Add docs for time dependent batching

Author: geraintpalmer

docs/Guides/batching.rst

@@ -49,4 +49,41 @@ Note:
      + :code:`TimeDependent`
   + If the keyword :code:`Batching_distributions` is omitted, then no batching is assumed. That is only one customer arrives at a time. Equivalent to :code:`['Deterministic', 1]`.
   + If some nodes/customer classes require no batching, but others do, please use :code:`['Deterministic', 1]`.
-  + Batch arrivals may lead to :ref:`simultaneous events <simultaneous_events>`, please take care.
+  + Batch arrivals may lead to :ref:`simultaneous events <simultaneous_events>`, please take care.
+
+
+---------------------------------
+How to Set Time Dependent Batches
+---------------------------------
+
+Ciw allows batching distributions to be time dependent.
+That is the batch size, the number of customers arriving simultaneously, is sampled from a distribution that varies with time.
+
+Let's show an example, we wish to have batch sizes of 2 for the first 10 time units, but batch sizes of 1 thereafter.
+Define a time dependent batching distribution::
+
+    >>> def time_dependent_batches(t):
+    ...     if t < 10.0:
+    ...         return 2
+    ...     return 1
+
+Now use this when defining a network:
+
+    >>> import ciw
+    >>> N = ciw.create_network(
+    ...     Arrival_distributions=[['Deterministic', 3.0]],
+    ...     Service_distributions=[['Deterministic', 0.5]],
+    ...     Batching_distributions=[['TimeDependent', time_dependent_batches]],
+    ...     Number_of_servers=[1]
+    ... )
+
+We'll simulate this for 16 time units.
+Now at times 3, 6, and 9 we would expect 2 customers arriving (a total of 6).
+And at times 12 and 15 we would expect 1 customer arriving (a total of 2).
+So 8 customers in total should finish service::
+
+    >>> Q = ciw.Simulation(N)
+    >>> Q.simulate_until_max_time(16.0)
+    >>> len(Q.nodes[-1].all_individuals)
+    8
+

docs/Guides/time_dependent.rst

@@ -4,7 +4,7 @@
 How to Define Time Dependent Distributions
 ==========================================
 
-In Ciw we can get a time dependent distribution, that is a service time or inter-arrival time distribution that changes as the simulation time progresses.
+In Ciw we can get a time dependent distribution, that is a service time, inter-arrival time, or batching distribution that changes as the simulation time progresses.
 In order to do this a time dependent function, that returns a sampled time, must be defined.
 This must take in a time variable `t`.
 
@@ -54,8 +54,3 @@ Therefore a total of 42 customers passed through the system::
 
    >>> len(Q.nodes[-1].all_individuals)
    42
-
-Time dependent function can be used for batching distributions too.
-In this case the time dependent function must return the number of individuals that must enter the simulation at the given time.
-Note that the function must return a float number, but the result will then be rounded to the nearest integer.
-The function may return zero, if no individuals are expected to enter the simulation at the given time.