[AutoSparkUT] log10 - Function has different output on GPU

[GaryShen2008]

Describe the bug

The log10 function test fails on GPU during expression evaluation.

---

Steps/Code to reproduce bug

Spark-shell Reproduction Code

Environment Setup:

export SPARK_HOME=/path/to/spark-3.3.0
export RAPIDS_JAR=rapids-4-spark_2.12-25.12.0-SNAPSHOT-cuda12.jar

Reproduction Script: test-failures/reproduce-math-expressions-log10.scala

// Test: log10 function with positive and negative values

// Positive values
val df1 = spark.range(1).selectExpr(
  "log10(0.1) as test1",
  "log10(1.0) as test2",
  "log10(10.0) as test3",
  "log10(100.0) as test4"
)
df1.show(false)

// Non-positive values (should return null)
val df2 = spark.range(1).selectExpr(
  "log10(0.0) as zero",
  "log10(-0.5) as negative",
  "log10(-1.0) as negative2"
)
df2.show(false)

Run Command (with all 32 configuration parameters):

cd test-failures
$SPARK_HOME/bin/spark-shell \
  --master local[2] \
  --conf spark.sql.optimizer.excludedRules=org.apache.spark.sql.catalyst.optimizer.ConvertToLocalRelation,org.apache.spark.sql.catalyst.optimizer.ConstantFolding \
  --conf spark.rapids.sql.enabled=true \
  --conf spark.plugins=com.nvidia.spark.SQLPlugin \
  --conf spark.sql.queryExecutionListeners=org.apache.spark.sql.rapids.ExecutionPlanCaptureCallback \
  --conf spark.rapids.sql.explain=ALL \
  --conf spark.rapids.sql.test.isFoldableNonLitAllowed=true \
  --conf spark.rapids.sql.csv.read.decimal.enabled=true \
  --conf spark.rapids.sql.format.avro.enabled=true \
  --conf spark.rapids.sql.format.avro.read.enabled=true \
  --conf spark.rapids.sql.format.hive.text.write.enabled=true \
  --conf spark.rapids.sql.format.json.enabled=true \
  --conf spark.rapids.sql.format.json.read.enabled=true \
  --conf spark.rapids.sql.incompatibleDateFormats.enabled=true \
  --conf spark.rapids.sql.python.gpu.enabled=true \
  --conf spark.rapids.sql.rowBasedUDF.enabled=true \
  --conf spark.rapids.sql.window.collectList.enabled=true \
  --conf spark.rapids.sql.window.collectSet.enabled=true \
  --conf spark.rapids.sql.window.range.byte.enabled=true \
  --conf spark.rapids.sql.window.range.short.enabled=true \
  --conf spark.rapids.sql.expression.Ascii=true \
  --conf spark.rapids.sql.expression.Conv=true \
  --conf spark.rapids.sql.expression.GetJsonObject=true \
  --conf spark.rapids.sql.expression.JsonToStructs=true \
  --conf spark.rapids.sql.expression.StructsToJson=true \
  --conf spark.rapids.sql.exec.CollectLimitExec=true \
  --conf spark.rapids.sql.exec.FlatMapCoGroupsInPandasExec=true \
  --conf spark.rapids.sql.exec.WindowInPandasExec=true \
  --conf spark.rapids.sql.hasExtendedYearValues=false \
  --conf spark.unsafe.exceptionOnMemoryLeak=true \
  --conf spark.sql.session.timeZone=America/Los_Angeles \
  --jars $RAPIDS_JAR \
  < reproduce-math-expressions-log10.scala

---

Expected behavior

The log10 function should compute base-10 logarithm correctly for positive values and return null for non-positive values.

Test cases:

• log10(x) for x in (1 to 20).map(_ * 0.1) should work correctly
• log10(x) for x in (-5 to 0).map(_ * 0.1) should return null

Actual behavior

GPU output:
+-------------------+-----+-----+-----+
|test1 |test2|test3|test4|
+-------------------+-----+-----+-----+
|-0.9999999999999998|0.0 |1.0 |2.0 |
+-------------------+-----+-----+-----+

+----+--------+---------+
|zero|negative|negative2|
+----+--------+---------+
| | | |
+----+--------+---------+

CPU:
+-----+-----+-----+-----+
|test1|test2|test3|test4|
+-----+-----+-----+-----+
|-1.0 |0.0 |1.0 |2.0 |
+-----+-----+-----+-----+

+----+--------+---------+
|zero|negative|negative2|
+----+--------+---------+
|null|null |null |
+----+--------+---------+

---

Environment details

• Spark Version: 3.3.0
• RAPIDS Version: 25.12.0-SNAPSHOT
• CUDA Version: 12.x
• Test Suite: MathExpressionsSuite
• Test Name: log10

---

Additional context

• Original Spark Test Location: spark/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/MathExpressionsSuite.scala (line 435)
• Test Logic: Validates log10 function with positive and negative inputs

[mattahrens]

P0 to validate that collect() will return the same results between CPU and GPU for the null values.

[mythrocks]

Just clarifying here:

The part where -1.0 is reported as -0.9999999998 is not the concerning part. (This is likely because of floating point epsilon, etc.)

We need to investigate the other part, where the GPU results for log10(0.0), log10(-0.5), log10(-1.0) are reported as NaN instead of null as the CPU does.

[revans2]

see #13741 (comment) This is generally a dupe of #13741 but I want to keep both around to make sure we enable all of the correct tests.