How To Debug

This module is based on the resources above. I've included the content that is most relevant to USACO.

Pre-Submit

• Your code should be readable (to yourself at the very least).

  • Following style tips from the Adding Solutions module may help with this.

Wrong Answer (or Runtime Error)

• Is your output format correct?

  • Did you remove debug output before submitting?

• Do you handle all corner cases (such as $N=1$) / special cases?

• For problems with multiple independent test cases (such as this one), are you clearing all data structures between test cases?

  • Keep in mind that your solution might only behave incorrectly when a test case is followed by a smaller test case.

• Have you understood the problem correctly? Read the full problem statement again.

• Read your code again.

  • Confusing $N$ and $M$, $i$ and $j$, etc.?

• Shadowed or unused or uninitialized variables?

  • In C++, compiling with warning options (-Wall -Wshadow) should detect these.

• Any undefined behavior? It can result in different outputs locally vs online (ex. maybe you are passing the sample case locally but not when you submit to the USACO judge). Try running your code in multiple places (ex. USACO Guide IDE, Codeforces Custom Test) and see if you always get the same result. Common examples of undefined behavior include:

  • (C++) Uninitialized variables

  • (C++) Not returning anything from non-void functions

  • (C++) Array out of bounds

    • Considering using ::at as mentioned here.

  • (C++ / Java) Signed integer overflow

    • USACO problems usually contain a note of the following form if the output format requires 64-bit rather than 32-bit integers, but it's easy to miss:

    Note that the large size of integers involved in this problem may require the use of 64-bit integer data types (e.g., a long long in C/C++)."

  In C++, compiling with additional options can help catch these.

• Add assertions and resubmit.

• Floating point numbers

  • Any NaNs (ex. taking the square root of a negative number)?
  • Try using a type with more precision (ex. long double instead of double in C++).
  • Are you printing the output to the correct amount of precision?

• Are you sure your algorithm works?

  • Go through the algorithm for a simple case / write some testcases to run your algorithm on.
  • Write a test case generator and compare the outputs of your solution against that of a (simpler) slow solution, or a model solution if available.

    • See stress testing for more information.

Runtime Error

• Any undefined behavior? (see above)
• Any assertions that might fail?
• Any possible division by 0? (mod 0 for example)
• Any possible infinite recursion?
• Invalidated pointers or iterators?
• Are you using too much memory?

Time Limit Exceeded

• Do you have any possible infinite loops?
• What is the complexity of your algorithm?
• Did you remove debug output before submitting (ex. are you printing a lot of information to stderr)?
• Unnecessary copying of data? C++ - Consider passing variables by reference.
• C++ - Try substituting arrays in place of vectors.

Last Resort

• Rewrite your solution from the start.

  • Be sure to save your original solution. It's always possible that you might introduce more bugs.

Before Posting on the USACO Guide Forum

• If you have found a small test case on which your program fails and you know why the expected output is correct, you should be able to figure out why your program is incorrect on your own.

  • Add print statements to your code and compare their outputs to what you get when you simulate your program by hand.
  • Check for undefined behavior as described above.
• If you haven't found a small test case on which your solution fails,

  • Try downloading the official test data and seeing if your solution fails on any small test cases.
  • If that doesn't work, then try generating a small test case on which your solution fails as described above.