Number Functions

Number functions cover value-level numeric utilities that are not really math formulas: display formatting, base conversion, Roman numeral interop, and predicates for validating imported data. For numeric formulas, trigonometry, rounding, constants, and random values, use Math Functions.

Most functions return null when an argument is null, has the wrong type, overflows, or uses an unsupported option. That keeps ingestion queries simple: invalid values can be filtered with IS NOT NULL or replaced with coalesce.

Overview

Goal	Function
Format for display	number.format(n[, precision[, thousands]])
Convert integer to base 2..36	number.to_base(n, radix)
Parse base 2..36 text	number.from_base(text, radix)
Roman numeral interop	number.to_roman(n), number.from_roman(text)
Integer predicates	number.is_integer(n), number.is_even(n), number.is_odd(n)
Sign predicates	number.is_positive(n), number.is_negative(n), number.is_zero(n)
Float predicates	number.is_nan(n), number.is_finite(n), number.is_infinite(n)
Bit operations	bits.and, bits.or, bits.xor, bits.not, bits.shift_left, bits.shift_right

Formatting

number.format(n[, precision[, thousands]]) returns a string. When precision is provided, the number is rounded to that many decimal places. When thousands is provided, it is inserted between groups of three digits in the integer part.

RETURN number.format(1234)             -- '1234'
RETURN number.format(1234.567, 2)      -- '1234.57'
RETURN number.format(1234567, 0, ',')  -- '1,234,567'
RETURN number.format(-12345.6, 1, '_') -- '-12_345.6'

Use math.round when the result should remain numeric:

RETURN math.round(1234.567, 2)         -- 1234.57
RETURN number.format(1234.567, 2)      -- '1234.57'

Radix Conversion

Use number.to_base(n, radix) and number.from_base(text, radix) for integer identifiers, compact codes, imported binary or hexadecimal fields, and similar data-shaping jobs. radix must be between 2 and 36. Output uses lowercase digits 0-9 and a-z.

RETURN number.to_base(255, 16)       -- 'ff'
RETURN number.to_base(42, 2)         -- '101010'
RETURN number.to_base(-10, 2)        -- '-1010'

RETURN number.from_base('ff', 16)    -- 255
RETURN number.from_base('101010', 2) -- 42
RETURN number.from_base('-1010', 2)  -- -10

Invalid digits, unsupported radix values, and integer overflow return null:

RETURN number.from_base('ff', 10)    -- null
RETURN number.to_base(10, 1)         -- null

Import Hex IDs

UNWIND $rows AS row
WITH row, number.from_base(row.hex_id, 16) AS id
WHERE id IS NOT NULL
MERGE (:ExternalThing {id: id})

Roman Numerals

number.to_roman(n) supports integers from 1 through 3999. number.from_roman(text) accepts uppercase or lowercase Roman numerals and returns an integer.

RETURN number.to_roman(1994)        -- 'MCMXCIV'
RETURN number.from_roman('MCMXCIV') -- 1994
RETURN number.to_roman(0)           -- null

Roman parsing is intended for interop and display cleanup, not strict historical validation. Prefer storing ordinary integers as properties.

Predicates

number.is_integer(n) accepts both INTEGER values and finite FLOAT values with no fractional part. number.is_even and number.is_odd require an INTEGER; they return null for floats, even when the float looks integral. The sign predicates accept both integers and floats.

RETURN number.is_integer(42)     -- true
RETURN number.is_integer(42.0)   -- true
RETURN number.is_integer(42.5)   -- false

RETURN number.is_even(42)        -- true
RETURN number.is_odd(41)         -- true
RETURN number.is_even(42.0)      -- null

RETURN number.is_positive(0.1)   -- true
RETURN number.is_negative(-1)    -- true
RETURN number.is_zero(0.0)       -- true

The finite-value predicates are mostly useful when a host binding or JSON import can produce special floating-point values:

RETURN number.is_finite(1.5)     -- true
RETURN number.is_nan($maybe_nan) -- true, when a host binding supplies NaN

Bit Operations

Bit operations live in the bits.* namespace because they are integer operations over the two's-complement representation, not ordinary numeric formulas.

RETURN bits.and(12, 10)          -- 8
RETURN bits.or(12, 10)           -- 14
RETURN bits.xor(12, 10)          -- 6
RETURN bits.not(0)               -- -1
RETURN bits.shift_left(3, 2)     -- 12
RETURN bits.shift_right(12, 2)   -- 3

For new queries, prefer the named bits.* helpers over number.bitop(a, op, b). The older number.bitop form remains available for generated queries that need to pass the operation name as a string.

RETURN number.bitop(12, 'and', 10) -- 8

Common Patterns

Validate Imported Integers

UNWIND $rows AS row
WITH row, cast.try(row.rank, INTEGER) AS rank
WHERE number.is_integer(rank) AND number.is_even(rank)
CREATE (:ImportedRank {rank: rank, source: row.source})

Build Stable Short Codes

MATCH (n:Thing)
RETURN n.id, string.upper(number.to_base(n.id, 36)) AS code
ORDER BY n.id

Keep Numeric And Display Values Separate

MATCH (invoice:Invoice)
WITH invoice, invoice.total_cents / 100.0 AS total
RETURN invoice.id,
       total,
       number.format(total, 2, ',') AS display_total