stanfordnlp/CoreNLP · error · java.lang.IllegalArgumentException

Invalid line search option for QNMinimizer.

Error message

Invalid line search option for QNMinimizer.

What it means

The switch over the configured line search type (QNMinimizer.eLineSearchType) hit its default branch because the enum value is not one of the supported options (e.g. BACKTRACKING, MINPACK). This indicates a corrupted or unrecognized line search setting.

Solutions

1. Set line search to a supported value: minimizer.setMinimizeType(QNMinimizer.eLineSearchType.BACKTRACKING) or .MINPACK.
2. Upgrade/downgrade so all modules use the same CoreNLP version and enum.
3. Inspect the current value with reflection/logging before calling minimize().

Example fix

// before
minimizer.setMinimizeType(unrecognizedType); // default branch throws
// after
minimizer.setMinimizeType(QNMinimizer.eLineSearchType.MINPACK);

QNMinimizer.eLineSearchType t = minimizer.lineSearchType; // via getter if available
if (t != QNMinimizer.eLineSearchType.BACKTRACKING && t != QNMinimizer.eLineSearchType.MINPACK)
  throw new IllegalStateException("Unsupported line search: " + t);

boolean supported = EnumSet.of(QNMinimizer.eLineSearchType.BACKTRACKING, QNMinimizer.eLineSearchType.MINPACK).contains(type);

try {
  minimizer.minimize(f, tol, init);
} catch (IllegalArgumentException e) {
  minimizer.setMinimizeType(QNMinimizer.eLineSearchType.MINPACK);
  minimizer.minimize(f, tol, init);
}

When it happens

Trigger: Setting the line search option via setMinimizeType()/reflection/config deserialization to an eLineSearchType value not handled by the switch (only BACKTRACKING and MINPACK cases are covered) and then calling minimize().

Common situations: Deserializing a serialized QNMinimizer whose enum constant comes from a different compiled version of the library; programmatic enum construction; version skew between code that sets the type and the minimizer code.

Understand the failure class

Background: Invalid enum value errors: "Unknown type", "Invalid scope", "must be one of" — when a string is not on the library's allowed list — this error's family across 23 libraries.

Related errors

• Not a valid NewlineIsSentenceBreak name
• Unknown format
• Unknown timeAnnotator
• Unknown LogPriorType:
• is not a legal LogPrior.

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/c6a62e74ddd17a02. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/optimization/QNMinimizer.java:1025

        double[] newPoint; // initialized in if/else/switch below
        if (useOWLQN) {
          // only linear search is allowed for OWL-QN
          newPoint = lineSearchBacktrackOWL(dFunction, dir, x, newX, grad, value, sb);
          sb.append('B');
        } else {
          // switch between line search options.
          switch (lsOpt) {
          case BACKTRACK:
            newPoint = lineSearchBacktrack(dFunction, dir, x, newX, grad, value, sb);
            sb.append('B');
            break;
          case MINPACK:
            newPoint = lineSearchMinPack(dFunction, dir, x, newX, grad, value,
                functionTolerance, sb);
            sb.append('M');
            break;
          default:
            throw new IllegalArgumentException("Invalid line search option for QNMinimizer.");
          }
        }

        newValue = newPoint[f];
        sb.append(' ').append(nf.format(newPoint[a])).append("] ");

        // This shouldn't actually evaluate anything since that should have been
        // done in the lineSearch.
        System.arraycopy(dFunction.derivativeAt(newX), 0, newGrad, 0, newGrad.length);

        // This is where all the s, y updates are applied.
        qn.update(newX, x, newGrad, rawGrad, newPoint[a]); // step (4) in Galen & Gao 2007

        if (useOWLQN) {
          System.arraycopy(newGrad, 0, rawGrad, 0, newGrad.length);
          // pseudo gradient
          newGrad = pseudoGradientOWL(newX, newGrad, dFunction);
        }

View on GitHub (pinned to 1b7edd19c4)