Re: Octave on Reddit

[Jaroslav Hajek]

2010/9/29 Jake <address@hidden>:
> I wouldn't be too concerned about the Python related comments; these are just 
> Python fanboys bragging about how much better Python is than every other 
> language and piece of software on the planet.  On a personal note, I'm pretty 
> sure all their comments about speed are moot based on the fact that NumPy's 
> sqrt function is twice as slow as using Python's built in ** exponent 
> operator.
>
> Back on topic though, this is kind of what I was talking about when I 
> mentioned the lack of (built-in) odexx functions.  Though there is lsode, I 
> think the typical person who starts with MATLAB gives Octave a chance until 
> they notice that a few essential functions are missing.  I'm not talking 
> about MATLAB's toolboxes necessarily, but rather a few spoils that MATLAB 
> provides while Octave does not.
>

This is going to improve somewhat in Octave 3.4. When you attempt to
use, e.g. ode23, Octave will present you with a sophisticated warning
telling you about lsode being the native solver and ode23 provided by
OctaveForge. In general, however, it is naive to expect that Octave
will provide the very same algorithms and solvers like Matlab. Yes,
many people are naive, but what can we do?

> I was particularly surprised recently that Octave does not have LTI functions 
> built-in (ss, tf, step, impulse, etc.)
>

The control functions were removed intentionally, because it was
simpler to maintain them outside the core. I also think the legacy
control package is becoming deprecated in favor of something new.


-- 
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz