def test_computable_mvars(self):
# for debugging we draw the sparse_powerset_graph of the actually present computers and mvars
spsg = sparse_powerset_graph(bgc_md2_computers())
f = plt.figure()
ax = f.add_subplot(1, 1, 1)
draw_ComputerSetMultiDiGraph_matplotlib(ax,
spsg,
bgc_md2_mvar_aliases(),
bgc_md2_computer_aliases(),
targetNode=frozenset(
{SmoothModelRun}))
f.savefig("spgs.pdf")
fig = plt.figure()
draw_update_sequence(bgc_md2_computers(), max_it=8, fig=fig)
fig.savefig("c1.pdf")
# https://docs.python.org/3/library/unittest.html#distinguishing-test-iterations-using-subtests
for mn in [
"Potter1993GlobalBiogeochemicalCycles",
"testVectorFree",
"Williams2005GCB", # just uncomment the one model you are working on and comment the others
]:
with self.subTest(mn=mn):
mvs = MVarSet.from_model_name(mn)
mvars = mvs.computable_mvar_types()
list_str = "\n".join(
["<li> " + str(var.__name__) + " </li>" for var in mvars])
print(list_str)
for var in mvars:
print("########################################")
print(str(var.__name__))
print(mvs._get_single_mvar_value(var))
def test_from_model_name(self):
# alternative constructor
mvs= MVarSet.from_model_name("testVectorFree")
res = frozenset(
[
InFluxesBySymbol,
OutFluxesBySymbol,
InternalFluxesBySymbol,
TimeSymbol,
StateVariableTuple,
]
)
self.assertSetEqual(mvs.provided_mvar_types, res)
def setUp(self):
I_vl, I_vw = symbols("I_vl I_vw")
vl, vw = symbols("vl vw")
k_vl, k_vw = symbols("k_vl k_vw")
self.provided_values=frozenset(
[
InFluxesBySymbol({vl: I_vl, vw: I_vw}),
OutFluxesBySymbol({vl: k_vl * vl, vw: k_vw * vw}),
InternalFluxesBySymbol({(vl, vw): k_vl * vl, (vw, vl): k_vw * vw}),
TimeSymbol("t"),
StateVariableTuple((vl, vw))
]
)
self.mvs=MVarSet(self.provided_values)
def setUp(self):
self.mn = "Williams2005GCB"
self.mvs = MVarSet.from_model_name(self.mn)
self.ref_provided_mvars = frozenset([
CompartmentalMatrix,
TimeSymbol,
StateVariableTuple,
VegetationCarbonInputPartitioningTuple,
VegetationCarbonInputScalar,
InputTuple,
NumericStartValueDict,
NumericSimulationTimes,
NumericParameterization,
QuantityStartValueDict,
QuantitySimulationTimes,
QuantityParameterization,
BibInfo,
# QuantityModelRun,
# QuantityParameterizedModel
])
def setUp(self):
self.mn = "Castanho2013Biogeosciences"
self.mvs = MVarSet.from_model_name(self.mn)
self.ref_provided_mvars = frozenset([
CompartmentalMatrix,
TimeSymbol,
StateVariableTuple,
VegetationCarbonInputPartitioningTuple,
VegetationCarbonInputScalar,
VegetationCarbonStateVariableTuple,
InputTuple,
# # NumericStartValueDict,
# # NumericSimulationTimes,
# NumericParameterization,
# # QuantityStartValueDict,
# # QuantitySimulationTimes,
# # QuantityParameterization,
# BibInfo,
# # QuantityModelRun,
# # QuantityParameterizedModel
])
def setUp(self):
# self.mn = ""
self.mn = "ElMasri2013AgricForMeteorol"
# self.mn = "Pavlick2013Biogeosciences"
# self.mn = "Hilbert1991AnnBot"
# self.mn = "Foley1996GBC"
# self.mn = "Gu2010EcologicalComplexity"
# self.mn = "Arora2005GCB-1"
# self.mn = "King1993TreePhysiol"
# self.mn = "Murty2000EcolModell"
# self.mn = "Wang2010Biogeosciences"
# self.mn = "DeAngelis2012TheorEcol"
# self.mn = "Comins1993EA"
# self.mn = "Running1988EcolModel"
# self.mn = "Luo2012TE"
self.mvs = MVarSet.from_model_name(self.mn)
self.ref_provided_mvars = frozenset(
[
CompartmentalMatrix,
TimeSymbol,
StateVariableTuple,
VegetationCarbonInputPartitioningTuple,
VegetationCarbonInputScalar,
VegetationCarbonStateVariableTuple,
InputTuple,
# # NumericStartValueDict,
# # NumericSimulationTimes,
# NumericParameterization,
# # QuantityStartValueDict,
# # QuantitySimulationTimes,
# # QuantityParameterization,
# BibInfo,
# # QuantityModelRun,
# # QuantityParameterizedModel
]
)
# "ElMasri2013AgricForMeteorol", #Paper shows results for soil carbon, and fig. 1 has litter and C pools, but the equations on table A2 (although very detailed) don't include them. There are also equations for Phenology, but they were not included in the source.py
# "Scheiter2009GlobalChangeBiol", #No soil compartments
# "Turgman2018EcologyLetters", #No soil compartments
# "Haverd2016Biogeosciences", #No soil compartments
# "Foley1996GBC", #No equations for litter and soil, but the figure has those compartments. See Markus’ Ibis.yaml?
# "Gu2010EcologicalComplexity", #No equations for litter and soil, but the model description (CEVSA) mentions them
# "King1993TreePhysiol", #No soil compartments
# "DeAngelis2012TheorEcol", #No soil compartments (model based on G’Day, but removed litter and soil compartments)
# "Potter1993GlobalBiogeochemicalCycles", #No equations for litter and soil, but the model description mentions them
# "testVectorFree",
# "Williams2005GCB",
"CARDAMOM",
]
######################################################################
####### MODELS NOT TRANSLATED FROM .YAML TO SOURCE.PY
# Sitch2003GlobChangBiol: Not included because allocation (see pg 8) has no ODEs; biomass increment is allocated to the tissue pools while satisfying the functional balance difference equations...
# VanDerWerf1993PlantSoil: this model has no compartment for wood. It is used to simulate effect of nitrogen on growth of a grass (Dactylis glomerata L.).
# ICBM: same as Andren1997EA but less parameter sets
# Schimel2003SoilBiologyandBiochemistry.yaml, Schimel2003SoilBiologyandBiochemistry_rMM.yaml, Schimel2003SoilBiologyandBiochemistry_rMM_improved.yaml: Original model has no outputs, corrected by Holger
# Ibis: includes soil, no metadata -added by Markus. Vero’s version: Foley
######################################################################
for mn in model_names:
mvs = MVarSet.from_model_name(mn)
mvars = mvs.computable_mvar_types()
list_str = "\n".join(["<li> " + str(var.__name__) + " </li>" for var in mvars])
print(list_str)
for var in mvars:
print("########################################")
print(str(var.__name__))
print(mvs._get_single_mvar_value(var))
mvs = MVarSet({
BibInfo(# Bibliographical Information
name="DALEC",
longName="Data Assimilation Linked Ecosystem model",
version="1",
entryAuthor="Verónika Ceballos-Núñez",
entryAuthorOrcid="0000-0002-0046-1160",
entryCreationDate="16/9/2016",
doi="10.1111/j.1365-2486.2004.00891.x",
further_references=BibInfo(doi="10.1016/j.agrformet.2009.05.002"),
# Also from PDF in Reflex experiment
sym_dict=sym_dict
),
#
# the following variables constitute the compartmental system:
#
A, # the overall compartmental matrix
Input, # the overall imput
t, # time for the complete system
x, # state vector of the the complete system
#
# the following variables constitute a numerical model run :
#
np1,
nsv1,
ntimes,
# Again the model run could be created explicitly (in several ways)
#
# the following variables constitute a quantiy model run :
#
qp1,
qsv1,
qtimes,
# Again the model run could be created explicitly (in several ways)
#
# the following variables give a more detailed view with respect to
# carbon and vegetation variables
# This imformation can be used to extract the part
# of a model that is concerned with carbon and vegetation
# in the case of this model all of the state variables
# are vegetation and carbon related but for an ecosystem model
# for different elements there could be different subsystems
# e.g. consisting of Nitrogen Soil state variables
#
# vegetation carbon
VegetationCarbonInputScalar(u),
# vegetation carbon partitioning.
VegetationCarbonInputPartitioningTuple(b),
VegetationCarbonStateVariableTuple((C_f, C_lab, C_w, C_r))
})
vl, vw = symbols("vl vw")
k_vl, k_vw = symbols("k_vl k_vw")
# the keys of the internal flux dictionary are tuples (source_pool,target_pool)
# srm:SmoothReservoirModel
# srm=SmoothReservoirModel.from_state_variable_indexed_fluxes(
# in_fluxes
# ,out_fluxes
# ,internal_fluxes
# )
#specialVars = {
mvs = MVarSet({
InFluxesBySymbol({
vl: I_vl,
vw: I_vw
}),
OutFluxesBySymbol({
vl: k_vl * vl,
vw: k_vw * vw
}),
InternalFluxesBySymbol({
(vl, vw): k_vl * vl,
(vw, vl): k_vw * vw
}),
TimeSymbol("t"),
StateVariableTuple((vl, vw))
# ,'SmoothReservoirModel':srm
})
def test_numeric_input_tuple(self):
# setup the paths to the testdata
cable_out_path = Path(
'/home/data/cable-data/example_runs/parallel_1901_2004_with_spinup/output/new4'
)
# cable_data_set = cH.cable_ds(cable_out_path)
time_slice = slice(0, None)
landpoint_slice = slice(1590, 1637) # cheated
# landpoint_slice = slice(None,None)
# time_slice=slice(None,None,None)
zarr_cache_path = cP.slice_dir_path(
cable_out_path,
sub_dir_trunk="zarr_mm11",
time_slice=time_slice,
landpoint_slice=landpoint_slice,
)
if "cable_data_set" not in dir():
cable_data_set = cH.cable_ds(cable_out_path)
args = {
"cable_data_set": cable_data_set,
"zarr_cache_path": zarr_cache_path,
"landpoint_slice": landpoint_slice,
"time_slice": time_slice,
#'batch_size': 128,
"batch_size": 12,
#'rm': True
}
x_org_iveg = cH.cacheWrapper(cC.x_org_iveg, **args)
time = cH.cacheWrapper(cC.time, **args)
patches, landpoints = cC.all_pools_vary_cond_nz(**args)
pcs = patches.compute()
lpcs = landpoints.compute()
pcs, lpcs
p = Path('plots')
p.mkdir(exist_ok=True)
ind = 0 # first pair that has a nonconstant solution for
lp = lpcs[ind]
patch = lpcs[ind]
# get the symbolic represantation from the database
mvs = self.mvs
# define some stuff to extend it with
def default(t):
return 1
leaf = Symbol('leaf')
fine_root = Symbol('fine_root')
Npp = Function("Npp")
bvec_leaf = Function("bvec_leaf")
bvec_fine_root = Function("bvec_fine_root")
xk_leaf_cold = Function("xk_leaf_cold")
xk_leaf_dry = Function("xk_leaf_dry")
kleaf = Function("kleaf")
kfroot = Function("kfroot")
# bvec_wood = Function("bvec_wood")
np1 = NumericParameterization(
par_dict={},
func_dict=frozendict({
Npp: default,
bvec_fine_root: default,
bvec_leaf: default,
xk_leaf_cold: default,
kleaf: default,
kfroot: default,
xk_leaf_dry: default,
}),
)
nsv1 = NumericStartValueDict({leaf: 0.3, fine_root: 3.96})
ntimes1 = NumericSimulationTimes(np.linspace(0, 1, 11))
# extend the symbolice version with the new stuff
pvs = mvs.provided_mvar_values
#from IPython import embed; embed()
pvs1 = pvs.union(frozenset({np1, nsv1, ntimes1}))
mvs1 = MVarSet(pvs1)
x = mvs1.get_StateVariableTuple()
Input = mvs1.get_InputTuple()
#B = mvs1.get_CompartmentalMatrix()
sym_times = mvs1.get_NumericSimulationTimes()
sol_smooth = mvs1.get_NumericSolutionArray()
comp_slice = slice(0, 100)
n = sol_smooth.shape[1]
fig = plt.figure()
for pool in range(n):
ax = fig.add_subplot(n + 1, 1, 2 + pool)
title = "\$" + latex(x[pool]) + "\$"
#ax.plot(
# sym_times[comp_slice],
# sol_smooth[comp_slice, pool],
# color='r'
#)
ax.plot(time[comp_slice],
x_org_iveg[comp_slice, pool, patch, lp],
color='b')
fontsize = 10
ax.set_title(title, fontsize=fontsize)
fig.savefig('solution.pdf')
def setUp(self):
self.mn = "cable_all"
self.mvs = MVarSet.from_model_name(self.mn)
# text_representation:
# extension: .py
# format_name: light
# format_version: '1.5'
# jupytext_version: 1.10.2
# kernelspec:
# display_name: Python 3
# language: python
# name: python3
# ---
from IPython.display import HTML
from bgc_md2.resolve.mvars import CompartmentalMatrix, StateVariableTuple, VegetationCarbonInputPartitioningTuple,VegetationCarbonInputTuple
from bgc_md2.resolve.MVarSet import MVarSet
display(HTML("<style>.container { width:100% !important; }</style>"))
mvs_mm = MVarSet.from_model_name('TECOmm')
import igraph
g=igraph.Graph(directed=True)
in_fluxes, internal_fluxes, out_fluxes = mvs_mm.get_InFluxesBySymbol(),mvs_mm.get_InternalFluxesBySymbol(),mvs_mm.get_OutFluxesBySymbol()
in_flux_targets, out_flux_sources = [[str(k) for k in d.keys()] for d in (in_fluxes, out_fluxes)]
internal_connections = [(str(s),str(t)) for s,t in internal_fluxes.keys()]
# +
G=igraph.Graph(directed=True)
for i,s in enumerate(mvs_mm.get_StateVariableTuple()):
G.add_vertex(str(s))# the name attribute is set automatically
{
VegetationCarbonInputPartitioningTuple,
NumericSolutionArray
}
),
# explicit_exclude_models=frozenset({'CARDAMOM'})
)
li
# From these we chose two models to investigate more thoroughly.
# We load the first one with a helper function and the second by using standard python tools.
#
#
mvs_Luo = MVarSet.from_model_name('Luo2012TE')
# or using a moduel import
from bgc_md2.models.TECO.source import mvs as mvs_TECO
# # ComputabilityGraphs
# Now that we actually have two records (MVarSets) we can exlore what we can comptute from them.
# Just add a dot "." behind the variable in the next cell and press the tab key!
#
mvs_TECO
# The options provided by the python interpreter are actually the result of a graph computation.
# To see all computable mvars of the TECO MVarSet execute the next cell!
#
mvs_TECO.computable_mvar_names