[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
norm problem
From: |
Ludger Leushacke |
Subject: |
norm problem |
Date: |
Mon, 21 Jun 93 13:13:01 GMT |
Hello again !
Thanks for your prompt and accurate answer to my help request.
Some comments:
:: My problem is that Octave first calls the functions and
:: collects all text for screen output and displays it after all
:: calculations have been finished.
:There are two possible solutions. You can turn paging off completely
:by setting the variable page_screen_output to the string "false", or
:you can force the output buffer to be flushed to the pager at any time
:by using the command fflush (stdout). For an example of this, look at
:the menu function (it's implemented as an M-file).
:Let me know if this isn't what you want, or it doesn't work quite the
:way you expect.
This is exactly what I meant. The simplest solution is to set
page_screen_output="false".
:: - The function norm(A,"fro") seems to work not correctly in
:: every case
:Can you send a specific case that fails?
If A is a scalar or vector, octave gives
error: failed to convert `fro' to a numeric type -- default
conversion turned off
In principle this might be correct, since the F-norm of a scalar or
vector makes no sense. But from the definition of the F-norm
sqrt(sum(diag(A'*A))) it should work and really does. MATLAB
reports no error (I guess "fro" is ignored for scalars and vectors).
I think you want to avoid overhead in your functions, but a simple
check of input dimensions could help.
A last question: What is the search hirarchy for commands in octave?
(internal command, octave library m-file, m-file in current directory, ...)
Thank you very much again
Ludger
-------------------------------------------------------------------------
Dr. Ludger Leushacke
Research Establishment for Applied Science (FGAN)
Research Institute for High Frequency Physics (FHP)
Neuenahrer Str. 20
D-53343 Wachtberg-Werthhoven, Germany
phone: 049-228-852-261
e-mail: address@hidden
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- norm problem,
Ludger Leushacke <=