如何在Jupyter Notebook中显示PyDrake的SNOPT打印文件？

Great question! Let's break down why your initial code didn't work, then walk through the correct ways to get SNOPT's print output showing up in your Jupyter Notebook.

Your line mp.SetSolverOption(SolverType.kSnopt, "Print file", 1) didn't work because SNOPT's "Print file" option expects either a file path string or a standard file descriptor integer—and 1 doesn't reliably map to Jupyter's stdout environment. Instead, we use descriptor 6, which is SNOPT's standard way to route output to the console (and thus your notebook cell).

Option 1: Send SNOPT Output Directly to the Notebook Cell #

This is the simplest approach—output will appear right below your code cell when you run it:

from pydrake.solvers.mathematicalprogram import MathematicalProgram, SolverType
from pydrake.solvers.snopt import SnoptSolver

# Set up your optimization problem
mp = MathematicalProgram()
# Add your variables, constraints, and cost function here...

# Tell SNOPT to send print output to stdout (file descriptor 6)
mp.SetSolverOption(SolverType.kSnopt, "Print file", 6)

# Solve and view the output
solver = SnoptSolver()
result = solver.Solve(mp)

When you execute this cell, you'll see SNOPT's full print output (including iteration logs, feasibility checks, and convergence details) directly in the notebook.

Option 2: Save Output to a File and Display It #

If you want to save the log for later reference, you can write it to a file and then display the content in Jupyter:

# Configure SNOPT to write to a local log file
mp.SetSolverOption(SolverType.kSnopt, "Print file", "snopt_output.log")

# Run the solver
solver = SnoptSolver()
result = solver.Solve(mp)

# Read and display the log in the notebook
with open("snopt_output.log", "r") as log_file:
    snopt_log = log_file.read()

# Print as plain text
print(snopt_log)

# Or format it nicely with Markdown (optional)
from IPython.display import display, Markdown
display(Markdown(f"```text\n{snopt_log}\n```"))

Quick Tips #

• Timing matters: Make sure you set the solver option before calling Solve()—options set after solving won't apply.
• Adjust verbosity: Pair this with the "Print level" option to control how much detail SNOPT outputs. For example, mp.SetSolverOption(SolverType.kSnopt, "Print level", 10) will give you more detailed logs, while 0 will suppress most output.
• Descriptor 6 is reliable: Avoid using 1 or 2 directly—SNOPT uses 6 as the standard file descriptor for stdout across most environments, which works consistently in Jupyter.

内容的提问来源于stack exchange，提问作者Daniel