Configuring Simulations¶
For a general overview regarding configuration, see Configuring Simulation Runs. For reading configuration keys in the model, see the model configuration FAQs.
How can I find the available model parameters?¶
The base configuration documents a lot of parameters directly in the configuration file; see here. For the model configuration, the model documentation usually includes the default configuration; for example, take a look at the ForestFire Model.
What’s with all these YAML !tags? What can I use them for?¶
YAML tags are really cool!
You’ve probably already discovered the !sweep tag, which is used to define a parameter sweep dimension. Another example is !param, used to mark parameters that need to be validated before a simulation run. When reading YAML files, the frontend can attach certain functionality to entries that are labelled with a !tag. Throughout Utopia (and its dependencies, dantro and paramspace), this functionality is used to make it more convenient to define configuration entries.
Which other tags are available?¶
Quite a bunch. The example below demonstrates them:
# --- Simple calculations ---------------------------------------------------
# Allowed symbols: whitespace, numbers, dot, +-/*, **, (), eE
fourtytwo: !expr 6. * 7
seconds_per_day: !expr 60 * 60 * 24
powers: !expr 2**10
parentheses: !expr (2 + 3) * 4 / (5 - 6)
exp_notation: !expr (2.34 / 3.45) * 1e-10
# --- List generation -------------------------------------------------------
# Generate an integer-filled list using: list(range(*args))
list_foo: !listgen [10] # == [0, 1, 2, ..., 9]
list_bar: !listgen [4, 20, 2] # == [4, 6, 8, ..., 18]
# Can also specify more advanced arguments to control the list's content
list_fancy: !listgen # == [0, 10]
from_range: [0, 100, 10]
append: [0, 25, 50, 75, 100] # append these to the list
remove: [0] # remove _all_ occurences of these
unique: true # removes duplicates, here: 50
sort: true # whether to sort the list, default: true
# NOTE The arguments are always applied in the same order:
# from_range, append, remove, unique, sort (default: true)
# Nesting is also possible!
list_nested: !listgen # == [0, 2, 4, 6, 8, 10, 20, 40, 60, 80]
from_range: [0, 11, 2]
append: !listgen [0, 100, 20]
unique: true
# --- Boolean logic with !any and !all --------------------------------------
will_be_true: !any [false, false, true]
will_be_false: !all [true, false, true, false]
# These are useful to combine with YAML anchors
flag1: &flag1 true
flag2: &flag2 false
flag3: &flag3 true
some_parameters:
enabled: !all [*flag1, *flag2, *flag3]
# And they can also be nested:
another_set_of_parameters:
enabled: !any
- !all [*flag1, *flag2]
- *flag3
There are some Python-only tags, which create Python objects that have no equivalent on C++-side. Make sure to not specify them inside your run or model config, but only in your evaluation config.
# NOTE All tags below cannot be used to pass information to the model, as
# they will create Python objects that have no equivalent in C++.
# However, they can be useful for the plotting of model runs.
# Can also specify Python NaN and INFs ...
some_nan: !expr NaN # creates np.nan
some_inf: !expr inf # == np.inf
some_ninf: !expr -inf # == - np.inf
# Create a Python range. !range == range(*args)
one_range: !range [10] # == range(10)
another_range: !range [5, 20, 2] # == range(5, 20, 2)
# Create a Python slice, !slice == slice(*args)
empty_slice: !slice [~] # == slice(None), i.e. [:]
some_slice: !slice [5, ~] # == slice(5, None), i.e. [5:]
Hint
Make sure to check out the paramspace docs for more YAML tags, which e.g. allow evaluating simple boolean operations or format strings.