[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: ode csets
From: |
c. |
Subject: |
Re: ode csets |
Date: |
Sat, 22 Oct 2016 21:48:03 +0200 |
On 22 Oct 2016, at 21:42, c. <address@hidden> wrote:
>
> On 19 Oct 2016, at 21:01, Rik <address@hidden> wrote:
>
>> The issue is made worse by the fact that there are no BIST tests for
>> "InitialSlope", "MvPattern", and "MaxOrder" which might have caught this.
>
> Rik,
>
> The options "InitialSlope" and "MaxOrder" cannot be tested either in
> ode45 or in ode23 as they make no sense for either of the two solvers
> and are actually removed completely from the option structure before
> invoking odemergeopts that is responsible of cecking option consistency.
>
> This is consistent with Matlab as also there no consistency check
> is made for options that are not relevant for the specific solver
> being used.
>
> The option "MvPattern", on the other hand, would make sense for the
> explicit Runge-Kutta solvers, but it is also not used right now
> and I would not know how to take advantage of the knowledge of the
> sparsity path either.
>
> I think the comment you added to both solvers
>
> ## FIXME: Missing tests.
> ## test for InitialSlope option is missing
> ## test for MaxOrder option is missing
> ## test for MvPattern option is missing
>
> should be removed.
>
> If you want you can add instead a reminder that these options are not
> tested in ode{45,23}.m as they are ignored and have no effect anyway.
>
> c.
The complete list of options that are treated similarly is
at line 151 of ode45.m:
{"Jacobian", "JPattern", "Vectorized", ...
"MvPattern", "MassSingular", ...
"InitialSlope", "MaxOrder", "BDF"}
c.