Fabian Bernhard

Note that if the ODE doesn't return `0.`, the callback at `t=0.` seems to work. ```Julia # With callback at 0.0, but f not returning 0.0, it solves without error...

Thanks, @ChrisRackauckas . Glad to hear that you are aware of the cause of this behaviour. Indeed, `dt=1e-3` solves the issue, and should be perfectly fine in my use case....

Thanks for this helpful answer, Jonnie! The first part solves my problem perfectly well. For the second part regarding multidimensional `ComponentMatrix`: in your example the second axis is simply `FlatAxis()`....

The underlying code in the package `LWFBrook90.jl` simplifies to the following MWE: ```julia using ComponentArrays # version: [b0b7db55] ComponentArrays v0.15.2 u0 = ComponentArray( GWAT = (mm = 0.0, d18O =...

Actually there might be two ways to fix: - either set `mins` and `maxs` based on `plot_vecs` as mentioned at bottom of first post - or refrain from setting `xlims`...

Regarding the two possible fixes: I would rather attempt to remove the C-layer than to require a user to use GCC 12. - This is supported by: the fact that...

Sounds good. (I mean your strategy sounds good. Random outputs sound rather bad.)

> I found the reason for why multiple consecutive runs yield slightly different outputs. In a lot of places in the code, some 'memory' variables are used. These variables persist...

Can we close this?

After updating to Big Sur, following message was presented to me.  To prevent the issues described by @zixuan96 I renamed the following file as...