"Zhang, Hong" <[email protected]> writes:

>> On Oct 20, 2017, at 3:15 PM, Jed Brown <[email protected]> wrote:
>> 
>> "Zhang, Hong" <[email protected]> writes:
>> 
>>> Another confusion which is not related to this topic is the usage of the 
>>> word "DAE".
>>> I disagree with the statement "In general, this (the general form) is a 
>>> differential algebraic equation (DAE)" on page 141 of the manual.
>>> The word "DAE" has been abused in the comments of many TS functions 
>>> (including TSComputeIFunction), where it actually should mean "DAE or ODE".
>>> PETSc uses the same interfaces for both DAE and ODE, but it is wrong to 
>>> consider ODE as a special case of DAE.
>>> They are fundamentally different from each other and should be 
>>> distinguished explicitly in the manual and the source code.
>> 
>> Is this not valid usage?
>> 
>>  "Thus, ODEs have index 0."
>
> One can think of scalar as 0d matrix. But how often do people use 0d matrix 
> for scalar?

The elegance of differential geometry would be lost if scalars were not
0-forms.

> When people say DAE, naturally it refers to DAE with index 1 or higher 
> (ODE+algebraic constraints).
> If PETSc TS developers are really DAE zealots, at least it should be 
> specified that their "DAE" includes index 0 DAE.

I'm fine with explicitly stating that ODE are index 0 DAE and having
phrases like "if your DAE is an ODE, the explicit interface can be
used".

Reply via email to