pingu_98/cara
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge remote-tracking branch 'origin/feature/mc' into feature/mc

Browse source
This commit is contained in:
Andrejh 2021-03-22 17:42:52 +01:00
commit 669898be3e
1 changed files with 47 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

48
cara/montecarlo.py
View file

@ -105,6 +105,11 @@ def lognormal(csi: float, lamb: float, samples: int) -> np.ndarray:
@dataclass(frozen=True)
class MCVirus:
"""
halflife and decay_constant are used indentically as before
qID replaces coefficient_of_infectivity to be more in-line with the paper (qID = 1 / COI)
no pre-defined virus types
"""
#: Biological decay (inactivation of the virus in air)
halflife: float
@ -119,6 +124,14 @@ class MCVirus:
@dataclass(frozen=True)
class MCPopulation:
"""
The mask-attribute of original CARA has been reduced to a simple bool indicating whether or not masks are being worn
expiratory_activity and breathing_category are integer values corresponding to pre-defined values
Unlike original CARA, expiratory activity is an attribute of a general population, not just an infected one. This
modification turned out to be unnecessary, so the expiratory activity can be places back into MCInfectedPopulation.
The exposed population in the MCExposureModel is currently an instance of the original Population class, as the new
MC-specific features were never needed for the exposed subject.
"""
"""
Represents a group of people all with exactly the same behaviour and
situation.
@ -147,6 +160,27 @@ class MCPopulation:
@dataclass(frozen=True)
class MCInfectedPopulation(MCPopulation):
"""
The new samples-attribute indicates the number of samples to use for the monte-carlo simulation. Having this as an
attribute of the infected population is quite messy, and should probably be changed.
The expiratory_activity_weights are added as a way to simulate scenarios of combined expiratory activities. Should
be placed in the same class as the expiratory_activity attribute.
viral_load can optionally be set to a single float in order to over-ride the use of a distribution. This was mainly
used for testing.
_generate_viral_loads uses the data defined at the top of this file to generate a kernel density estimation
distribution in order to sample randomly from the distribution suggested by the data. This is messy and should be
changed.
_generate_breathing_rates similarly samples an array from a lognormal distribution specified by parameters at the
top of the file.
The functions for generating these samples are cached in order to avoid having to re-sample from
the distributions when we wanted to plot the histograms of the distributions after running the model.
The function _concentration_distribution_without_mask returns a lambda-function which determines the concentration
of particles of a given diameter being expirated by the infected subject, which depends on their expiratory
activity. In the case that masks are being worn, a pre-defined function for mask-leakage vs diameter is applied in
order to simulate the filtration of particles as a function of their diameter. This function is defined at the top
of this file in mask_leakage_out. If it is desirable to model different kinds of masks, this function should
probably a member function of some Mask-class.
"""
#: The virus with which the population is infected.
virus: MCVirus
@ -259,6 +293,9 @@ class MCInfectedPopulation(MCPopulation):
@dataclass(frozen=True)
class MCConcentrationModel:
"""
No significant modifications compared to models.py
"""
room: models.Room
ventilation: models._VentilationBase
infected: MCInfectedPopulation
@ -320,7 +357,8 @@ class MCConcentrationModel:
class BuonannoSpecificInfectedPopulation:
"""
A class representing a specific case described in a paper by Buonanno et al., previously used to compare our
results with the results of the paper.
results with the results of the paper. Not used anywhere in the paper or elsewhere as far as I am aware, and can
most likely be discarded.
"""
#: The virus with which the population is infected.
virus: MCVirus
@ -343,6 +381,14 @@ class BuonannoSpecificInfectedPopulation:
@dataclass(frozen=True)
class MCExposureModel:
"""
The quanta_exposure function has been modified in order to process arrays of a different shape compared to original
CARA, but essentially does the exact same thing.
The infection_probability function now takes into account a deposition factor (f_dep) which indicates the portion
of inhaled particles which are not subsequently exhaled.
The reproduction_number function was never needed and thus not implemented for the MC approach, but doing so would
probably be trivial.
"""
#: The virus concentration model which this exposure model should consider.
concentration_model: MCConcentrationModel