horzcat — Concatenate arrays horizontally in MATLAB and RunMat.
horzcat(A1, A2, ...) concatenates inputs along dimension 2, matching MATLAB square-bracket syntax [A1 A2 ...]. Size compatibility and type coercion follow MATLAB semantics.
Syntax
B = horzcat()
B = horzcat(A1, An...)Inputs
A1- Type
Any- Required
- Yes
- Description
- First input array.
An- Type
Any- Required
- Variadic
- Description
- Additional input arrays.
Returns
B- Type
Any- Description
- Concatenated array, retaining its class and provider placement where supported.
Errors
RunMat:horzcat:InvalidInput- When
- Inputs are malformed or incompatible for horizontal concatenation.
- Message
horzcat: invalid input arguments
RunMat:horzcat:TypeMismatch- When
- Input classes cannot be concatenated together.
- Message
horzcat: incompatible input classes for concatenation
Behavior
Operates on numeric, logical, complex, character, string, and cell arrays with MATLAB-compatible type checking.
All inputs must have the same number of rows (dimension 1). Higher dimensions are padded with singleton sizes where necessary.
Scalars act as 1×1 building blocks, so horzcat(1, 2, 3) produces the row vector [1 2 3].
Empty inputs participate naturally. With no operands, the result is the canonical 0×0 double; nonzero all-empty operands retain compatible geometry and the applicable input class.
When the trailing 'like', prototype pair is supplied, the output matches the prototype's residency (host or GPU) and numeric category.
Compatible mixed gpuArray and host numeric operands use exact gather/concatenate/restore fallback; convert explicitly with gpuArray or gather when you need a specific placement.
GPU execution
horzcat delegates to cat(dim = 2, …). When the owning acceleration provider implements the cat hook for the operands, concatenation runs on the GPU. Otherwise RunMat gathers, concatenates on host, and attempts to upload the result; if upload is unavailable, floating output can remain on host.
Delegates to cat with dim=2; provider execution is used where supported, otherwise exact gather/host assembly and attempted restoration apply.
GPU floating precisions are f32 and f64; there is no broadcasting, elementwise or reduction fusion, and constants may be inlined.
Placement and options
Numeric, logical, complex, character, string, and cell arrays use MATLAB-compatible class checks. Inputs have matching row counts; higher dimensions are singleton-padded. Scalars are 1-by-1 building blocks. With no operands, the result is the canonical 0-by-0 double; nonzero all-empty operands retain compatible geometry and the applicable input class.
A trailing 'like', prototype pair selects output residency and numeric category. Compatible mixed host and gpuArray numeric operands use the delegated cat fallback; convert explicitly with gpuArray or gather to control placement.
The operation delegates to cat(dim=2). An owning provider with a cat hook executes on device for supported f32/f64 data; otherwise exact values gather to host for concatenation and are restored through that owner. There is no broadcasting, elementwise/reduction fusion, or independent horzcat kernel.
Examples
Concatenating matrices by columns
A = [1 2; 3 4];
B = [10 20; 30 40];
C = horzcat(A, B)Expected output:
C =
1 2 10 20
3 4 30 40Building a row vector from scalars
row = horzcat(1, 2, 3, 4)Expected output:
row = [1 2 3 4]Extending character arrays into words
lhs = ['Run' ; 'GPU'];
rhs = ['Mat' ; 'Fun'];
words = horzcat(lhs, rhs)Expected output:
words =
RunMat
GPUFunKeeping gpuArray inputs resident on the device
G1 = gpuArray(rand(256, 128));
G2 = gpuArray(rand(256, 64));
wide = horzcat(G1, G2)Preserving empties when all inputs are empty
emptyBlock = zeros(0, 3);
result = horzcat(emptyBlock, emptyBlock)Matching an output prototype with the 'like' syntax
proto = gpuArray.zeros(5, 1);
combo = horzcat(ones(5, 1), zeros(5, 1), "like", proto)Using horzcat with coding agents
Open a RunMat example with live inputs, then ask the agent to explain how horzcat changes the result.
Run a small horzcat example, explain the result, then change one input and compare the output.
Related Array functions
Shape
blkdiag · cat · circshift · diag · flip · fliplr · flipud · ipermute · kron · permute · repelem · repmat · reshape · rot90 · shiftdim · squeeze · toeplitz · tril · triu · vertcat
Accumulation
Sorting Sets
argsort · intersect · ismember · ismembertol · issorted · issortedrows · setdiff · setxor · sort · sortrows · union · unique
Creation
colon · createArray · empty · eye · false · full · inf · int16.empty · int32.empty · int64.empty · int8.empty · linspace · logspace · magic · meshgrid · nan · ndgrid · nonzeros · ones · peaks · rand · randi · randn · randperm · range · sparse · spdiags · speye · spones · sprand · true · uint16.empty · uint32.empty · uint64.empty · uint8.empty · zeros
Combinatorics
Binning
Grouping
Source you can read
Every RunMat function's source is public. Read exactly how horzcat is executed, line by line, in Rust.
- View the source for horzcat 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.