update integrator interface

docs/src/basics/integrator.md

@@ -127,44 +127,68 @@ for safe modifications of the integrator type, and allows for uniform usage
 throughout the ecosystem (for packages/algorithms which implement the functions).
 The following functions make up the interface:
 
-* `u_modified!(integrator,bool)`: Bool which states whether a change to `u` occurred,
-  allowing the solver to handle the discontinuity.
+### Saving Controls
+
 * `savevalues!(integrator)`: Adds the current state to the `sol`.
-* `modify_proposed_dt(integrator,factor)`:  Multiplies the proposed `dt` for the
-  next timestep by the scaling `factor`.
-* `proposed_dt(integrator)`: Returns the `dt` of the proposed step.
+
+### Cache Iterators
+
 * `user_cache(integrator)`: Returns an iterator over the user-facing cache arrays.
 * `u_cache(integrator)`:  Returns an iterator over the cache arrays for `u` in the method.
   This can be used to change internal values as needed.
 * `du_cache(integrator)`:  Returns an iterator over the cache arrays for rate quantities the method.
   This can be used to change internal values as needed.
 * `full_cache(integrator)`:  Returns an iterator over the cache arrays of the method.
   This can be used to change internal values as needed.
+
+### Stepping Controls
+
+* `u_modified!(integrator,bool)`: Bool which states whether a change to `u` occurred,
+  allowing the solver to handle the discontinuity.
+* `modify_proposed_dt(integrator,factor)`:  Multiplies the proposed `dt` for the
+  next timestep by the scaling `factor`.
+* `proposed_dt(integrator)`: Returns the `dt` of the proposed step.
+* `terminate!(integrator)`: Terminates the integrator by emptying `tstops`. This
+  can be used in events and callbacks to immediately end the solution process.
+* `change_t_via_interpolation(integrator,t,modify_save_endpoint=Val{false})`: This
+  option lets one modify the current `t` and changes all of the corresponding
+  values using the local interpolation. If the current solution has already
+  been saved, one can provide the optional value `modify_save_endpoint` to also
+  modify the endpoint of `sol` in the same manner.
+
+### Resizing
+
 * `resize!(integrator,k)`: Resizes the DE to a size `k`. This chops off the end
   of the array, or adds blank values at the end, depending on whether `k>length(integrator.u)`.
 * `resize_non_user_cache!(integrator,k)`: Resizes the non-user facing caches to be
   compatible with a DE of size `k`. This includes resizing Jacobian caches. Note
   that in many cases, `resize!` simple resizes `user_cache` variables and then
   calls this function. This finer control is required for some `AbstractArray`
   operations.
-* `terminate!(integrator)`: Terminates the integrator by emptying `tstops`. This
-  can be used in events and callbacks to immediately end the solution process.
-* `deleteat!(integrator,k)`: Shrinks the ODE by deleting the `i`th component.
+* `deleteat_non_user_cache!(integrator,idxs)`: `deleteat!`s the non-user facing caches
+  at indices `idxs`. This includes resizing Jacobian caches. Note
+  that in many cases, `deleteat!` simple `deleteat!`s `user_cache` variables and then
+  calls this function. This finer control is required for some `AbstractArray`
+  operations.
+* `addat_non_user_cache!(integrator,idxs)`: `addat!`s the non-user facing caches
+  at indices `idxs`. This includes resizing Jacobian caches. Note
+  that in many cases, `addat!` simple `addat!`s `user_cache` variables and then
+  calls this function. This finer control is required for some `AbstractArray`
+  operations.
+* `deleteat!(integrator,idxs)`: Shrinks the ODE by deleting the `idxs` components.
+* `addat!(integrator,idxs)`: Grows the ODE by adding the `idxs` components.
+  Must be contiguous indices.
+
+### Misc
+
 * `get_du(integrator)`: Returns the derivative at `t`.
-* `change_t_via_interpolation(integrator,t,modify_save_endpoint=Val{false})`: This
-  option lets one modify the current `t` and changes all of the corresponding
-  values using the local interpolation. If the current solution has already
-  been saved, one can provide the optional value `modify_save_endpoint` to also
-  modify the endpoint of `sol` in the same manner.
+
+#### Note
 
 Note that not all of these functions will be implemented for every algorithm.
 Some have hard limitations. For example, Sundials.jl cannot resize problems.
 When a function is not limited, an error will be thrown.
 
-#### Note
-
-Currently, many of these functions are not implemented.
-
 ## Additional Options
 
 The following options can additionally be specified in `init` (or be mutated in

docs/src/tutorials/discrete_stochastic_example.md

@@ -55,15 +55,15 @@ Reaction(rate_constant,reactants,stoichiometry)
 
 The first value is the rate constant. We will use `1e-4`. Secondly, we pass in the
 indices for the reactants. In this case, since it uses the susceptible and infected
-persons, the indices are `[2,3]`. Lastly, we detail the stoichometric changes. These
+persons, the indices are `[1,2]`. Lastly, we detail the stoichometric changes. These
 are tuples `(i,j)` where `i` is the reactant and `j` is the number to change by.
 Thus `(1,-1)` means "decrease the number of susceptible persons by 1" and
 `(2,1)` means "increase the number of infected persons by 1".
 
 Therefore, in total, our reaction is:
 
 ```julia
-r1 = Reaction(1e-4,[2,3],[(1,-1),(2,1)])
+r1 = Reaction(1e-4,[1,2],[(1,-1),(2,1)])
 ```
 
 To finish the model, we define one more reaction. Over time, infected people become