# Changing the righthand side in a callback

**URL:** <https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462>\
**Category:** Modelling & Simulations\
**Tags:** diffeq\
**Created:** [August 12, 2019, 3:17pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462 "2019-08-12T15:17:53Z")\
**Posts on this page:** 7\
**Page:** 1

<div class="post-metadata">

**Author:** ![FHell](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/fhell/32/2421_2.png) [@FHell](https://discourse.julialang.org/u/FHell)\
**Post date:** [August 12, 2019, 3:17pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462/1 "2019-08-12T15:17:53Z")

</div>

Hi everyone,  
I am wondering how to correctly change the right hand side inside a DifferentialEquations callback.  
If I don’t reinit the system runs through but gives wrong results (compared to just running new differential equations). If I reinit the integrator I get an infinite loop.  
Any pointers?

```julia
mutable struct ode_switch
    original_ode
    fault_ode
    fault::Bool
end

function (pgs::ode_switch)(u, t, integrator)
    println("time $t") # To make sure the call back is really called
    if ! pgs.fault
        integrator.f = rhs(pgs.fault_ode)
        integrator.u = find_valid_ic(integrator.f, integrator.u) # This takes care of the DAE/Massmatrix aspect
        # reinit!(integrator, t0=t) # If I use reinit, I get an infinite loop with the callback being called over and over.
        # step!(integrator) # I tried to do a step in order for that not to happen but it didn't help
        pgs.fault = true
    else
        integrator.f = rhs(pgs.original_ode)
        integrator.u = find_valid_ic(integrator.f, integrator.u) # This takes care of the DAE/Massmatrix aspect
        # reinit!(integrator, t0=t) # If I use reinit, I get an infinite loop with the callback being called over and over.
        # step!(integrator) # I tried to do a step in order for that not to happen but it didn't help
        pgs.fault = false
    end
end

timespan = (0., 10.)
faulttime = (1., 2.)
pgs = ode_switch(func1,func2)
cb = FunctionCallingCallback(pgs; funcat=faultime, func_start = false)
prob = ODEProblem(func1, x0, timespan, callback=cb)
solve(prob, Rodas4(autodiff=false))

```

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [August 12, 2019, 3:50pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462/2 "2019-08-12T15:50:32Z")

</div>

I would just use a `DiscerteCallback` to change a parameter that is the boolean in `p`, and then check for that in `f`. By making it integrator-local it won’t have re-use issues.

---

<div class="post-metadata">

**Author:** ![FHell](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/fhell/32/2421_2.png) [@FHell](https://discourse.julialang.org/u/FHell)\
**Post date:** [August 12, 2019, 4:00pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462/3 "2019-08-12T16:00:56Z")

</div>

Thanks for the quick reply, I saw that as a design in the other answers. In this case it’s a bit of an awkward fit because I typically will get an ODEFunction and a fault function that changes the ODEFunction from the user. So it’s a bit of effort to build a wrapper that switches between them, thus I thought I’d try it this way.

---

<div class="post-metadata">

**Author:** ![FHell](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/fhell/32/2421_2.png) [@FHell](https://discourse.julialang.org/u/FHell)\
**Post date:** [August 12, 2019, 4:07pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462/4 "2019-08-12T16:07:38Z")

</div>

For example, the different functions (func1, func2) could have different mass matrices, that could not be handled by a change of parameters, correct?

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [August 12, 2019, 4:08pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462/5 "2019-08-12T16:08:39Z")

</div>

That probably could be, since it can be a DiffEqOperator with an update\_func, but that is not well documented…

---

<div class="post-metadata">

**Author:** ![FHell](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/fhell/32/2421_2.png) [@FHell](https://discourse.julialang.org/u/FHell)\
**Post date:** [August 12, 2019, 4:25pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462/6 "2019-08-12T16:25:32Z")

</div>

In that case I think we will stick to calling solve several times with initial conditions taken from the last simulation.

Is there a documented way to combine several solutions into one? (or the next solve on top of the existing solution object)?

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [August 18, 2019, 6:31pm UTC](https://discourse.julialang.org/t/changing-the-righthand-side-in-a-callback/27462/7 "2019-08-18T18:31:31Z")

</div>

There’s no documented way to do that. Instead of doing that, why not use the integrator interface?

[http://docs.juliadiffeq.org/latest/basics/integrator.html](http://docs.juliadiffeq.org/latest/basics/integrator.html)