如何为scipy.optimize.curve_fit的拟合函数传递参数？

Got it, let's break this down. You want to use scipy.optimize.curve_fit where a is the parameter we fit for, and b is a fixed value that changes depending on which dataset we're working with. Here are two straightforward, clean ways to make this work:

Method 1: Use a Closure (Nested Function) #

A closure lets you define a "wrapper" function that takes your fixed b value, then returns the actual fitting function that uses this b. This keeps each dataset's b value isolated and avoids confusion.

import numpy as np
from scipy.optimize import curve_fit

def create_fit_func(b):
    # This inner function uses the `b` passed to the outer wrapper
    def fitfun(x, a):
        return np.exp(a * (x - b))
    return fitfun

# --- Example usage with two datasets ---
# First dataset: b = 10
x_data1 = np.linspace(0, 20, 100)
y_data1 = np.exp(0.1 * (x_data1 - 10)) + np.random.normal(0, 0.05, size=len(x_data1))

# Create a fit function bound to b=10
fit_func_b10 = create_fit_func(10)
# Fit: curve_fit will only optimize `a` here
popt1, pcov1 = curve_fit(fit_func_b10, x_data1, y_data1)
print(f"Dataset 1 (b=10): Fitted a = {popt1[0]:.4f}")

# Second dataset: b = 20
x_data2 = np.linspace(0, 30, 100)
y_data2 = np.exp(0.08 * (x_data2 - 20)) + np.random.normal(0, 0.05, size=len(x_data2))

fit_func_b20 = create_fit_func(20)
popt2, pcov2 = curve_fit(fit_func_b20, x_data2, y_data2)
print(f"Dataset 2 (b=20): Fitted a = {popt2[0]:.4f}")

Method 2: Use functools.partial to Bind Fixed Parameters #

If you prefer a more concise approach, functools.partial lets you "freeze" the b parameter in a base fitting function, leaving only a as the variable to optimize.

import numpy as np
from scipy.optimize import curve_fit
from functools import partial

# Define a base function that includes both a and b as parameters
def base_fitfun(x, a, b):
    return np.exp(a * (x - b))

# --- Example usage ---
# For dataset 1, fix b=10 (use keyword argument to avoid ordering issues)
fit_func_b10 = partial(base_fitfun, b=10)
popt1, pcov1 = curve_fit(fit_func_b10, x_data1, y_data1)

# For dataset 2, fix b=20
fit_func_b20 = partial(base_fitfun, b=20)
popt2, pcov2 = curve_fit(fit_func_b20, x_data2, y_data2)

A Quick Note to Avoid Mistakes #

Steer clear of using global variables for b—this can lead to bugs when switching between datasets, since the global value might get overwritten accidentally. Both methods above keep each dataset's b value contained and explicit.

内容的提问来源于stack exchange，提问作者Nisfa