Matplotlib中ticker formatter.limits配合次坐标轴的小数位数调整问题

To force your secondary axis to display a fixed number of decimal places in scientific notation (while retaining Matplotlib's mathtext formatting), you have two reliable approaches:

Approach 1: Use ScalarFormatter for Significant Figures #

If you want to control the number of significant digits (aligning with Matplotlib's default behavior but enforcing consistency), use ScalarFormatter with mathtext enabled:

1. Import the ticker module from Matplotlib.
2. Create a ScalarFormatter instance, enable mathtext, set the desired precision, and match your formatter.limits configuration.
3. Assign the formatter to the secondary axis.

Modified code snippet:

import numpy as np
from matplotlib import pyplot as plt
from matplotlib import ticker  # Add this import

plt.rcParams['axes.formatter.limits'] = [-4, 6]
plt.rcParams["axes.autolimit_mode"] = "round_numbers" 
plt.rcParams["axes.formatter.use_mathtext"] = True

x = np.linspace(0, 4)
y = np.linspace(0, 2.5)

fig, ax = plt.subplots(dpi=300)
ax.plot(x, y)
secax = ax.twiny()
axlim = ax.get_xlim()
secax_xlim = [axlim[0]*22.76847e-6,
              axlim[1]*22.76847e-6]
secax.set_xlim(secax_xlim)
axticks = ax.get_xticks()
secax_xtics = axticks*22.76847e-6
secax.xaxis.set_ticks(secax_xtics)

# Customize secondary axis formatter
formatter = ticker.ScalarFormatter(useMathText=True)
formatter.set_precision(2)  # Show 2 significant digits
formatter.set_powerlimits((-4, 6))  # Match rcParam limits
secax.xaxis.set_major_formatter(formatter)

plt.show()

This will display ticks like $2.3 \times 10^{-5}$ (for 2 significant digits) or $2.28 \times 10^{-5}$ if you set precision=3.

Approach 2: Custom FuncFormatter for Exact Decimal Places #

If you need exact decimal places in the mantissa (e.g., always 2 decimal places regardless of significant digits), create a custom FuncFormatter to format each tick value into mathtext-compliant scientific notation:

Modified code snippet:

import numpy as np
from matplotlib import pyplot as plt
from matplotlib import ticker  # Add this import

plt.rcParams['axes.formatter.limits'] = [-4, 6]
plt.rcParams["axes.autolimit_mode"] = "round_numbers" 
plt.rcParams["axes.formatter.use_mathtext"] = True

x = np.linspace(0, 4)
y = np.linspace(0, 2.5)

fig, ax = plt.subplots(dpi=300)
ax.plot(x, y)
secax = ax.twiny()
axlim = ax.get_xlim()
secax_xlim = [axlim[0]*22.76847e-6,
              axlim[1]*22.76847e-6]
secax.set_xlim(secax_xlim)
axticks = ax.get_xticks()
secax_xtics = axticks*22.76847e-6
secax.xaxis.set_ticks(secax_xtics)

# Custom formatter for exact 2 decimal places in mantissa
def custom_format(value, tick_number):
    if value == 0:
        return r'$0$'
    # Format to 2 decimal places in scientific notation
    sci_str = "{0:.2e}".format(value)
    mantissa, exponent = sci_str.split('e')
    # Clean up exponent (remove leading zeros and plus sign)
    exponent = exponent.lstrip('+').lstrip('0') or '0'
    # Return mathtext-formatted string
    return r'$%s \times 10^{%s}$' % (mantissa, exponent)

secax.xaxis.set_major_formatter(ticker.FuncFormatter(custom_format))

plt.show()

This will consistently display ticks like $2.28 \times 10^{-5}$ (exactly 2 decimal places in the mantissa) across all your charts, ensuring visual consistency.

Key Notes: #

• Both methods preserve Matplotlib's mathtext rendering (e.g., proper superscript exponents).
• The FuncFormatter approach gives full control over decimal places, while ScalarFormatter adheres to significant digits (closer to Matplotlib's default behavior).
• Adjust the .2e in the FuncFormatter to .1e for 1 decimal place, or .3e for 3, etc.

内容的提问来源于stack exchange，提问作者tetukowski