stanfordnlp/CoreNLP · error · IllegalArgumentException

n > size of collection

Error message

n > size of collection: ${n}, ${c.size()}

What it means

CollectionUtils.sampleWithoutReplacement throws IllegalArgumentException when n exceeds the collection size, because you cannot draw more distinct elements (without replacement) than the collection contains. This is a hard precondition of the method's contract.

Solutions

1. Clamp n to the collection size before calling: Math.min(n, c.size()).
2. Guard with an explicit check and choose a fallback (return all elements, or use sampleWithReplacement which allows repeats).
3. Validate that the source collection has at least n elements before sampling; skip or rescale otherwise.
4. If duplicates are acceptable, switch to sampleWithReplacement(c, n, r).

Example fix

// before
Collection<E> s = CollectionUtils.sampleWithoutReplacement(c, 100, r);
// after
Collection<E> s = CollectionUtils.sampleWithoutReplacement(c, Math.min(100, c.size()), r);

if (n > c.size()) throw new IllegalArgumentException("cannot sample " + n + " without replacement from " + c.size());

boolean canSampleWithoutReplacement(int n, Collection<?> c) { return n >= 0 && n <= c.size(); }

try { return CollectionUtils.sampleWithoutReplacement(c, n, r); } catch (IllegalArgumentException e) { return new ArrayList<>(c); }

When it happens

Trigger: Calling sampleWithoutReplacement(c, n, r) with n > c.size(), e.g. sampling a fixed count from a possibly smaller collection or an empty collection with n > 0.

Common situations: Hardcoded sample sizes applied to small datasets; collections filtered down before sampling without adjusting n; off-by-one (n == c.size() + 1).

Related errors

• n < 0
• Bad arguments: " + x + " and " + lambda
• BiLexPCFGParser doesn't support k sampled parses
• Call next() before calling remove()!
• conditionalLogProbGivenFirst requires of one less than…

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/2f1e9e05d1a92f0f. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/util/CollectionUtils.java:369

   */
  public static <E> Collection<E> sampleWithoutReplacement(Collection<E> c, int n) {
    return sampleWithoutReplacement(c, n, new Random());
  }

  /**
   * Samples without replacement from a collection, using your own
   * {@link Random} number generator.
   *
   * @param c The collection to be sampled from
   * @param n The number of samples to take
   * @param r The random number generator
   * @return a new collection with the sample
   */
  public static <E> Collection<E> sampleWithoutReplacement(Collection<E> c, int n, Random r) {
    if (n < 0)
      throw new IllegalArgumentException("n < 0: " + n);
    if (n > c.size())
      throw new IllegalArgumentException("n > size of collection: " + n + ", " + c.size());
    List<E> copy = new ArrayList<>(c.size());
    copy.addAll(c);
    Collection<E> result = new ArrayList<>(n);
    for (int k = 0; k < n; k++) {
      double d = r.nextDouble();
      int x = (int) (d * copy.size());
      result.add(copy.remove(x));
    }
    return result;
  }

  public static <E> E sample(List<E> l, Random r) {
    int i = r.nextInt(l.size());
    return l.get(i);
  }

  /**
   * Samples with replacement from a collection.

View on GitHub (pinned to 1b7edd19c4)