Merge branch 'development' into diffusion_analytic

Author: zhichen3

.github/workflows/dependencies/dependencies_hip.sh

@@ -33,7 +33,9 @@ sudo apt-key add rocm.gpg.key
 
 source /etc/os-release # set UBUNTU_CODENAME: focal or jammy or ...
 
-echo "deb [arch=amd64] https://repo.radeon.com/rocm/apt/${1-latest} ${UBUNTU_CODENAME} main" \
+VERSION=${1-6.3.2}
+
+echo "deb [arch=amd64] https://repo.radeon.com/rocm/apt/${VERSION} ${UBUNTU_CODENAME} main" \
   | sudo tee /etc/apt/sources.list.d/rocm.list
 echo 'export PATH=/opt/rocm/llvm/bin:/opt/rocm/bin:/opt/rocm/profiler/bin:/opt/rocm/opencl/bin:$PATH' \
   | sudo tee -a /etc/profile.d/rocm.sh
@@ -53,16 +55,16 @@ sudo apt-get install -y --no-install-recommends \
     libnuma-dev     \
     libopenmpi-dev  \
     openmpi-bin     \
-    rocm-dev        \
-    roctracer-dev   \
-    rocprofiler-dev \
-    rocrand-dev     \
-    rocfft-dev      \
-    rocprim-dev     \
-    rocsparse-dev
+    rocm-dev${VERSION}        \
+    roctracer-dev${VERSION}   \
+    rocprofiler-dev${VERSION} \
+    rocrand-dev${VERSION}     \
+    rocfft-dev${VERSION}      \
+    rocprim-dev${VERSION}     \
+    rocsparse-dev${VERSION}
 
 # hiprand-dev is a new package that does not exist in old versions
-sudo apt-get install -y --no-install-recommends hiprand-dev || true
+sudo apt-get install -y --no-install-recommends hiprand-dev${VERSION} || true
 
 # activate
 #

.github/workflows/hip.yml

@@ -8,7 +8,7 @@ concurrency:
 
 jobs:
   hip-compile:
-    runs-on: ubuntu-20.04
+    runs-on: ubuntu-24.04
     steps:
       - uses: actions/checkout@v4
         with:
@@ -24,7 +24,7 @@ jobs:
           cd ../..
 
       - name: Dependencies
-        run: .github/workflows/dependencies/dependencies_hip.sh
+        run: .github/workflows/dependencies/dependencies_hip.sh 6.3.2
 
       - name: compile flame_wave
         run: |

CHANGES.md

@@ -1,3 +1,39 @@
+# 25.03
+
+  * nova problem: update the inputs to match Smith Clark & Zingale
+    (2025) and support 3D (#3031, #3040)
+
+  * docs updates (#3039)
+
+  * update the network names for the subchandra and subch_base
+    problems (#3033)
+
+  * update the HIP CI action to support a later runner (#3034)
+
+  * add support for printing from `check_for_name` from GPUs (#3036)
+
+  * remove some old sparse Jacobian stuff from the build system
+    (#3035)
+
+# 25.02
+
+  * docs updates (#3027)
+
+  * add the `simple_convection` problem (#3016)
+
+  * codespell CI updates (#3025)
+
+  * only print mid-timestep `estTimeStep` outputs if relevant (#3023)
+
+  * CI updates to match Microphysics changes (#3021, #3024)
+
+  * remove some old comments related to Fortran (#3022)
+
+  * fix a tiling issue with shock detection on CPUs (#3019)
+
+  * remove unneeded files from problem setups (#3018) and fix comments
+    in problem setup `GNUmakefile`s
+
 # 25.01
 
   * address rotation source in spherical 2d coordinate (#2967)

Docs/source/Introduction.rst

@@ -24,8 +24,8 @@ Castro's major capabilities:
   * adaptive mesh refinement with subcycling; jumps of 2x and 4x
     between levels (see :ref:`ch:amr`)
 
-  * arbitrary equation of state (provided by the companion StarKiller
-    Microphysics project)
+  * arbitrary equation of state (provided by the companion `AMReX-Astro
+    Microphysics <https://github.com/amrex-astro/Microphysics>`_ project)
 
   * general nuclear reaction networks
 
@@ -110,5 +110,3 @@ throughout the code documentation and papers.
 
 Physical constants, again using the CGS system are available
 in ``Microphysics/constants/``.
-
-

Docs/source/build_system.rst

@@ -135,12 +135,8 @@ Microphysics Parameters
 
     .. index:: USE_REACT, general_null, GENERAL_NET_INPUTS, NETWORK_DIR, INTEGRATOR_DIR
 
-  * ``USE_REACT_SPARSE_JACOBIAN``
-
-  * ``USE_SPARSE_STOP_ON_OOB``
-
-  * ``EOS_DIR``: the equation of state to use.  This will be a subdirectory under the
-    Microphysics repo.
+    * ``EOS_DIR``: the equation of state to use.  This will be a subdirectory under the
+      Microphysics repo.
 
     .. index:: EOS_DIR
 
@@ -275,5 +271,3 @@ This is the current build system process.
 
 For all of this to work, we need the ``tmp_build_dir/s`` directory to
 be first in the vpath, so our modified sources are found and used.
-
-

Docs/source/creating_a_problem.rst

@@ -152,6 +152,8 @@ plane-parallel hydrostatic atmosphere, using the hydrostatic boundary
 conditions instead of a simple symmetry boundary is essential when
 using the standard CTU PPM solver.
 
+.. index:: castro.fill_ambient_bc, castro.ambient_fill_dir, castro.ambient_outflow_vel
+
 A different special boundary condition, based on outflow, is available at
 the upper boundary.  This works together with the ``model_parser``
 module to fill the ghost cells at the upper boundary with the initial
@@ -232,6 +234,8 @@ each of these in the main source tree.
 Model Parser
 ------------
 
+.. index:: USE_MODEL_PARSER, MAX_NPTS_MODEL
+
 Many problem setups begin with a 1-d initial model that is mapped onto
 the grid.  The ``model_parser.H`` provides the functions that read in
 the initial model and map it on the Castro grid.  To enable this, add::
@@ -282,5 +286,3 @@ This fills ``dens`` with the density at the position ``height``.  In
 addition to density, you can specify temperature (``model::itemp``),
 pressure (``model::ipres``), species (indexed from ``model::ispec``),
 or an auxiliary quantity (indexed from ``model::iaux``).
-
-

Docs/source/getting_started.rst

@@ -16,7 +16,7 @@ general requirements to build Castro are:
 
  * A C++17 (or later) compiler (for GCC, we need >= 9.0 for CUDA compilation)
 
- * python (>= 3.7)
+ * python (>= 3.10)
 
  * GNU make (>= 3.82)
 
@@ -26,7 +26,7 @@ For running in parallel, an MPI library is required.  For running on GPUs:
 
 * CUDA 11 or later is required for NVIDIA GPUs
 
-* ROCM 4.5 or later is required for AMD GPUs
+* ROCM 6.3.1 or later is required for AMD GPUs (earlier versions have a register allocation bug)
 
 More information on parallel builds is given in section
 :ref:`ch:mpiplusx`.

Exec/Make.Castro

@@ -201,17 +201,6 @@ ifeq ($(USE_REACT), TRUE)
   DEFINES += -DREACTIONS
 endif
 
-ifeq ($(USE_REACT_SPARSE_JACOBIAN), TRUE)
-  DEFINES += -DREACT_SPARSE_JACOBIAN
-
-  # The following is sometimes useful to turn on for debugging sparse J indices
-  # (if a get/set/scale is called with (row, col) not in the sparse J, stop)
-  # Otherwise, set/scale do nothing, and get returns 0.
-  ifeq ($(USE_SPARSE_STOP_ON_OOB), TRUE)
-    DEFINES += -DSPARSE_STOP_ON_OOB
-  endif
-endif
-
 ifeq ($(USE_SIMPLIFIED_SDC), TRUE)
   DEFINES += -DSDC_EVOLVE_ENERGY
 endif

Exec/science/Detonation/GNUmakefile

@@ -18,7 +18,7 @@ EOS_DIR     := helmholtz
 
 # This sets the network directory in $(MICROPHYSICS_HOME)/networks
 ifeq ($(USE_NSE_NET), TRUE)
-	NETWORK_DIR := subch_base
+	NETWORK_DIR := he-burn/he-burn-18a
 	SCREEN_METHOD := chabrier1998
 else
 	NETWORK_DIR := aprox19

Exec/science/nova/GNUmakefile

@@ -1,7 +1,7 @@
 PRECISION        = DOUBLE
 PROFILE          = FALSE
 DEBUG            = FALSE
-DIM              = 2
+DIM              = 3
 
 COMP	         = gnu
 
@@ -10,7 +10,7 @@ USE_OMP          = FALSE
 USE_CUDA         = FALSE
 
 USE_REACT        = TRUE
-USE_DIFFUSION    = TRUE
+USE_DIFFUSION    = FALSE
 USE_GRAV         = TRUE
 
 CASTRO_HOME := ../../..
@@ -22,7 +22,7 @@ MAX_NPTS_MODEL = 20000
 EOS_DIR     := helmholtz
 
 # This sets the network directory in $(MICROPHYSICS_HOME)/networks
-NETWORK_DIR := nova2
+NETWORK_DIR := nova
 
 # Thi sets the conductivity EOS directory in $(MICROPHYSICS_HOME)/conductivity
 CONDUCTIVITY_DIR := stellar