Merge branch 'main' into first_shot_memory

.cudaq_version

@@ -1,6 +1,6 @@
 {
   "cudaq": {
     "repository": "NVIDIA/cuda-quantum",
-    "ref": "bd499151326950ffb163b2afb928ca882f926ab4"
+    "ref": "d8930461128a9c564d636d7531d21f836eb1a59e"
   }
 }

.github/workflows/build_dev.yaml

@@ -0,0 +1,179 @@
+name: Build dev images
+
+on:
+  workflow_dispatch:
+    inputs:
+      force_rebuild:
+        type: boolean
+        required: true
+        description: 'Force rebuild even if image tags already exist'
+        default: false
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  build-cudaqx-dev:
+    name: Build CUDA-QX Dev Image
+    if: ${{ github.repository == 'NVIDIA/cudaqx' }}
+    strategy:
+      matrix:
+        platform: [amd64, arm64]
+      fail-fast: false
+    runs-on: linux-${{ matrix.platform }}-cpu8
+    steps:
+    - name: Login to GitHub CR
+      uses: docker/login-action@v3
+      with:
+        registry: ghcr.io
+        username: ${{ github.actor }}
+        password: ${{ github.token }}
+
+    - name: Get code
+      uses: actions/checkout@v4
+      with:
+        set-safe-directory: true
+
+    - name: Get required CUDA-Q version
+      id: get-cudaq-version
+      uses: ./.github/actions/get-cudaq-version
+
+    - name: Get additional metadata
+      id: get-cudaq-version-short
+      run: |
+        shortref=$(echo "${{ steps.get-cudaq-version.outputs.ref }}" | head -c8)
+        commit_date=$(curl -s "https://api.github.com/repos/NVIDIA/cuda-quantum/commits/${{ steps.get-cudaq-version.outputs.ref }}" | jq -r '.commit.committer.date' | cut -dT -f1)
+        echo "shortref=$shortref" >> $GITHUB_OUTPUT
+        echo "commit_date=$commit_date" >> $GITHUB_OUTPUT
+
+    - name: Check if image already exists
+      id: check-image
+      run: |
+        IMAGE_TAG="${{ steps.get-cudaq-version-short.outputs.commit_date }}-${{ steps.get-cudaq-version-short.outputs.shortref }}-${{ matrix.platform }}"
+        CONTAINER_VERSION_METADATA=$(curl -s -L -H "Accept: applicat/vnd.github+json" -H "Authorization: Bearer ${{ github.token }}" "https://api.github.com/orgs/NVIDIA/packages/container/cudaqx-dev/versions")
+        CONTAINER_TAGS=$(echo $CONTAINER_VERSION_METADATA | jq -r '.[] | .metadata.container.tags[]')
+
+        if echo "$CONTAINER_TAGS" | grep -qx "$IMAGE_TAG"; then
+          echo "Image tag $IMAGE_TAG already exists. Skipping build."
+          echo "IMAGE_EXISTS=true" >> $GITHUB_OUTPUT
+        else
+          echo "Image tag $IMAGE_TAG does not exist. Proceeding with build."
+          echo "IMAGE_EXISTS=false" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Set up context for buildx
+      if: ${{ inputs.force_rebuild || steps.check-image.outputs.IMAGE_EXISTS == 'false' }}
+      run: |
+        docker context create builder_context
+      shell: bash --noprofile --norc -euo pipefail {0}
+
+    - name: Set up buildx runner
+      if: ${{ inputs.force_rebuild ||steps.check-image.outputs.IMAGE_EXISTS == 'false' }}
+      uses: docker/setup-buildx-action@v3
+      with:
+        endpoint: builder_context
+        driver-opts: network=host
+
+    - name: Build dev image
+      if: ${{ inputs.force_rebuild ||steps.check-image.outputs.IMAGE_EXISTS == 'false' }}
+      uses: docker/build-push-action@v5
+      with:
+        file: ./docker/build_env/cudaqx.dev.Dockerfile
+        tags: |
+          ghcr.io/nvidia/cudaqx-dev:${{ steps.get-cudaq-version-short.outputs.commit_date }}-${{ steps.get-cudaq-version-short.outputs.shortref }}-${{ matrix.platform }}
+          ghcr.io/nvidia/cudaqx-dev:${{ steps.get-cudaq-version-short.outputs.shortref }}-${{ matrix.platform }}
+          ghcr.io/nvidia/cudaqx-dev:latest-${{ matrix.platform }}
+        platforms: linux/${{ matrix.platform }}
+        push: true
+
+  build-cudaqx-pydev:
+    name: Build CUDA-QX Python Dev Image
+    if: ${{ github.repository == 'NVIDIA/cudaqx' }}
+    strategy:
+      matrix:
+        python: ['3.10', '3.11', '3.12']
+        platform: ['amd64', 'arm64']
+      fail-fast: false
+    runs-on: linux-${{ matrix.platform }}-cpu8
+    steps:
+    - name: Login to GitHub CR
+      uses: docker/login-action@v3
+      with:
+        registry: ghcr.io
+        username: ${{ github.actor }}
+        password: ${{ github.token }}
+
+    - name: Get code
+      uses: actions/checkout@v4
+      with:
+        set-safe-directory: true
+
+    - name: Get required CUDA-Q version
+      id: get-cudaq-version
+      uses: ./.github/actions/get-cudaq-version
+
+    - name: Get additional metadata
+      id: get-cudaq-version-short
+      run: |
+        shortref=$(echo "${{ steps.get-cudaq-version.outputs.ref }}" | head -c8)
+        commit_date=$(curl -s "https://api.github.com/repos/NVIDIA/cuda-quantum/commits/${{ steps.get-cudaq-version.outputs.ref }}" | jq -r '.commit.committer.date' | cut -dT -f1)
+        echo "shortref=$shortref" >> $GITHUB_OUTPUT
+        echo "commit_date=$commit_date" >> $GITHUB_OUTPUT
+
+    - name: Check if image already exists
+      id: check-image
+      run: |
+        IMAGE_TAG="${{ steps.get-cudaq-version-short.outputs.commit_date }}-${{ steps.get-cudaq-version-short.outputs.shortref }}-${{ matrix.platform }}"
+        CONTAINER_VERSION_METADATA=$(curl -s -L -H "Accept: applicat/vnd.github+json" -H "Authorization: Bearer ${{ github.token }}" "https://api.github.com/orgs/NVIDIA/packages/container/cudaqx-dev/versions")
+        CONTAINER_TAGS=$(echo $CONTAINER_VERSION_METADATA | jq -r '.[] | .metadata.container.tags[]')
+
+        if echo "$CONTAINER_TAGS" | grep -qx "$IMAGE_TAG"; then
+          echo "Image tag $IMAGE_TAG already exists. Skipping build."
+          echo "IMAGE_EXISTS=true" >> $GITHUB_OUTPUT
+        else
+          echo "Image tag $IMAGE_TAG does not exist. Proceeding with build."
+          echo "IMAGE_EXISTS=false" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Set up context for buildx
+      if: ${{ inputs.force_rebuild ||steps.check-image.outputs.IMAGE_EXISTS == 'false' }}
+      run: |
+        docker context create builder_context
+      shell: bash --noprofile --norc -euo pipefail {0}
+
+    - name: Set up buildx runner
+      if: ${{ inputs.force_rebuild ||steps.check-image.outputs.IMAGE_EXISTS == 'false' }}
+      uses: docker/setup-buildx-action@v3
+      with:
+        endpoint: builder_context
+        driver-opts: network=host
+
+    - name: Build pydev image
+      if: ${{ inputs.force_rebuild ||steps.check-image.outputs.IMAGE_EXISTS == 'false' }}
+      uses: docker/build-push-action@v5
+      with:
+        file: ./docker/build_env/cudaqx.wheel.Dockerfile
+        build-args: |
+          base_image=ghcr.io/nvidia/cuda-quantum-devdeps:manylinux-${{ matrix.platform }}-cu12.0-gcc11-main
+          python_version=${{ matrix.python }}
+        tags: |
+          ghcr.io/nvidia/cudaqx-dev:${{ steps.get-cudaq-version-short.outputs.commit_date }}-${{ steps.get-cudaq-version-short.outputs.shortref }}-py${{ matrix.python }}-${{ matrix.platform }}
+          ghcr.io/nvidia/cudaqx-dev:${{ steps.get-cudaq-version-short.outputs.shortref }}-py${{ matrix.python }}-${{ matrix.platform }}
+          ghcr.io/nvidia/cudaqx-dev:latest-py${{ matrix.python }}-${{ matrix.platform }}
+        platforms: linux/${{ matrix.platform }}
+        push: true
+
+  cleanup:
+    name: Cleanup
+    needs: [build-cudaqx-dev, build-cudaqx-pydev]
+    runs-on: ubuntu-latest
+    if: ${{ github.repository == 'NVIDIA/cudaqx' }}
+    steps:
+    - name: Delete untagged images
+      uses: actions/delete-package-versions@v5
+      with: 
+        package-name: cudaqx-dev
+        package-type: 'container'
+        min-versions-to-keep: 0
+        delete-only-untagged-versions: 'true'

.github/workflows/update-cudaq-dep.yml

@@ -1,5 +1,12 @@
 on:
   workflow_dispatch:
+    inputs:
+      cudaq_branch_name:
+        type: string
+        required: true
+        description: 'Branch name for upstream CUDA-Q repo (e.g. main or pull-request/2407)'
+        default: main
+
   schedule:
     - cron: 0 1 * * 6
 
@@ -22,7 +29,7 @@ jobs:
 
       - name: Fetch the latest commit
         run: |
-          SHA=$(curl -s "https://api.github.com/repos/NVIDIA/cuda-quantum/commits/main" | jq -r '.sha')
+          SHA=$(curl -s "https://api.github.com/repos/NVIDIA/cuda-quantum/commits/${{ inputs.cudaq_branch_name }}" | jq -r '.sha')
           echo "Latest SHA: $SHA"
           echo "sha=$SHA" >> $GITHUB_ENV
 
@@ -64,8 +71,13 @@ jobs:
           GITHUB_TOKEN: ${{ github.token }}
           BRANCH_NAME: ${{ steps.commit_and_push.outputs.BRANCH_NAME }}
         run: |
+          if [[ "${{ inputs.cudaq_branch_name }}" == "main" ]]; then
+            TITLE_STR="Bump CUDA-Q commit"
+          else
+            TITLE_STR="Bump CUDA-Q commit (${{ inputs.cudaq_branch_name }} DO NOT MERGE)"
+          fi
           gh pr create \
-            --title "Bump CUDA-Q commit" \
-            --body "Auto update to the latest CUDA-Q commit" \
+            --title "${TITLE_STR}" \
+            --body "Auto update to the latest CUDA-Q commit on ${{ inputs.cudaq_branch_name }} branch" \
             --head "${BRANCH_NAME}" \
             --base "main"

docker/build_env/cudaqx.dev.Dockerfile

@@ -0,0 +1,34 @@
+# ============================================================================ #
+# Copyright (c) 2025 NVIDIA Corporation & Affiliates.                          #
+# All rights reserved.                                                         #
+#                                                                              #
+# This source code and the accompanying materials are made available under     #
+# the terms of the Apache License 2.0 which accompanies this distribution.     #
+# ============================================================================ #
+
+FROM ghcr.io/nvidia/cuda-quantum-devdeps:ext-cu12.0-gcc11-main
+
+LABEL org.opencontainers.image.description="Dev tools for building and testing CUDA-QX libraries"
+LABEL org.opencontainers.image.source="https://github.com/NVIDIA/cudaqx"
+LABEL org.opencontainers.image.title="cuda-qx-dev"
+LABEL org.opencontainers.image.url="https://github.com/NVIDIA/cudaqx"
+
+RUN apt-get update && apt-get install -y gfortran libblas-dev jq cuda-nvtx-12-0 \
+  && python3 -m pip install cmake --user \
+  && apt-get autoremove -y --purge && apt-get clean && rm -rf /var/lib/apt/lists/*
+
+COPY .cudaq_version /cudaq_version
+
+RUN mkdir -p /workspaces/cudaq && cd /workspaces/cudaq \
+  && git init \
+  && CUDAQ_REPO=$(jq -r '.cudaq.repository' /cudaq_version) \
+  && CUDAQ_COMMIT=$(jq -r '.cudaq.ref' /cudaq_version) \
+  && git remote add origin https://github.com/${CUDAQ_REPO} \
+  && git fetch -q --depth=1 origin ${CUDAQ_COMMIT} \
+  && git reset --hard FETCH_HEAD \
+  && bash scripts/build_cudaq.sh -v \
+  && rm -rf build
+
+#RUN mkdir -p /workspaces/cudaqx && cd /workspaces/cudaqx \
+#  && ~/.local/bin/cmake -G Ninja -DCMAKE_BUILD_TYPE=Release -DCUDAQ_DIR=~/.cudaq/lib/cmake/cudaq .. \
+#  && ninja install

docker/build_env/cudaqx.wheel.Dockerfile

@@ -0,0 +1,32 @@
+# ============================================================================ #
+# Copyright (c) 2025 NVIDIA Corporation & Affiliates.                          #
+# All rights reserved.                                                         #
+#                                                                              #
+# This source code and the accompanying materials are made available under     #
+# the terms of the Apache License 2.0 which accompanies this distribution.     #
+# ============================================================================ #
+
+ARG base_image=ghcr.io/nvidia/cuda-quantum-devdeps:manylinux-amd64-cu12.0-gcc11-main
+FROM ${base_image}
+
+ARG python_version=3.10
+
+LABEL org.opencontainers.image.description="Dev tools for building and testing CUDA-QX libraries"
+LABEL org.opencontainers.image.source="https://github.com/NVIDIA/cudaqx"
+LABEL org.opencontainers.image.title="cuda-qx-dev"
+LABEL org.opencontainers.image.url="https://github.com/NVIDIA/cudaqx"
+
+RUN dnf install -y jq
+RUN mkdir -p /workspaces/cudaqx
+COPY .cudaq_version /workspaces/cudaqx
+COPY .github/workflows/scripts/build_cudaq.sh /workspaces/cudaqx
+
+RUN mkdir -p /workspaces/cudaqx/cudaq && cd /workspaces/cudaqx/cudaq \
+  && git init \
+  && CUDAQ_REPO=$(jq -r '.cudaq.repository' /workspaces/cudaqx/.cudaq_version) \
+  && CUDAQ_COMMIT=$(jq -r '.cudaq.ref' /workspaces/cudaqx/.cudaq_version) \
+  && git remote add origin https://github.com/${CUDAQ_REPO} \
+  && git fetch -q --depth=1 origin ${CUDAQ_COMMIT} \
+  && git reset --hard FETCH_HEAD \
+  && bash ../build_cudaq.sh --python-version ${python_version} \
+  && rm -rf build

libs/solvers/lib/stateprep/uccsd.cpp

@@ -407,10 +407,10 @@ __qpu__ std::size_t getNumVirtualBeta(std::size_t numElectrons,
 __qpu__ void uccsd(cudaq::qview<> qubits, const std::vector<double> &thetas,
                    std::size_t numElectrons, std::size_t spin) {
 
-  auto numOccAlpha = getNumOccupiedAlpha(numElectrons, spin, qubits.size());
-  auto numOccBeta = getNumOccupiedBeta(numElectrons, spin, qubits.size());
-  auto numVirtAlpha = getNumVirtualAlpha(numElectrons, spin, qubits.size());
-  auto numVirtBeta = getNumVirtualBeta(numElectrons, spin, qubits.size());
+  int numOccAlpha = getNumOccupiedAlpha(numElectrons, spin, qubits.size());
+  int numOccBeta = getNumOccupiedBeta(numElectrons, spin, qubits.size());
+  int numVirtAlpha = getNumVirtualAlpha(numElectrons, spin, qubits.size());
+  int numVirtBeta = getNumVirtualBeta(numElectrons, spin, qubits.size());
   std::vector<std::size_t> occupiedAlpha(numOccAlpha),
       virtualAlpha(numVirtAlpha), occupiedBeta(numOccBeta),
       virtualBeta(numVirtBeta);
@@ -433,10 +433,11 @@ __qpu__ void uccsd(cudaq::qview<> qubits, const std::vector<double> &thetas,
       occupiedBeta[counter] = i * 2 + 1;
       counter++;
     }
-    counter = 0;
 
+    counter = 0;
     for (std::size_t i = 0; i < numVirtBeta; i++) {
       virtualBeta[counter] = i * 2 + numElectrons - 1;
+      counter++;
     }
 
   } else {
@@ -507,18 +508,18 @@ __qpu__ void uccsd(cudaq::qview<> qubits, const std::vector<double> &thetas,
         }
 
   counter = 0;
-  for (std::size_t p = 0; p < numOccAlpha - 1; p++)
-    for (std::size_t q = p + 1; q < numOccAlpha; q++)
-      for (std::size_t r = 0; r < numVirtAlpha - 1; r++)
-        for (std::size_t s = r + 1; s < numVirtAlpha; s++)
+  for (int p = 0; p < numOccAlpha - 1; p++)
+    for (int q = p + 1; q < numOccAlpha; q++)
+      for (int r = 0; r < numVirtAlpha - 1; r++)
+        for (int s = r + 1; s < numVirtAlpha; s++)
           counter++;
 
   std::vector<std::size_t> doublesAlpha(4 * counter);
   counter = 0;
-  for (std::size_t p = 0; p < numOccAlpha - 1; p++)
-    for (std::size_t q = p + 1; q < numOccAlpha; q++)
-      for (std::size_t r = 0; r < numVirtAlpha - 1; r++)
-        for (std::size_t s = r + 1; s < numVirtAlpha; s++) {
+  for (int p = 0; p < numOccAlpha - 1; p++)
+    for (int q = p + 1; q < numOccAlpha; q++)
+      for (int r = 0; r < numVirtAlpha - 1; r++)
+        for (int s = r + 1; s < numVirtAlpha; s++) {
           doublesAlpha[counter] = occupiedAlpha[p];
           counter++;
           doublesAlpha[counter] = occupiedAlpha[q];
@@ -530,17 +531,17 @@ __qpu__ void uccsd(cudaq::qview<> qubits, const std::vector<double> &thetas,
         }
 
   counter = 0;
-  for (std::size_t p = 0; p < numOccBeta - 1; p++)
-    for (std::size_t q = p + 1; q < numOccBeta; q++)
-      for (std::size_t r = 0; r < numVirtBeta - 1; r++)
-        for (std::size_t s = r + 1; s < numVirtBeta; s++)
+  for (int p = 0; p < numOccBeta - 1; p++)
+    for (int q = p + 1; q < numOccBeta; q++)
+      for (int r = 0; r < numVirtBeta - 1; r++)
+        for (int s = r + 1; s < numVirtBeta; s++)
           counter++;
   std::vector<std::size_t> doublesBeta(4 * counter);
   counter = 0;
-  for (std::size_t p = 0; p < numOccBeta - 1; p++)
-    for (std::size_t q = p + 1; q < numOccBeta; q++)
-      for (std::size_t r = 0; r < numVirtBeta - 1; r++)
-        for (std::size_t s = r + 1; s < numVirtBeta; s++) {
+  for (int p = 0; p < numOccBeta - 1; p++)
+    for (int q = p + 1; q < numOccBeta; q++)
+      for (int r = 0; r < numVirtBeta - 1; r++)
+        for (int s = r + 1; s < numVirtBeta; s++) {
           doublesBeta[counter] = occupiedBeta[p];
           counter++;
           doublesBeta[counter] = occupiedBeta[q];