comphy-lab · pull · Mar 9, 2026 · Mar 7, 2026 · Mar 7, 2026 · Mar 7, 2026
diff --git a/cmake/Modules/Packages/GPU.cmake b/cmake/Modules/Packages/GPU.cmake
@@ -146,11 +146,18 @@ if(GPU_API STREQUAL "CUDA")
       if(CUDA_VERSION VERSION_GREATER_EQUAL "11.1")
         string(APPEND GPU_CUDA_GENCODE " -gencode arch=compute_86,code=[sm_86,compute_86]")
       endif()
-      # Lovelace (GPU Arch 8.9) is supported by CUDA 11.8 and later
+      # Ada Lovelace (GPU Arch 8.9) and Hopper (GPU Arch 9.0) are supported by CUDA 11.8 and later
       if(CUDA_VERSION VERSION_GREATER_EQUAL "11.8")
         string(APPEND GPU_CUDA_GENCODE " -gencode arch=compute_90,code=[sm_90,compute_90]")
       endif()
-      # newer GPU Arch versions require CUDA 12.0 or later which is handled above
+      # Backwell (GPU Arch 100) is supported by CUDA 12.4 and later
+      if(CUDA_VERSION VERSION_GREATER_EQUAL "12.4")
+        string(APPEND GPU_CUDA_GENCODE " -gencode arch=compute_100,code=[sm_100,compute_100]")
+      endif()
+      # Rubin (GPU Arch 120) require CUDA 12.8 and later
+      if(CUDA_VERSION VERSION_GREATER_EQUAL "12.8")
+        string(APPEND GPU_CUDA_GENCODE " -gencode arch=compute_120,code=[sm_120,compute_120]")
+      endif()
     endif()
   endif()
 
@@ -331,11 +338,18 @@ elseif(GPU_API STREQUAL "HIP")
       if(CUDA_VERSION VERSION_GREATER_EQUAL "11.1")
         string(APPEND HIP_CUDA_GENCODE " -gencode arch=compute_86,code=[sm_86,compute_86]")
       endif()
-      # Lovelace (GPU Arch 8.9) is supported by CUDA 11.8 and later
+      # Ada Lovelace (GPU Arch 8.9) and Hopper (GPU Arch 9.0) are supported by CUDA 11.8 and later
       if(CUDA_VERSION VERSION_GREATER_EQUAL "11.8")
         string(APPEND HIP_CUDA_GENCODE " -gencode arch=compute_90,code=[sm_90,compute_90]")
       endif()
-      # newer GPU Arch versions require CUDA 12.0 or later which is handled above
+      # Backwell (GPU Arch 100) is supported by CUDA 12.4 and later
+      if(CUDA_VERSION VERSION_GREATER_EQUAL "12.4")
+        string(APPEND HIP_CUDA_GENCODE " -gencode arch=compute_100,code=[sm_100,compute_100]")
+      endif()
+      # Rubin (GPU Arch 120) require CUDA 12.8 and later
+      if(CUDA_VERSION VERSION_GREATER_EQUAL "12.8")
+        string(APPEND HIP_CUDA_GENCODE " -gencode arch=compute_120,code=[sm_120,compute_120]")
+      endif()
     endif()
   endif()
 

diff --git a/doc/src/Build_basics.rst b/doc/src/Build_basics.rst
@@ -137,16 +137,15 @@ to: e.g. ``KSPACE``, and ``COLVARS``.  See the :doc:`Packages details
 <Packages_details>` page for more info on these packages, and the pages
 for their respective commands for OpenMP threading info.
 
-For CMake, if you use ``BUILD_OMP=yes``, you can use these packages
-and turn on their native OpenMP support and turn on their native OpenMP
-support at run time, by setting the ``OMP_NUM_THREADS`` environment
-variable before you launch LAMMPS.
-
-For building via conventional make, the ``CCFLAGS`` and ``LINKFLAGS``
-variables in Makefile.machine need to include the compiler flag that
-enables OpenMP. For the GNU compilers or Clang, it is ``-fopenmp``\ .
-For (recent) Intel compilers, it is ``-qopenmp``\ .  If you are using a
-different compiler, please refer to its documentation.
+For CMake, if you use ``BUILD_OMP=yes``, you can use these packages and
+turn on their native OpenMP support at run time by setting the
+``OMP_NUM_THREADS`` environment variable before you launch LAMMPS.
+
+When building LAMMPS with conventional make, the ``CCFLAGS`` and
+``LINKFLAGS`` variables in Makefile.machine need to include the compiler
+flag that enables OpenMP.  For the GNU compilers or Clang, it is
+``-fopenmp``\ .  For (recent) Intel compilers, it is ``-qopenmp``\ .  If
+you are using a different compiler, please refer to its documentation.
 
 .. _default-none-issues:
 

diff --git a/doc/src/Build_cmake.rst b/doc/src/Build_cmake.rst
@@ -206,14 +206,14 @@ Configuration and build options
 
 The CMake commands have one mandatory argument: a folder containing a
 file called ``CMakeLists.txt`` (for LAMMPS it is located in the
-``cmake`` folder, in that case the current working directory becomes
-the build folder) or a build folder containing a file called
+``cmake`` folder, in that case the current working directory becomes the
+build folder) or a build folder containing a file called
 ``CMakeCache.txt``, which is generated at the end of the CMake
 configuration step.  The cache file contains all current CMake settings.
-This is a "legacy mode" or running CMake and thus often found
-when searching the web.  We recommend to use the ``-S`` and ``-B``
-folders to explicitly set the path to the folder containing the
-``CMakeLists.txt`` file and the build folder, respectively.
+This is a "legacy mode" of running CMake and thus often found when
+searching the web.  We recommend to use the ``-S`` and ``-B`` folders to
+explicitly set the path to the folder containing the ``CMakeLists.txt``
+file and the build folder, respectively.
 
 To modify settings, enable or disable features, you need to set
 *variables* with either the ``-D`` command-line flag (``-D
@@ -252,19 +252,21 @@ of them you can use the command:
 
    cmake --build build --target lmp
 
-Especially, when programming LAMMPS and you are making just local
-changes, or only want to make certain that the code still compiles
-and links, you may want to skip may of the extra steps that are
-run by default and then append "/fast" to the target.  Example:
+Especially, when programming LAMMPS and you are making only local
+changes, or only want to make certain that the code still compiles and
+links, you may want to skip many of the extra steps that are run by
+default and then append "/fast" to the target.  Example:
 
 .. code-block:: bash
 
    cmake --build build --target lammps/fast
 
-Note that this speeds up the build by skipping a lot of checks for
-dependencies, and avoiding to re-run CMake, so you may occasionally need
-to compile without the "/fast" suffix or use the "all" (= default)
-target to update everything.
+Note that this speeds up the build by avoiding to re-run CMake and
+skipping a lot of checks for possible dependencies, so you may
+occasionally need to compile without the "/fast" suffix or use the "all"
+(= default) target to update everything.  This would be *required* when
+adding new files, additional dependencies, or making changes to the
+CMake scripts.
 
 Some custom targets, e.g. "install-python" are explicitly excluded
 from the "all" target and must be built individually.

diff --git a/doc/src/Build_development.rst b/doc/src/Build_development.rst
@@ -55,8 +55,8 @@ during CMake configuration.
    -D ENABLE_IWYU=value    # value = no (default) or yes
 
 This will check if the required binary (include-what-you-use or iwyu)
-and python script script (iwyu-tool or iwyu_tool or iwyu_tool.py) can
-be found in the path.  The analysis can then be started with:
+and python script (iwyu-tool or iwyu_tool or iwyu_tool.py) can be found
+in the path.  The analysis can then be started with:
 
 .. code-block:: bash
 

diff --git a/doc/src/Build_extras.rst b/doc/src/Build_extras.rst
@@ -838,7 +838,7 @@ This list was last updated for version 4.7.1 of the Kokkos library.
 
       .. versionchanged:: 11Feb2026
 
-      The KOKKOS package no longer supports the the traditional make build.
+      The KOKKOS package no longer supports the traditional make build.
       You need to build LAMMPS with CMake.
 
 Advanced KOKKOS compilation settings
@@ -1945,5 +1945,5 @@ your system.
 
       .. versionchanged:: 10Sep2025
 
-      The VTK package no longer supports the the traditional make build.
+      The VTK package no longer supports the traditional make build.
       You need to build LAMMPS with CMake.
diff --git a/doc/src/Build_link.rst b/doc/src/Build_link.rst
@@ -128,7 +128,7 @@ POEMS package installed becomes:
 
       .. code-block:: bash
 
-         gcc -c -O -I${HOME}/lammps/src -caller.c
+         gcc -c -O -I${HOME}/lammps/src caller.c
          g++ -o caller caller.o -L${HOME}/lammps/src -llammps_serial
 
 Locating liblammps.so at runtime
@@ -156,7 +156,7 @@ would add something like this to your ``${HOME}/.profile`` file:
 
 .. code-block:: bash
 
-   LD_LIBRARY_PATH ${LD_LIBRARY_PATH-/usr/lib64}:${HOME}/lammps/src
+   LD_LIBRARY_PATH=${LD_LIBRARY_PATH-/usr/lib64}:${HOME}/lammps/src
    export LD_LIBRARY_PATH
 
 For the csh or tcsh shells, you would equivalently add something like this

diff --git a/doc/src/Build_manual.rst b/doc/src/Build_manual.rst
@@ -34,7 +34,7 @@ the rest the build process will attempt to download and install them into
 a python virtual environment and local folders.
 
 A current version of the manual (latest feature release, that is the state
-of the *release* branch) is is available online at:
+of the *release* branch) is available online at:
 `https://docs.lammps.org/ <https://docs.lammps.org/>`_.
 A version of the manual corresponding to the ongoing development (that is
 the state of the *develop* branch) is available online at:

diff --git a/doc/src/Build_prerequisites.rst b/doc/src/Build_prerequisites.rst
@@ -34,7 +34,7 @@ Compiler and OS compatibility
 
 The LAMMPS developers aim to keep LAMMPS very portable and usable -
 at least in parts - on most operating systems commonly used for
-running MD simulations.  Please see the :doc:`section on portablility
+running MD simulations.  Please see the :doc:`section on portability
 <Intro_portability>` for more details.
 
 .. admonition:: Warning: LLVM based Intel Compilers
@@ -52,10 +52,10 @@ running MD simulations.  Please see the :doc:`section on portablility
 
    Unfortunately there is no simple way to detect whether a binary is
    working correctly outside of running the unit and regression tests,
-   but those do not cover all of LAMMPS and and would be reliable only
-   for no or moderate optimization anyway.  For most of LAMMPS there is
-   not much of a benefit (if any) to use the Intel compilers over the
-   GCC or Clang compilers, except for the INTEL package (which *can* be
+   but those do not cover all of LAMMPS and would be reliable only for
+   no or moderate optimization anyway.  For most of LAMMPS there is not
+   much of a benefit (if any) to use the Intel compilers over the GCC or
+   Clang compilers, except for the INTEL package (which *can* be
    compiled with other compilers, but most vectorization directives are
    inactive for those) or KOKKOS with SYCL.
 

diff --git a/doc/src/Build_settings.rst b/doc/src/Build_settings.rst
@@ -254,8 +254,8 @@ use and parallel communication costs for transposing 3d FFT data.
 When using ``-DFFT_SINGLE`` with FFTW3, you may need to ensure that
 the FFTW3 installation includes support for single-precision.
 
-When compiler FFTW3 from source, you can do the following, which should
-produce the additional libraries ``libfftw3f.a`` and/or ``libfftw3f.so``\ .
+When compiling FFTW3 from source, you can do the following, which should
+produce the additional libraries ``libfftw3f.a`` and ``libfftw3f.so``\ .
 
 .. code-block:: bash
 

diff --git a/doc/src/Commands_all.rst b/doc/src/Commands_all.rst
@@ -124,7 +124,7 @@ An alphabetic list of general LAMMPS commands.
 
 Additional general LAMMPS commands provided by packages.  A few
 commands have accelerated versions.  This is indicated by an
-additional letter in parenthesis: k = KOKKOS.
+additional letter in parentheses: k = KOKKOS.
 
 .. table_from_list::
    :columns: 6

diff --git a/doc/src/Commands_parse.rst b/doc/src/Commands_parse.rst
@@ -159,11 +159,11 @@ LAMMPS:
 
 .. note::
 
-   If the argument is itself a command that requires a quoted
-   argument (e.g. using a :doc:`print <print>` command as part of an
-   :doc:`if <if>` or :doc:`run every <run>` command), then single, double, or
+   If the argument is itself a command that requires a quoted argument
+   (e.g. using a :doc:`print <print>` command as part of an :doc:`if
+   <if>` or :doc:`run every <run>` command), then single, double, or
    triple quotes can be nested in the usual manner.  See the doc pages
-   for those commands for examples.  Only one of level of nesting is
+   for those commands for examples.  Only one level of nesting is
    allowed, but that should be sufficient for most use cases.
 
 .. admonition:: ASCII versus UTF-8

diff --git a/doc/src/Commands_removed.rst b/doc/src/Commands_removed.rst
@@ -30,12 +30,11 @@ ATC, AWPMD, and POEMS packages
 
 .. deprecated:: 10Sep2025
 
-The ATC, AWPMD, and POEMS packages are removed.because there were
+The ATC, AWPMD, and POEMS packages are removed because they were
 unmaintained for a long time and their legacy C++ programming style
-started to create problems with modern C++ compilers.  LAMMPS version
-22 July 2025 is the last version that contains them.  You have to
-download and compile this version, if you want to use any of these
-packages.
+started to create problems with modern C++ compilers.  LAMMPS version 22
+July 2025 is the last version that contains them.  You have to download
+and compile this version, if you want to use any of these packages.
 
 .. toctree::
    :maxdepth: 0

diff --git a/doc/src/Developer_platform.rst b/doc/src/Developer_platform.rst
@@ -80,6 +80,12 @@ to the path as a C++ string class instance.
 .. doxygenfunction:: file_is_readable
    :project: progguide
 
+.. doxygenfunction:: file_is_writable
+   :project: progguide
+
+.. doxygenfunction:: file_redirect
+   :project: progguide
+
 .. doxygenfunction:: is_console
    :project: progguide
 

diff --git a/doc/src/Errors_common.rst b/doc/src/Errors_common.rst
@@ -57,7 +57,7 @@ sometimes be confusing or need additional explanations.
   do, but very strictly applies the syntax as described in the
   documentation.  If you run into errors that LAMMPS does not catch that
   you think it should flag, please send an email to the `developers
-  <https://www.lammps.org/authors.html>`_ or create an new topic on the
+  <https://www.lammps.org/authors.html>`_ or create a new topic on the
   dedicated `MatSci forum section <https://matsci.org/lammps/>`_.
 
 - If you get an error message about an invalid command in your input

diff --git a/doc/src/Errors_debug.rst b/doc/src/Errors_debug.rst
@@ -248,7 +248,7 @@ can have multiple reasons:
   system size, e.g. :doc:`kspace styles like ewald or pppm
   <kspace_style>`. There is very little that can be done in this case.
 - The output interval is not set or set to a large value with the
-  :doc:`thermo <thermo>` command. I the first case, there will be output
+  :doc:`thermo <thermo>` command. In the first case, there will be output
   only at the first and last step.
 - The output is block-buffered and instead of line-buffered. The output
   will only be written to the screen after 4096 or 8192 characters of

diff --git a/doc/src/Errors_details.rst b/doc/src/Errors_details.rst
@@ -31,8 +31,9 @@ Create a small test system
 Debugging problems often requires running a simulation many times with
 small modifications, thus it can be a huge time saver to first assemble
 a small test system input that has the same issue, but will take much
-time until it triggers the error condition.  Also, it will be easier to
-see what happens.
+less time until it triggers the error condition.  Also, it will be
+easier to see what happens when visualizing the system or looking at
+output files.
 
 .. _hint02:
 
@@ -81,15 +82,15 @@ While a segmentation fault is likely an indication of a bug in LAMMPS,
 it need not always be; it can also be the consequence of too aggressive
 simulation settings.  For time critical code paths, LAMMPS will assume
 the user has chosen the settings carefully and will not make any checks
-to avoid to avoid performance penalties.
+to avoid performance penalties.
 
 A crucial step in resolving a segmentation fault is to identify the
-exact location in the code where it happens.  Please see `Errors_debug`
-for a couple of examples showing how to do this on a Linux machine.
-With this information -- a simple way to reproduce the segmentation
-fault and the exact :doc:`LAMMPS version <Manual_version>` and platform
-you are running on -- you can contact the LAMMPS developers or post in
-the LAMMPS forum to get assistance.
+exact location in the code where it happens.  Please see
+:doc:`Errors_debug` for a couple of examples showing how to do this on a
+Linux machine.  With this information -- a simple way to reproduce the
+segmentation fault and the exact :doc:`LAMMPS version <Manual_version>`
+and platform you are running on -- you can contact the LAMMPS developers
+or post in the LAMMPS forum to get assistance.
 
 .. _hint05:
 
@@ -100,7 +101,7 @@ Fast moving atoms may be "lost" or "missing" when their velocity becomes
 so large that they can cross a sub-domain within one timestep.  This
 often happens when atoms are too close, but atoms may also "move" too
 fast from sub-domain to sub-domain if the box changes rapidly.
-E.g. when setting a large an initial box with :doc:`shrink-wrap boundary
+E.g. when setting a large initial box with :doc:`shrink-wrap boundary
 conditions <boundary>` that collapses on the first step (in this case
 the solution is often using 'm' instead of 's' as a boundary condition).
 
@@ -266,7 +267,7 @@ before the box (e.g. a compute or fix reference or a thermo keyword).
 Illegal ... command
 ^^^^^^^^^^^^^^^^^^^
 
-These are a catchall error messages that used to be used a lot in LAMMPS
+These are catchall error messages that used to be used a *lot* in LAMMPS
 (also programmers are sometimes lazy).  They usually include the name of
 the source file and the line where the error happened.  This can be used
 to track down what caused the error (most often some form of syntax error)
@@ -433,7 +434,7 @@ Nose-Hoover (or other) barostat, and thus it may be advisable to run
 with only a thermostat for a bit until the potential energy has
 stabilized.
 
-.. _err007:
+.. _err0007:
 
 Fix used in ... not computed at compatible time
 -----------------------------------------------
@@ -521,7 +522,7 @@ executable.  In rare cases, there may be licensing or portability issues
 that prevent including a package in publicly accessible binaries or in a
 specific environment.
 
-.. _err011:
+.. _err0011:
 
 Energy or stress was not tallied by pair style
 ----------------------------------------------
@@ -570,7 +571,7 @@ For example:
    variable cutoff index 10.0
    pair_style lj/cut ${cutoff}  # this is correct
    pair_style lj/cut $cutoff    # this is incorrect, LAMMPS looks for 'c' instead of 'cutoff'
-   variable c      index 5.0    # if $c is defined, LAMMPS subsitutes only '$c' and reads: 5utoff
+   variable c      index 5.0    # if $c is defined, LAMMPS substitutes only '$c' and reads: 5utoff
 
 Another potential source of this error may be invalid command line
 variables (-var or -v argument) used when launching LAMMPS from an
@@ -908,7 +909,7 @@ the lack of a time-integrating fix is intentional or not.
 System is not charge neutral, net charge = ...
 ----------------------------------------------
 
-the sum of charges in the system is not zero.  When a system is not
+The sum of charges in the system is not zero.  When a system is not
 charge-neutral, methods that evolve/manipulate per-atom charges,
 evaluate Coulomb interactions, evaluate Coulomb forces, or
 evaluate/manipulate other properties relying on per-atom charges may
@@ -1021,7 +1022,7 @@ Error messages ending in 'Please contact the LAMMPS developers'
 Such error messages indicate that something unexpected has happened and
 that it will require a good understanding of the details of the design
 of LAMMPS to resolve this.  This can be due to some bug in contributed
-code, and oversight when updating functionality, a feature that is
+code, an oversight when updating functionality, a feature that is
 scheduled to be removed or reaching a combination of flags and settings
 that should not be possible or similar.