diff --git a/doc/concepts/benchmark_cell.rst b/doc/concepts/benchmark_cell.rst
index b3f3075002b00154dea7c50efc8194c8267e5cf7..d443e180e2b4dcee89f9b15c233411a1d556a804 100644
--- a/doc/concepts/benchmark_cell.rst
+++ b/doc/concepts/benchmark_cell.rst
@@ -10,4 +10,3 @@ API
---
* :ref:`Python <pybenchcell>`
-* :ref:`C++ <cppbenchcell>`
diff --git a/doc/concepts/decor.rst b/doc/concepts/decor.rst
index d1b7f9fac870263dfb1c0cbc7288272e1c2b3c40..5cc973f3aa79dd14b3d7635d872808756e0d014c 100644
--- a/doc/concepts/decor.rst
+++ b/doc/concepts/decor.rst
@@ -125,7 +125,7 @@ specialised on specific regions.
Regions can have density mechanisms defined over their extents.
:ref:`Density mechanisms <mechanisms-density>` are a kind of
-:ref:`NMODL mechanism <nmodl>` which describe biophysical processes.
+:ref:`NMODL mechanism <formatnmodl>` which describe biophysical processes.
These are processes that are distributed in space, but whose behaviour is
defined purely by the state of the cell and the process at any given point.
@@ -305,7 +305,7 @@ pile-up effects similar to reflective bounds.
The diffusive concentration is *separate* from the internal concentration for
reasons innate to the cable model, which require resetting it to its starting
point at every timestep. It can be accessed from NMODL density and point
-mechanisms as an independent quantity, see :ref:`NMODL mechanism <nmodl>`. It is
+mechanisms as an independent quantity, see :ref:`NMODL mechanism <formatnmodl>`. It is
present on the full morphology if its associated diffusivity is set to a
non-zero value on any subset of the morphology, ie ``region``. It is initialised
to the value of the internal concentration at time zero.
@@ -328,7 +328,7 @@ locset.
Similar to how regions can have density mechanisms defined over their extents,
locsets can have point mechanisms placed on their individual locations.
-:ref:`Point mechanisms <mechanisms-point>` are a kind of :ref:`NMODL mechanism <nmodl>`
+:ref:`Point mechanisms <mechanisms-point>` are a kind of :ref:`NMODL mechanism <formatnmodl>`
which describe synaptic processes such as the ``expsyn`` mechanism provided by
NEURON and Arbor, which models an exponential synapse.
@@ -384,7 +384,7 @@ See also :term:`threshold detector`.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Locsets can also have junction mechanisms placed on their individual locations.
-:ref:`Junction mechanisms <mechanisms-junction>` are a kind of :ref:`NMODL mechanism <nmodl>`
+:ref:`Junction mechanisms <mechanisms-junction>` are a kind of :ref:`NMODL mechanism <formatnmodl>`
which describe gap-junction processes such as the ``gj`` mechanism provided by Arbor,
which models a basic, linear, constant-conductance based gap-junction.
@@ -452,5 +452,3 @@ API
---
* :ref:`Python <pycablecell-decor>`
-* :ref:`C++ <cppcablecell-decor>`
-
diff --git a/doc/concepts/interconnectivity.rst b/doc/concepts/interconnectivity.rst
index 68cfa7d75f4c21ffd0966108131efc62eab4aa2a..d4dd8d98fd8585067498811d36e4d3c96332a1b1 100644
--- a/doc/concepts/interconnectivity.rst
+++ b/doc/concepts/interconnectivity.rst
@@ -108,7 +108,7 @@ full recipe.
threshold detector
:ref:`Placed <cablecell-place>` on a cell. Possible source of a connection.
Detects crossing of a fixed threshold and generates corresponding events.
- Also used to :ref:`record spikes <>` for analysis. See :ref:`here
+ Also used to record spikes for analysis. See :ref:`here
<cablecell-threshold-detectors>` for more information.
spike source cell
diff --git a/doc/concepts/mechanisms.rst b/doc/concepts/mechanisms.rst
index d3125971be46d744b1a4bdffbf45d7818f4d520f..42518b71cebaeb0d6848da2cd8725fe7aaa68c9f 100644
--- a/doc/concepts/mechanisms.rst
+++ b/doc/concepts/mechanisms.rst
@@ -6,7 +6,7 @@ Cable cell mechanisms
Mechanisms describe biophysical processes such as ion channels, synapses and gap-junctions.
Mechanisms are assigned to regions and locations on a cell morphology
through the process of :ref:`decoration <cablecell-decoration>`.
-Mechanisms are described using a dialect of the :ref:`NMODL <nmodl>` domain
+Mechanisms are described using a dialect of the :ref:`NMODL <formatnmodl>` domain
specific language that is similarly used in `NEURON <https://neuron.yale.edu/neuron/>`_.
Arbor supports mechanism descriptions using the NMODL language through our ``modcc``
@@ -321,4 +321,3 @@ API
---
* :ref:`Python <py_mechanisms>`
-* :ref:`C++ <cpp_mechanisms>`
diff --git a/doc/contrib/test.rst b/doc/contrib/test.rst
index f04eb6bc187f20f9eddcfcbb4fc6970a207d109e..7ccfa474a60aef89413115e2ae1807367a213e0d 100644
--- a/doc/contrib/test.rst
+++ b/doc/contrib/test.rst
@@ -32,16 +32,19 @@ that you decided to add your own radix sort algorithm, since you expect it to be
faster for this particular usecase.
Thus, the tests added should be
+
1. sorting algorithm
- the application of `sort` sorts the given array. This seems trivial, but is
really the core of what you are doing!
- corner cases like: empty array, all elements equal, ... are treated greacefully
- if the sort is intended to be stable, check that equal elements do not switch order
+
2. local sorting
- spikes are -- after sorting -- ordered by their source id and in case of ties by time
- corner cases: NaN, negative numbers, ...
+
3. global sorting
- after MPI exchange, each sub-array is still sorted
diff --git a/doc/cpp/morphology.rst b/doc/cpp/morphology.rst
index ba0c2473429b20facd555b411cca30eb819281c7..8800e02a905d317286051c6ff1f7860ffcdd9801 100644
--- a/doc/cpp/morphology.rst
+++ b/doc/cpp/morphology.rst
@@ -10,7 +10,7 @@ file formats; see :ref:`cppcablecell-morphology-construction` for details.
Segment tree
------------
-A ``segment_tree` is -- as the name implies -- a set of segments arranged in a
+A ``segment_tree`` is -- as the name implies -- a set of segments arranged in a
tree structure, ie each segment has exactly one parent and no child is the
parent of any of its ancestors. The tree starts at a *root* segment which has no
parent. Each segment comprises two points in 3d space together with the radii at
diff --git a/doc/dev/cable_cell.rst b/doc/dev/cable_cell.rst
index 248c0e8d2b3a5d5e55fb6bdc3bfd71750df1ae14..945b50788a3e619a8a2e5144672bef2452c12a2e 100644
--- a/doc/dev/cable_cell.rst
+++ b/doc/dev/cable_cell.rst
@@ -103,11 +103,11 @@ axonal/synaptic delays. This works since we know that an event at time :math:`t`
can have an effect at time :math:`t + T_{min}` at the soonest. The factor of one
half stems from double-buffering to overlap communication and computation. So,
Arbor collects all events in an epoch and transmits them in bulk, see
-:ref:`Communication <mpi_comm>` for details.
+:ref:`Communication <communication>` for details.
Integration in Arbor is then split into three parts:
-1. apply effect of events to mechanisms :ref:`Event Handling <events>`
+1. apply effect of events to mechanisms :ref:`Event Handling <event_distribution>`
2. evolve mechanisms and apply currents :ref:`Mechanisms <mechanisms>`
3. solve voltage equations, see :ref:`Solver <matrix_solver>`
diff --git a/doc/dev/communication.rst b/doc/dev/communication.rst
index 5ac1c5f3f25fcee21129a2181ad94106d4f09a85..b283e9db23e5716a80d76776f6ba1a9ce929192e 100644
--- a/doc/dev/communication.rst
+++ b/doc/dev/communication.rst
@@ -103,6 +103,8 @@ each sub-array of spikes is sorted, ie between ``offsets[ix]`` and
and its ``i``'th element gives the position of the first spike sent by
task ``i``.
+.. _event_distribution:
+
Distribution of Events to Targets
=================================
diff --git a/doc/dev/matrix_solver.rst b/doc/dev/matrix_solver.rst
index 5733f1b27393e74dd443dc15f210762b7b2a57e8..92b0d0a8452aaafa1cd33b3abaaa43c45e99e9ba 100644
--- a/doc/dev/matrix_solver.rst
+++ b/doc/dev/matrix_solver.rst
@@ -3,6 +3,8 @@
Matrix Solvers
==============
+.. _cable_equation:
+
Cable Equation
--------------
diff --git a/doc/install/build_install.rst b/doc/install/build_install.rst
index a7486fbe823db55a01f1aa3ba70b15acc1121940..6584dc27265a5adf7374e374c4c1a2ea52d871b8 100644
--- a/doc/install/build_install.rst
+++ b/doc/install/build_install.rst
@@ -556,7 +556,7 @@ component ``neuroml``. The corresponding CMake library target is ``arbor::arbori
# ...
target_link_libraries(myapp arbor::arborio)
-.. install-profiling:
+.. _install-profiling:
Profiling
---------
diff --git a/doc/python/decor.rst b/doc/python/decor.rst
index bf7fe413f5653362d78c712e9d25e722e9d0c5eb..c405b3d9c7f8ab5f1e661511e6f83cb609e6237f 100644
--- a/doc/python/decor.rst
+++ b/doc/python/decor.rst
@@ -169,6 +169,7 @@ Cable cell decoration
:type policy: :py:class:`cv_policy`
.. method:: discretization(policy)
+ :noindex:
Set the cv_policy used to discretise the cell into control volumes for simulation.
diff --git a/doc/python/hardware.rst b/doc/python/hardware.rst
index 76969f735963de95d28c806008967d290c2ea7de..af51f6b2b1d5f6ef2af7a9372f905840dcece4e4 100644
--- a/doc/python/hardware.rst
+++ b/doc/python/hardware.rst
@@ -51,14 +51,14 @@ Helper functions for checking cmake or environment variables, as well as configu
.. class:: mpi_comm
- .. function:: mpi_comm()
+ .. method:: mpi_comm()
By default sets MPI_COMM_WORLD as communicator.
- .. function:: mpi_comm(object)
+ .. method:: mpi_comm(object)
:noindex:
- Converts a Python object to an MPI Communicator.
+ :param object: Converts a Python object to an MPI Communicator.
.. function:: mpi_finalize()
@@ -67,6 +67,7 @@ Helper functions for checking cmake or environment variables, as well as configu
.. function:: mpi_is_finalized()
Check if MPI is finalized.
+ :rtype: bool
Env: Helper functions
---------------------
@@ -97,13 +98,15 @@ The Python wrapper provides an API for:
Enumerates the computational resources on a node to be used for a simulation,
specifically the number of threads and identifier of a GPU if available.
- .. function:: proc_allocation([threads=1, gpu_id=None])
+ .. method:: proc_allocation([threads=1, gpu_id=None])
- Constructor that sets the number of :attr:`threads` and the id :attr:`gpu_id` of the available GPU.
+ :param int threads: Number of threads.
+ :param int gpu_id: Device ID.
.. attribute:: threads
The number of CPU threads available, 1 by default. Must be set to 1 at minimum.
+
.. attribute:: gpu_id
The identifier of the GPU to use.
@@ -115,7 +118,7 @@ The Python wrapper provides an API for:
See ``cudaSetDevice`` and ``cudaDeviceGetAttribute`` provided by the
`CUDA API <https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__DEVICE.html>`_.
- .. cpp:function:: has_gpu()
+ .. method:: has_gpu()
Indicates whether a GPU is selected (i.e., whether :attr:`gpu_id` is ``None``).
@@ -145,43 +148,37 @@ The Python wrapper provides an API for:
provided by :class:`context`, instead they configure contexts, which are passed to
Arbor interfaces for domain decomposition and simulation.
- .. function:: context(threads, gpu_id, mpi)
- :noindex:
-
- Create a context that uses a set number of :attr:`threads` and gpu identifier :attr:`gpu_id` and MPI communicator :attr:`mpi` for distributed calculation.
-
- .. attribute:: threads
-
- The number of threads available locally for execution. Must be set to 1 at minimum. 1 by default.
- Passing ``"avail_threads"`` (as string) will query and use the maximum number of threads the system makes available.
-
- .. attribute:: gpu_id
-
- The identifier of the GPU to use, ``None`` by default.
- Must be ``None``, or a non-negative integer.
+ .. method:: context(threads, gpu_id, mpi)
+
+ Create a distributed context.
+
+ :param int threads:
+ The number of threads available locally for execution.
+ Must be set to 1 at minimum. 1 by default.
+ Passing ``"avail_threads"`` (as string) will query and use the maximum number
+ of threads the system makes available.
+ :param int gpu_id:
+ The non-negative identifier of the GPU to use, ``None`` by default.
Can only be set when Arbor was built with GPU support.
-
- .. attribute:: mpi
-
- The MPI communicator (see :class:`mpi_comm`), ``None`` by default.
- Must be ``None``, or an MPI communicator.
+ :type gpu_id: int or None
+ :param mpi:
+ The MPI communicator, ``None`` by default for distributed calculation.
Can only be set when Arbor was built with MPI support.
+ :type mpi: :py:class:`arbor.mpi_comm` or None.
.. function:: context(alloc, mpi)
:noindex:
- Create a distributed context, that uses the local resources described by :class:`proc_allocation`, and
- uses the MPI communicator for distributed calculation.
-
- .. attribute:: alloc
+ Create a distributed context.
+ :param alloc:
The computational resources, one thread and no GPU by default.
-
- .. attribute:: mpi
-
- The MPI communicator (see :class:`mpi_comm`). ``None`` by default.
- mpi must be ``None``, or an MPI communicator.
+ :type alloc: :py:class:`proc_allocation`
+ :param mpi:
+ The MPI communicator, ``None`` by default for distributed calculation.
Can only be set when Arbor was built with MPI support.
+ :type mpi: :py:class:`arbor.mpi_comm` or None.
+
Contexts can be queried for information about which features a context has enabled,
whether it has a GPU, how many threads are in its thread pool.
diff --git a/doc/python/labels.rst b/doc/python/labels.rst
index 86e7cad33a126ce405c2731377626fd317bd8b62..f85c0b45786432580122d3b8da2727bc9f7fff43 100644
--- a/doc/python/labels.rst
+++ b/doc/python/labels.rst
@@ -14,6 +14,7 @@ Cable cell labels
Create an empty label dictionary
.. method:: label_dict(dictionary)
+ :noindex:
Initialize a label dictionary from a ``dictionary`` with string labels as keys,
and corresponding string definitions as values.
diff --git a/doc/python/mechanisms.rst b/doc/python/mechanisms.rst
index 97b9312cc94f34d150add05f905823e9ccb66252..9634a0dd33a15b9ea804b713560b943afb0ee016 100644
--- a/doc/python/mechanisms.rst
+++ b/doc/python/mechanisms.rst
@@ -126,7 +126,6 @@ Cable cell mechanisms
The underlying mechanism.
.. method:: density(name)
- :noindex:
constructs :attr:`mech` with *name* and default parameters.
@@ -134,6 +133,7 @@ Cable cell mechanisms
:type name: str
.. method:: density(name, params)
+ :noindex:
constructs :attr:`mech` with *name* and range parameter overrides *params*.
for example: ``arbor.density('pas', {'g': 0.01})``.
@@ -152,6 +152,7 @@ Cable cell mechanisms
:type mech: :py:class:`mechanism`
.. method:: density(mech, params)
+ :noindex:
constructs :attr:`mech` from *mech* and sets the range parameter overrides *params*.
@@ -171,7 +172,6 @@ Cable cell mechanisms
The underlying mechanism.
.. method:: synapse(name)
- :noindex:
constructs :attr:`mech` with *name* and default parameters.
@@ -179,6 +179,7 @@ Cable cell mechanisms
:type name: str
.. method:: synapse(name, params)
+ :noindex:
constructs :attr:`mech` with *name* and range parameter overrides *params*.
for example: ``arbor.synapse('expsyn', {'tau': 0.01})``.
@@ -197,6 +198,7 @@ Cable cell mechanisms
:type mech: :py:class:`mechanism`
.. method:: synapse(mech, params)
+ :noindex:
constructs :attr:`mech` from *mech* and sets the range parameter overrides *params*.
@@ -217,7 +219,6 @@ Cable cell mechanisms
The underlying mechanism.
.. method:: junction(name)
- :noindex:
constructs :attr:`mech` with *name* and default parameters.
@@ -225,6 +226,7 @@ Cable cell mechanisms
:type name: str
.. method:: junction(name, params)
+ :noindex:
constructs :attr:`mech` with *name* and range parameter overrides *params*.
for example: ``arbor.junction('gj', {'g': 2})``.
@@ -243,6 +245,7 @@ Cable cell mechanisms
:type mech: :py:class:`mechanism`
.. method:: junction(mech, params)
+ :noindex:
constructs :attr:`mech` from *mech* and sets the range parameter overrides *params*.
diff --git a/doc/python/probe_sample.rst b/doc/python/probe_sample.rst
index a6625298a04668f2da9a10d3c52365b3e4f993c9..da12e17a5245b9738c3624c2e4971a39145c5ddd 100644
--- a/doc/python/probe_sample.rst
+++ b/doc/python/probe_sample.rst
@@ -4,6 +4,7 @@ Cable cell probing and sampling
===============================
.. module:: arbor
+ :noindex:
.. figure:: probe_sample-diag.svg
:width: 800
diff --git a/doc/tutorial/network_ring_gpu.rst b/doc/tutorial/network_ring_gpu.rst
index 1f649ac7758973bd47195c15e2fce6861ddcc1d5..834a6fb805d8436c64460b64778803ceebb5164c 100644
--- a/doc/tutorial/network_ring_gpu.rst
+++ b/doc/tutorial/network_ring_gpu.rst
@@ -1,4 +1,4 @@
-.. _tutorialmpi:
+.. _tutorialgpu:
GPU and profiling
=================
diff --git a/doc/tutorial/single_cell_detailed_recipe.rst b/doc/tutorial/single_cell_detailed_recipe.rst
index 66f694ac31f7028211677d5dd85e3d17a5cd7f51..7391f69a0538ae5cf129efad64f34050c22d8bd4 100644
--- a/doc/tutorial/single_cell_detailed_recipe.rst
+++ b/doc/tutorial/single_cell_detailed_recipe.rst
@@ -105,12 +105,12 @@ This was handled by the :class:`arbor.single_cell_model` object in the original
We would like to get a list of the spikes on the cell during the runtime of the simulation, and we would like
to plot the voltage registered by the probe on the "custom_terminal" locset.
-The lines handling probe sampling warrant a second look. First, we declared :term:`probe_id` to be a
+The lines handling probe sampling warrant a second look. First, we declared ``probeset_id`` to be a
:class:`arbor.cell_member`, with :class:`arbor.cell_member.gid` = 0 and :class:`arbor.cell_member.index` = 0.
This variable serves as a global identifier of a probe on a cell, namely the first declared probe on the
cell with ``gid = 0``, which is id of the only probe we created on the only cell in the model.
-Next, we instructed the simulation to sample :term:`probe_id` at a frequency of 50 kHz. That function returns a
+Next, we instructed the simulation to sample ``probeset_id`` at a frequency of 50 kHz. That function returns a
:term:`handle` which we will use to :ref:`extract the results <pycablecell-probesample>` of the sampling after running the simulation.
We can now run the simulation we just instantiated for a duration of 100 ms with a time step of 0.025 ms.