Re: [Maxima-discuss] complete_square

[Barton W. <wi...@un...>]

A simple example:


(%i22) larry(x) := block([a],
  a[1] : x)$

(%i23) buddy(x) := block([a],
  local(a),
  a[1] : x)$

(%i24) a[2] : 2015;
(%o24) 2015

(%i25) buddy(0);
(%o25) 0


OK:

    (%i26) a[1];
    (%o26) a[1]

    (%i27) a[2];
    (%o27) 2015


Let's try Larry:

    (%i28) larry(0);
    (%o28) 0


Oops--be careful:

    (%i29) a[1];
    (%o29) 0

    (%i30) a[2];
    (%o30) 2015


Also, I  think the translator doesn't handle local correctly--maybe this is reported?


(%i1) buddy(x) := block([a],
  local(a),
  a[1] : x)$
(%i2) translate(buddy);
(%o2) [buddy]
(%i3) a[2] : 2015;
(%o3) 2015
(%i4) buddy(0);
(%o4) 0


--Barton
________________________________
From: Mike Valenzuela <mic...@gm...>
Sent: Monday, June 1, 2015 16:19
To: Barton Willis
Cc: Luigi Marino; max...@li...
Subject: Re: [Maxima-discuss] complete_square

Oh thank you Barton Willis,

I thought block([whatever], ...) made whatever local. I never knew about local. Thank you.

On Mon, Jun 1, 2015 at 1:50 PM, Barton Willis <wi...@un...<mailto:wi...@un...>> wrote:

  1.  Experiment with replacing ev(b_vec[i,1], map(lambda([avar], avar=0), vars)) with  subst(map(lambda([avar], avar=0), vars), b_vec[i,1]) .
The subst function has clean semantics, but ev doesn't.

  2.  Likely b_vec and c_mat should be declared local; see for example, Robert Dodier<http://search.gmane.org/?author=Robert+Dodier&sort=date> | 19 Jun 16:50 2008 on
http://blog.gmane.org/gmane.comp.mathematics.maxima.general/month=20080601/page=36


--Barton

> complete_square_alt_form:0$
> complete_square(expr, vars):=block([a_const:expr, b_vec, c_mat, c_mat_inv, new_const, offset, x],
 ...