Decrease relative tolerance from 1e-5 to 1e-7

core/integration_test/hws_integration_test.jl

@@ -17,13 +17,13 @@
 
  @testset "Results values" begin
  @test basin.node_id == basin_bench.node_id
- @test all(q -> abs(q) < 0.06, basin.level - basin_bench.level)
+ @test all(q -> abs(q) < 0.2, basin.level - basin_bench.level)
  end
 
  timed = @timed Ribasim.run(toml_path)
 
- # current benchmark is 600s
- benchmark_runtime = 600
+ # current benchmark in seconds, TeamCity is up to 4x slower than local
+ benchmark_runtime = 32
  performance_diff =
  round((timed.time - benchmark_runtime) / benchmark_runtime * 100; digits = 2)
  if performance_diff < 0.0

core/regression_test/regression_test.jl

@@ -8,7 +8,7 @@
  config = Ribasim.Config(toml_path)
 
  solver_list =
- ["QNDF", "Rosenbrock23", "TRBDF2", "Rodas5", "KenCarp4", "Tsit5", "ImplicitEuler"]
+ ["QNDF", "Rosenbrock23", "TRBDF2", "Rodas5P", "KenCarp4", "Tsit5", "ImplicitEuler"]
  sparse_on = [true, false]
  autodiff_on = [true, false]
 
@@ -38,13 +38,13 @@
  # subgrid = Arrow.Table(subgrid_bytes)
 
  @testset "Results values" begin
- @test basin.storage[1] ≈ 1.0
- @test basin.level[1] ≈ 0.044711584
- @test basin.storage[end] ≈ 16.530443267
+ @test basin.storage[1] ≈ 1.0f0
+ @test basin.level[1] ≈ 0.044711584f0
+ @test basin.storage[end] ≈ 16.530443267f0
  @test basin.level[end] ≈ 0.181817438
- @test flow.flow_rate[1] == basin.outflow_rate[1]
+ @test flow.flow_rate[1] ≈ basin.outflow_rate[1]
  @test all(q -> abs(q) < 1e-7, basin.balance_error)
- @test all(q -> abs(q) < 0.01, basin.relative_error)
+ @test all(err -> abs(err) < 0.01, basin.relative_error)
  end
  end
  end
@@ -105,7 +105,7 @@ end
  @test all(q -> abs(q) < 1.0, basin.storage - basin_bench.storage)
  @test all(q -> abs(q) < 0.5, basin.level - basin_bench.level)
  @test all(q -> abs(q) < 1e-3, basin.balance_error)
- @test all(q -> abs(q) < 2.5, basin.relative_error)
+ @test all(err -> abs(err) < 2.5, basin.relative_error)
  end
  end
  end
@@ -127,7 +127,7 @@ end
  flow_bench = Arrow.Table(flow_bytes_bench)
  basin_bench = Arrow.Table(basin_bytes_bench)
 
- # TODO "Rosenbrock23" and "Rodas5" solver are resulting unsolvable gradients
+ # TODO "Rosenbrock23" and "Rodas5P" solver are resulting unsolvable gradients
  solver_list = ["QNDF"]
  sparse_on = [true, false]
  autodiff_on = [true, false]
@@ -167,7 +167,7 @@ end
  @test basin.node_id == basin_bench.node_id
  @test all(q -> abs(q) < 100.0, basin.storage - basin_bench.storage)
  @test all(q -> abs(q) < 0.5, basin.level - basin_bench.level)
- @test all(q -> abs(q) < 1e-3, basin.balance_error)
+ @test all(err -> abs(err) < 1e-3, basin.balance_error)
  end
  end
  end

core/src/bmi.jl

@@ -48,7 +48,7 @@ function BMI.get_value_ptr(model::Model, name::AbstractString)::AbstractVector{F
  p.subgrid.level
  elseif name == "user_demand.demand"
  vec(p.user_demand.demand)
- elseif name == "user_demand.inflow"
+ elseif name == "user_demand.cumulative_inflow"
  u.user_demand_inflow
  else
  error("Unknown variable $name")

core/src/config.jl

@@ -100,8 +100,8 @@ const nodetypes = collect(keys(nodekinds))
  dtmin::Float64 = 0.0
  dtmax::Union{Float64, Nothing} = nothing
  force_dtmin::Bool = false
- abstol::Float64 = 1e-6
- reltol::Float64 = 1e-5
+ abstol::Float64 = 1e-7
+ reltol::Float64 = 1e-7
  water_balance_abstol::Float64 = 1e-3
  water_balance_reltol::Float64 = 1e-2
  maxiters::Int = 1e9

core/test/bmi_test.jl

@@ -11,11 +11,12 @@
  @test BMI.get_end_time(model) ≈ 3.16224e7
  BMI.update(model)
  @test BMI.get_current_time(model) ≈ dt0 atol = 5e-3
- # cannot go back in time
- @test_throws ErrorException BMI.update_until(model, dt0 / 2.0)
  @test BMI.get_current_time(model) ≈ dt0 atol = 5e-3
  BMI.update_until(model, 86400.0)
  @test BMI.get_current_time(model) == 86400.0
+ # cannot go back in time
+ @test_throws ErrorException BMI.update_until(model, 3600.0)
+ @test BMI.get_current_time(model) == 86400.0
 end
 
 @testitem "fixed timestepping" begin
@@ -67,7 +68,7 @@ end
  "basin.cumulative_drainage",
  "basin.subgrid_level",
  "user_demand.demand",
- "user_demand.inflow",
+ "user_demand.cumulative_inflow",
  ]
  value_first = BMI.get_value_ptr(model, name)
  BMI.update_until(model, 86400.0)
@@ -86,7 +87,7 @@ end
  config = Ribasim.Config(toml_path; allocation_use_allocation = false)
  model = Ribasim.Model(config)
  demand = BMI.get_value_ptr(model, "user_demand.demand")
- inflow = BMI.get_value_ptr(model, "user_demand.inflow")
+ inflow = BMI.get_value_ptr(model, "user_demand.cumulative_inflow")
  # One year in seconds
  year = model.integrator.p.user_demand.demand_itp[2][1].t[2]
  demand_start = 1e-3

core/test/control_test.jl

@@ -247,7 +247,7 @@ end
 
  t_switch = Ribasim.datetime_since(record.time[2], p.starttime)
  flow_table = DataFrame(Ribasim.flow_table(model))
- @test all(filter(:time => time -> time <= t_switch, flow_table).flow_rate .> 0)
+ @test all(filter(:time => time -> time <= t_switch, flow_table).flow_rate .> -1e-12)
  @test all(
  isapprox.(
  filter(:time => time -> time > t_switch, flow_table).flow_rate,

core/test/docs.toml

@@ -31,8 +31,8 @@ dt = 60.0 # optional, remove for adaptive time stepping
 dtmin = 0.0 # optional, default 0.0
 dtmax = 0.0 # optional, default length of simulation
 force_dtmin = false # optional, default false
-abstol = 1e-6 # optional, default 1e-6
-reltol = 1e-5 # optional, default 1e-5
+abstol = 1e-7 # optional, default 1e-7
+reltol = 1e-7 # optional, default 1e-7
 water_balance_abstol = 1e-3 # optional, default 1e-3
 water_balance_reltol = 1e-2 # optional, default 1e-2
 maxiters = 1e9 # optional, default 1e9

docs/concept/equations.qmd

@@ -157,8 +157,8 @@ For more a more in-depth discussion of numerical computations see [Numerical con
 
 There are many things that can influence the calculations times, for instance:
 
-- [Solver tolerance](https://diffeq.sciml.ai/stable/basics/faq/#What-does-tolerance-mean-and-how-much-error-should-I-expect):
- By default we use absolute and relative tolerances of `1e-6` and `1e-5` respectively.
+- [Solver tolerance](https://docs.sciml.ai/DiffEqDocs/stable/basics/faq/#What-does-tolerance-mean-and-how-much-error-should-I-expect):
+ By default both the absolute and relative tolerance is `1e-7`.
 - [ODE solvers](https://diffeq.sciml.ai/stable/solvers/ode_solve/):
  The `QNDF` method we use is robust to oscillations and massive stiffness, however other solvers should be tried as well.
 - Forcing: Every time new forcing data is injected into the model, it needs to pause.

python/ribasim/ribasim/config.py

@@ -99,9 +99,9 @@ class Solver(ChildModel):
  If a smaller dt than dtmin is needed to meet the set error tolerances, the simulation stops, unless force_dtmin = true
  (Optional, defaults to False)
  abstol : float
- The absolute tolerance for adaptive timestepping (Optional, defaults to 1e-6)
+ The absolute tolerance for adaptive timestepping (Optional, defaults to 1e-7)
  reltol : float
- The relative tolerance for adaptive timestepping (Optional, defaults to 1e-5)
+ The relative tolerance for adaptive timestepping (Optional, defaults to 1e-7)
  maxiters : int
  The total number of linear iterations over the whole simulation. (Defaults to 1e9, only needs to be increased for extremely long simulations)
  sparse : bool

python/ribasim_api/tests/test_bmi.py

@@ -168,7 +168,7 @@ def test_get_value_ptr_allocation(libribasim, user_demand, tmp_path):
  libribasim.update_until(60.0)
 
  # Realized inflow
- actual_inflow = libribasim.get_value_ptr("user_demand.inflow")
+ actual_inflow = libribasim.get_value_ptr("user_demand.cumulative_inflow")
  expected_inflow = np.array([1e-4 * 60.0, 0.0, 0.0])
  assert_array_almost_equal(actual_inflow, expected_inflow)