stanfordnlp/CoreNLP · error · RuntimeException

total mass must be positive!

Error message

total mass must be positive!

What it means

The Multinomial constructor normalizes the supplied Counter by its total mass; a total mass of zero or negative makes normalization meaningless, so the constructor throws immediately. This guards against building a distribution from an empty or all-zero counter.

Solutions

1. Verify totalCount() > 0 before constructing; if not, fall back to a uniform distribution
2. Add smoothing (e.g. add-k smoothing) so total mass is positive even for unseen events
3. Fix upstream counting code that produced an empty/zero counter

Example fix

// before
Multinomial<String> m = new Multinomial<>(counts); // throws if total is 0
// after
counts.incrementCount(UNKNOWN, 1.0); // smooth
Multinomial<String> m = new Multinomial<>(counts);

if (counter == null || counter.totalCount() <= 0.0) {
  counter = new ClassicCounter<>(); counter.incrementCount(DEFAULT_KEY, 1.0); // uniform fallback
}

boolean isPositiveMass(Counter<?> c) { return c != null && c.totalCount() > 0.0; }

try {
  return new Multinomial<>(counter);
} catch (RuntimeException e) {
  if ("total mass must be positive!".equals(e.getMessage())) {
    return uniformMultinomial(counter == null ? Collections.emptySet() : counter.keySet());
  }
  throw e;
}

When it happens

Trigger: new Multinomial(counter) where counter.totalCount() <= 0.0 — an empty counter, a counter whose entries are all 0, or one with only negative counts.

Common situations: Estimating a multinomial from training data that produced no counts for a context (e.g. unseen n-gram context); forgetting to set counts before constructing; smearing/counting bugs that zero out all entries.

Understand the failure class

Background: "Must be a positive integer", "Invalid value", "Unsupported": the invalid-argument-value error family, when a library rejects the value you pass — this error's family across 35 libraries.

Related errors

• no negative parameters allowed!
• Bad arguments: " + x + " and " + lambda
• Can't parse a zero-length sentence!
• Cannot combine MWT out of only
• Cannot make a Lemmatize with no nodeName

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/d9b1feaeee854531. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/stats/Multinomial.java:25

 * a counter.  It is assumed that the Counter's keySet() contains all of the parameters (i.e., there are not other
 * possible values which are set to 0).  It makes a copy of the Counter, so tha parameters cannot be changes,
 * and it normalizes the values if they are not already normalized.
 *
 * @author Jenny Finkel
 */

public class Multinomial<E> implements ProbabilityDistribution<E> {

  /**
   * 
   */
  private static final long serialVersionUID = -697457414113362926L;
  private Counter<E> parameters;

  public Multinomial(Counter<E> parameters) {
    double totalMass = parameters.totalCount();
    if (totalMass <= 0.0) {
      throw new RuntimeException("total mass must be positive!");
    }

    this.parameters = new ClassicCounter<>();
    for (E object : parameters.keySet()) {
      double oldCount = parameters.getCount(object);
      if (oldCount < 0.0) {
        throw new RuntimeException("no negative parameters allowed!");
      }
      this.parameters.setCount(object, oldCount/totalMass);
    }
  }

  public Counter<E> getParameters() {
    return new ClassicCounter<>(parameters);
  }
  
  public double probabilityOf(E object) {
    if (!parameters.keySet().contains(object)) {

View on GitHub (pinned to 1b7edd19c4)