Fitting Discontinuous Functions

[janzenchoi]

Hello,

I am using TemplateExpressionSpec to fit an expression (i.e., z0) that is undefined/complex after a certain input value (e.g., x0 = L). The template returns the relative error, which the loss function interprets as the loss (as per your previous suggestions). However, in fitting the expression (z0), the loss function may try to evaluate the relative error where the fitted expression is undefined (i.e., x0 > L).

Is there any way to configure the library's methods to have the loss function only conduct calculate the loss for defined / real values (i.e., x0 < L)? The following is a snippet of my code to fit the discontinuous expression.

expression_spec = TemplateExpressionSpec(
    expressions    = ["f0", "fA", "fn", "fM", "fphi", "fchi"],
    variable_names = ["x0", "x1", "x2", "y0"],
    combine        = """
        A = 10^fA(x2); n = abs(fn(x2)); M = 10^fM(x2); phi = abs(fphi(x2)); chi = abs(fchi(x2));
        z0 = A*x1^n * ((1-(phi+1)*M*x1^chi*x0)^((phi+1-n)/(phi+1))-1) / (M*x1^chi*(n-phi-1)) + f0(x0,x1,x2);
        ((y0-z0)/y0)^2
    """
)

regressor = PySRRegressor(
    expression_spec  = expression_spec,
    populations      = 32,
    population_size  = 32,
    maxsize          = 16,
    niterations      = 128,
    precision        = 64,
    verbosity        = 0,
    binary_operators = ["+", "*", "/"],
    unary_operators  = ["log"],
    elementwise_loss = "take_first(p, t, w) = p",
)

Thanks again for the help.