Skip to content

anypinn.cli.scaffold.seir.ode_csv

SEIR epidemic model — mathematical definition.

BETA_KEY = 'beta' module-attribute

E0 = 0.01 module-attribute

E_KEY = 'E' module-attribute

GAMMA_KEY = 'gamma' module-attribute

I0 = 0.001 module-attribute

I_KEY = 'I' module-attribute

NOISE_STD = 0.0005 module-attribute

S0 = 0.99 module-attribute

SIGMA_KEY = 'sigma' module-attribute

S_KEY = 'S' module-attribute

TRUE_BETA = 0.5 module-attribute

TRUE_GAMMA = 1 / 10 module-attribute

TRUE_SIGMA = 1 / 5.2 module-attribute

T_DAYS = 160 module-attribute

validation: ValidationRegistry = {} module-attribute

SEIR(x: Tensor, y: Tensor, args: ArgsRegistry) -> Tensor

Scaled SEIR ODE system.

Source code in src/anypinn/cli/scaffold/seir/ode_csv.py 
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
def SEIR(x: Tensor, y: Tensor, args: ArgsRegistry) -> Tensor:
    """Scaled SEIR ODE system."""
    S, E, I = y
    b = args[BETA_KEY]
    sigma = args[SIGMA_KEY]
    gamma = args[GAMMA_KEY]

    dS = -b(x) * S * I
    dE = b(x) * S * I - sigma(x) * E
    dI = sigma(x) * E - gamma(x) * I

    dS = dS * T_DAYS
    dE = dE * T_DAYS
    dI = dI * T_DAYS
    return torch.stack([dS, dE, dI])

create_data_module(hp: ODEHyperparameters)

Source code in src/anypinn/cli/scaffold/seir/ode_csv.py 
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
def create_data_module(hp: ODEHyperparameters):
    from anypinn.catalog.seir import SEIRDataModule

    # Unscaled SEIR for data generation
    def SEIR_unscaled(x: Tensor, y: Tensor, args: ArgsRegistry) -> Tensor:
        S, E, I = y
        b = args[BETA_KEY]
        sigma = args[SIGMA_KEY]
        gamma = args[GAMMA_KEY]
        dS = -b(x) * S * I
        dE = b(x) * S * I - sigma(x) * E
        dI = sigma(x) * E - gamma(x) * I
        return torch.stack([dS, dE, dI])

    gen_props = ODEProperties(
        ode=SEIR_unscaled,
        y0=torch.tensor([S0, E0, I0]),
        args={
            BETA_KEY: Argument(TRUE_BETA),
            SIGMA_KEY: Argument(TRUE_SIGMA),
            GAMMA_KEY: Argument(TRUE_GAMMA),
        },
    )

    return SEIRDataModule(
        hp=hp,
        gen_props=gen_props,
        noise_std=NOISE_STD,
        validation=validation,
        callbacks=[DataScaling(y_scale=1.0)],
    )

create_problem(hp: ODEHyperparameters) -> ODEInverseProblem

Source code in src/anypinn/cli/scaffold/seir/ode_csv.py 
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
def create_problem(hp: ODEHyperparameters) -> ODEInverseProblem:
    props = ODEProperties(
        ode=SEIR,
        y0=torch.tensor([S0, E0, I0]),
        args={
            SIGMA_KEY: Argument(TRUE_SIGMA),
            GAMMA_KEY: Argument(TRUE_GAMMA),
        },
    )

    fields = FieldsRegistry(
        {
            S_KEY: Field(config=hp.fields_config),
            E_KEY: Field(config=hp.fields_config),
            I_KEY: Field(config=hp.fields_config),
        }
    )
    params = ParamsRegistry(
        {
            BETA_KEY: Parameter(config=hp.params_config),
        }
    )

    def predict_data(x_data: Tensor, fields: FieldsRegistry, _params: ParamsRegistry) -> Tensor:
        I_pred = fields[I_KEY](x_data)
        return I_pred.unsqueeze(1)

    return ODEInverseProblem(
        props=props,
        hp=hp,
        fields=fields,
        params=params,
        predict_data=predict_data,
    )