Reference Updated May 13, 2026

Math & numeric

Use this category when you need numeric transformations and helpers in formulas.

These functions change values row-wise and support tasks such as rounding, caps and floors, powers, roots, logarithms, and element-wise adjustments.

Start here if…

  • You want numeric helpers such as rounding, caps/floors, sign, or modulo
  • You want to apply an element-wise constant change
  • You need powers, roots, or logarithms
  • You need controlled multiplication or division with optional mismatch validation

Not here if…


Mental model

  • Most functions in this category work row-wise and usually keep the input shape
  • Constants are 0-dimensional; if + 1 causes unexpected aggregation effects, use ADDEACH(...) for an explicit element-wise adjustment

Common patterns

Add a constant to every cell ADDEACH('Sales', 1) Use when you want an element-wise change without aggregation side effects.

Compute a ratio while keeping only shared dimensions RATIO('Gross Profit', 'Revenue') Use when you want a display-style ratio that keeps only the joint dimensionality of both inputs. RATIO must be an end node.

Cap / floor values MAX('Margin', 0) or MIN('Node1', 'Node2') Use when you need simple numeric thresholds without IF(...).

Round for reporting ROUND('Revenue') Use when you need stable rounding for display or downstream steps.

Safe divide or multiply with validation DIVIDE('Node1', 'Node2', "FailOnMissingFirst") Use when you want clearer mismatch feedback instead of raw / or *.

Transform with logs, power, or roots LN('Node'), LOG('Node1', 'Node2'), POWER('Node1', 'Node2'), SQRT('Node') Use when your method requires non-linear scaling.


Functions in this category

FunctionDescription
ABSReturns the absolute value of each value in the node.
ADDEACHAdds the specified amount to each row in the node without aggregation.
DIVIDEDivides the first input by the second with optional validation that provides detailed errors on mismatches.
MULTIPLYMultiplies inputs with optional validation that provides detailed errors on mismatches.
EXPCalculates e raised to the power of the input values.
FACTReturns the factorial of each input value.
LNReturns the natural logarithm of the input.
LOGReturns the logarithm of a value with respect to a specified base.
MAXReturns the row-wise maximum of two nodes.
MINReturns the row-wise minimum of two nodes.
MODReturns the remainder after dividing the first input by the second.
POWERRaises values from the first node to exponents from the second input.
RATIODivides the first node by the second while retaining only joint dimensions. Must be an end node, the result cannot be used in further calculations.
ROUNDRounds each value to a specified number of decimal digits.
SIGNUMReturns the sign of each number: 1 for positive, −1 for negative, 0 for zero.
SQRTReturns the square root of each value in the node.

Choosing between similar functions

ADDEACH vs + with a constant

  • Use ADDEACH when you want an explicit element-wise adjustment
  • Use + with a constant only if you intentionally want standard arithmetic behavior with a 0-dimensional value

DIVIDE/DIVISION (/) vs RATIO

  • Use DIVISION (/)for standard division
  • Use DIVIDE when you want standard division with optional mismatch validation
  • Use RATIO when you want a ratio that keeps only joint dimensions for display purposes; RATIO must be an end node

MIN/MAX vs IF

  • Use MIN/MAX for simple caps and floors
  • Use IF when the rule depends on broader conditional logic

LN vs LOG vs EXP

  • Use LN for natural logarithm
  • Use LOG when you need a specific base
  • Use EXP for exponentiation with base e

Pitfalls & troubleshooting

  • Unexpected aggregation after adding a constant: replace + 1 style adjustments with ADDEACH(...)
  • RATIO cannot be reused downstream: if the result must feed other nodes, use / or DIVIDE and set aggregation appropriately
  • Division looks wrong: check denominators for zero or missing values and validate both inputs at the same intersections
  • LN/LOG/SQRT return missing results: confirm the input domain (negative values or zeros where not supported)
  • Rounding drift: apply ROUND late, not in intermediate calculations

Was this page helpful?