From 7de54ecc029b6bae0ba6103e034abfa9bf0dace7 Mon Sep 17 00:00:00 2001 From: Talley Lambert Date: Tue, 2 Jun 2026 11:44:17 -0400 Subject: [PATCH 1/3] Update README with cuOpt installation instructions Added installation instructions for cuOpt and updated pip commands to use quotes. --- README.md | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index b4da5ad..23644eb 100644 --- a/README.md +++ b/README.md @@ -39,7 +39,7 @@ suite](https://www.scipopt.org), via explicitly: ```sh -pip install ilpy[scip] +pip install "ilpy[scip]" ``` ### ... with Gurobi @@ -48,9 +48,22 @@ If you want to use Gurobi (which requires a license), you can bring in the [`gurobipy`](https://pypi.org/project/gurobipy/) dependency with: ```sh -pip install ilpy[gurobi] +pip install "ilpy[gurobi]" ``` +### ... with cuOpt + +`cuOpt` is NVIDIA's GPU-accelerated optimization engine. `ilpy[cuopt]` +includes `cuopt-cu12`, which is hosted on NVIDIA's package index. +Install it with pip by adding NVIDIA as an extra index: + +```sh +pip install "ilpy[cuopt]" --extra-index-url https://pypi.nvidia.com +``` + +*if you want to install a different version, such as `cuopt-cu13`, you +may, but you will also need to install `scipy`* + ### On conda If you prefer to use conda: From 44ebdc69d8b00f5ae8fb2c9d76cac25068db86f1 Mon Sep 17 00:00:00 2001 From: Talley Lambert Date: Thu, 12 Nov 2026 09:16:50 +0100 Subject: [PATCH 2/3] docs: cuopt-cu12 is on PyPI now, fix cu13 install note Co-Authored-By: Claude Opus 5.5 --- README.md | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) diff --git a/README.md b/README.md index 23644eb..4de43a0 100644 --- a/README.md +++ b/README.md @@ -53,16 +53,15 @@ pip install "ilpy[gurobi]" ### ... with cuOpt -`cuOpt` is NVIDIA's GPU-accelerated optimization engine. `ilpy[cuopt]` -includes `cuopt-cu12`, which is hosted on NVIDIA's package index. -Install it with pip by adding NVIDIA as an extra index: +[cuOpt](https://github.com/NVIDIA/cuopt) is NVIDIA's GPU-accelerated +optimization engine (Linux only, requires an NVIDIA GPU): ```sh -pip install "ilpy[cuopt]" --extra-index-url https://pypi.nvidia.com +pip install "ilpy[cuopt]" ``` -*if you want to install a different version, such as `cuopt-cu13`, you -may, but you will also need to install `scipy`* +This installs `cuopt-cu12`. For CUDA 13, skip the extra and install +`cuopt-cu13` directly: `pip install ilpy cuopt-cu13` ### On conda From 47339d9bd7a76679e89e3aef953800e9c744e018 Mon Sep 17 00:00:00 2001 From: Talley Lambert Date: Thu, 12 Nov 2026 09:19:07 +0100 Subject: [PATCH 3/3] docs: install cuopt-cu12/cu13 directly instead of the extra Co-Authored-By: Claude Opus 5.5 --- README.md | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/README.md b/README.md index 4de43a0..3afa36d 100644 --- a/README.md +++ b/README.md @@ -54,15 +54,14 @@ pip install "ilpy[gurobi]" ### ... with cuOpt [cuOpt](https://github.com/NVIDIA/cuopt) is NVIDIA's GPU-accelerated -optimization engine (Linux only, requires an NVIDIA GPU): +optimization engine (Linux only, requires an NVIDIA GPU). Install the +package that matches your CUDA version: ```sh -pip install "ilpy[cuopt]" +pip install ilpy cuopt-cu12 # CUDA 12 +pip install ilpy cuopt-cu13 # CUDA 13 ``` -This installs `cuopt-cu12`. For CUDA 13, skip the extra and install -`cuopt-cu13` directly: `pip install ilpy cuopt-cu13` - ### On conda If you prefer to use conda: