might be not be always updated to the latest state. We will extend this documentation over time. Don't hesitate to contact us via GitHub issues or mail when you have questions or want to collaborate.
 
 For the differentiability of our model we rely on [Enzyme.jl](https://github.com/EnzymeAD/Enzyme.jl). If you've used Enzyme before, just go ahead and try to differentiate the model! It should work. We have checked the correctness of the gradients extensively against a finite differences differentiation with [FiniteDifferences.jl](https://github.com/JuliaDiff/FiniteDifferences.jl/). In the following we present a simple example how we can take the gradient of a single timestep of the primitive equation model with respect to one of the model parameter.
 
-!!! warning Enzyme with Julia 1.11
+!!! warning "Enzyme with Julia 1.11"
     Currently there are still some issues with Enzyme in Julia 1.11, we recommend to use Julia 1.10 for the following
 
 ## Differentiating through a single timestep