prestodb/presto · error · PrestoException

INVALID_FUNCTION_ARGUMENT

INVALID_FUNCTION_ARGUMENT

Error message

In differential_entropy UDF, max samples must be positive: %s

What it means

The weighted reservoir sampling strategy for differential_entropy requires a strictly positive max_samples because the reservoir must be able to hold at least one sample. The constructor rejects zero or negative capacities up front with INVALID_FUNCTION_ARGUMENT rather than failing later during aggregation.

Source

Thrown at presto-main-base/src/main/java/com/facebook/presto/operator/aggregation/differentialentropy/WeightedReservoirSampleStateStrategy.java:33

import com.facebook.presto.spi.PrestoException;
import io.airlift.slice.SliceInput;
import io.airlift.slice.SliceOutput;

import static com.facebook.presto.operator.aggregation.differentialentropy.EntropyCalculations.calculateFromSamplesUsingVasicek;
import static com.facebook.presto.spi.StandardErrorCode.INVALID_FUNCTION_ARGUMENT;
import static com.google.common.base.Verify.verify;
import static java.lang.String.format;

public class WeightedReservoirSampleStateStrategy
        implements DifferentialEntropyStateStrategy
{
    private final WeightedDoubleReservoirSample reservoir;

    public WeightedReservoirSampleStateStrategy(long maxSamples)
    {
        if (maxSamples <= 0) {
            throw new PrestoException(
                    INVALID_FUNCTION_ARGUMENT,
                    format("In differential_entropy UDF, max samples must be positive: %s", maxSamples));
        }
        if (maxSamples >= WeightedDoubleReservoirSample.MAX_SAMPLES_LIMIT) {
            throw new PrestoException(
                    INVALID_FUNCTION_ARGUMENT,
                    format("In differential_entropy UDF, max samples  must be capped: max_samples=%s, cap=%s", maxSamples, WeightedDoubleReservoirSample.MAX_SAMPLES_LIMIT));
        }

        reservoir = new WeightedDoubleReservoirSample((int) maxSamples);
    }

    private WeightedReservoirSampleStateStrategy(WeightedReservoirSampleStateStrategy other)
    {
        reservoir = other.reservoir.clone();
    }

    private WeightedReservoirSampleStateStrategy(WeightedDoubleReservoirSample reservoir)

View on GitHub (pinned to 55bb57d202)

Solutions

1. Pass a max_samples value of at least 1 (typically a few thousand for meaningful entropy estimates).
2. Fix the expression producing max_samples so it cannot evaluate to 0 or a negative number.
3. Add a CHECK/CASE guard in the query or coerce non-positive values to a sane default before aggregation.

Example fix

// before
SELECT differential_entropy(x, w, s, 0);
// after
SELECT differential_entropy(x, w, s, 1000);

-- reject non-positive max_samples before the call
SELECT * FROM cfg WHERE max_samples IS NULL OR max_samples <= 0; -- must be empty

When it happens

Trigger: Constructing WeightedReservoirSampleStateStrategy with maxSamples <= 0, i.e. calling differential_entropy(...) with a max_samples argument of 0 or a negative number.

Common situations: Passing a computed max_samples expression that evaluates to 0 (e.g. an empty/zero-valued column or bad default); typos such as -1 used as a sentinel.

Related errors

• DRUID_PUSHDOWN_UNSUPPORTED_EXPRESSION
• DRUID_PUSHDOWN_UNSUPPORTED_EXPRESSION
• Timeout is negative
• GENERIC_INSUFFICIENT_RESOURCES
• INVALID_FUNCTION_ARGUMENT

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/247a0fbf939b9faa. Report an issue: GitHub.