isnan — Determine which array elements are NaN.
isnan returns a same-shaped logical mask that identifies NaN components in real and complex values.
Syntax
tf = isnan(A)Inputs
A- Type
Any- Required
- Yes
- Description
- Numeric, logical, character, or string input to classify element by element.
Returns
tf- Type
LogicalArray- Description
- Same-shaped logical mask that is true where either numeric component is NaN.
Errors
RunMat:isnan:InvalidInput- When
- Input is sparse or is not numeric, logical, character, or string data.
- Message
isnan: invalid input
RunMat:isnan:InternalError- When
- Internal mask construction, provider execution, or residency restoration fails.
- Message
isnan: internal error
RunMat:isnan:TooManyOutputs- When
- More than one output is requested.
- Message
isnan: too many output arguments
Values and classes
isnan(A) returns true where an element contains an IEEE NaN. A complex element is true when either its real or imaginary component is NaN.
Fixed-width integers, logical values, and character code points cannot contain NaN and therefore produce false masks. String scalars return false and string arrays produce same-shaped false masks. Sparse and unrelated container values are rejected.
Shape, residency, and fusion
The result is a logical scalar for scalar input and a same-shaped logical array for nonscalar or empty input.
Resident floating input uses the exact owner's logical_isnan operation. A typed unsupported result may gather once and restore the logical output to that owner. Integer input needs no payload download because its class determines an all-false result. Compatible elementwise expressions may fuse.
Examples
Classify a NaN scalar
tf = isnan(NaN)Expected output:
tf = trueBuild a NaN mask
A = [1 NaN 2; 3 4 NaN];
tf = isnan(A)Expected output:
tf = [false true false; false false true]Classify either complex component
Z = complex([1 NaN 3], [2 0 NaN]);
tf = isnan(Z)Expected output:
tf = [false true true]Classify character code points
tf = isnan(['R' 'u' 'n'])Expected output:
tf = [false false false]Classify fixed-width integers exactly
A = [uint64(0), uint64(9007199254740992) + uint64(1), intmax('uint64')];
tf = isnan(A)Expected output:
tf = [false false false]Keep the NaN mask resident
A = gpuArray(single([1 NaN 3]));
gtf = isnan(A);
tf = gather(gtf)Expected output:
gtf remains a gpuArray and tf = [false true false]Using isnan with coding agents
Open a RunMat example with live inputs, then ask the agent to explain how isnan changes the result.
Run a small isnan example, explain the result, then change one input and compare the output.
FAQ
Does isnan modify its input?⌄
No. It returns a new logical mask and leaves host- or provider-resident input unchanged.
How are complex values classified?⌄
An element is true when either its real or imaginary component is NaN. Infinity alone does not count as NaN.
How are logical, character, and string values classified?⌄
These classes cannot contain NaN, so they produce false values with the corresponding shape.
How should NaN values be filtered?⌄
Use the logical result as an index or combine it with any along the dimension whose rows or columns should be selected.
Does gpuArray input use a provider operation?⌄
Yes. RunMat asks the exact owner for logical_isnan; only a typed unsupported result enters the class-preserving fallback path.
What happens for an empty array?⌄
The result is an empty logical array with the same dimensions as the input.
How do isnan, isinf, and isfinite differ?⌄
isnan identifies NaN, isinf identifies positive or negative infinity, and isfinite identifies values that are neither. For an ordinary real floating value, exactly one predicate is true.
Can isnan participate in a fused expression?⌄
Yes. The placement and fusion planner can combine compatible elementwise work when the selected provider supports the complete region.
Related Logical functions
Source you can read
Every RunMat function's source is public. Read exactly how isnan is executed, line by line, in Rust.
- View the source for isnan in Rust on GitHub
- Learn how the RunMat runtime works
- Found a bug? Open an issue with a minimal reproduction.
About RunMat
RunMat is a runtime that executes MATLAB-syntax code fast on any GPU. The runtime is free for individuals, including freelancers; students and faculty at an academic institution, for teaching or research there; contributors to an open-source project, for work on that project; organizations with 250 or fewer people and US$10M or less in annual revenue; and anyone, for non-production use. See runmat.com/license.
- RunMat automatically optimizes your math for GPU execution on Apple, Nvidia, and AMD hardware. No code changes needed. Simulations that took hours now take minutes.
- Start running code in seconds. RunMat runs in the browser, on the desktop, or from the CLI. No license server, no IT ticket.