{
"id": "fn-gpu-inverse",
"dataComponent": "gpu",
"heading": {
"title": "inverse",
"badges": ["GPU"]
},
"synopsis": "Computes the matrix inverse of a (2×2) tensor on the GPU. In many builds, real GPU code is not implemented yet, so it may fallback to CPU.",
"codeBlocks": [
"extend(\"gpu\")\\n\\n# 1) Invert a 2×2 matrix => shape [2,2]\\nM = gpu:to_tensor([[1,2],[3,4]])\\ninvM = gpu:inverse(M)\\nslog(gpu:to_array(invM))\\n# => e.g. [[-2.0,1.0],[1.5,-0.5]]\\n\\n# 2) If determinant is near zero => bridging raises an error.\\n# 3) Some builds have no real GPU kernel => logs fallback CPU.\\n"
],
"notes": [
"gpu:inverse => only supports 2×2. If shape is not [2,2], bridging raises an error. If the matrix is nearly singular, it errors with 'cannot invert'.",
"Often used as a minimal demonstration. For bigger NxN, see advanced GPU solvers or third-party libraries.",
"No partial usage by default; expects exactly 1 argument."
]
}