NEWS for the group


To ng-spice-devel@ieee.ing.uniroma1.it
From Paolo Nenzi <pnenzi@ieee.ing.uniroma1.it>
Date Thu, 6 Apr 2000 15:28:30 +0200 (CEST)
cc 09097737@spv.ing.uniroma1.it
Delivered-To mailing list ng-spice-devel@ieee.ing.uniroma1.it
Mailing-List contact ng-spice-devel-help@ieee.ing.uniroma1.it; run by ezmlm
Reply-To ng-spice-devel@ieee.ing.uniroma1.it

Hi all,

Some news for the group.

1) The letter to CAD Group ?
   Can someone send me the draft. Prof. Vincentelli is "waiting it". He
offered us to be our "ambassador" (well, not so officially) with
Berkeley's people. He asked for a small paper describing our work and our
directions to obtain the permission to link our pages to the EECS site,
giving us credit. I think that we should ride the wave.

2) Bug fixing and debauch
    Debauch seems very good. I have tried it (sorry, not yet with Arno's
script) and removed a couple of bugs (as in my previous message). I
encourage other people to use it, it is a very good piece of software, but
be prepared to uge output.There are some bugs that you cannot close simply
by putting a free(), in these cases the program structure should be
altered.

3) Improvements
   In his seminar Sangiovanni ha pointed some ways to improve spice3f:
   a) Sparse package: He stated that the sparse1.3a found on netlib
       repository is better (numerically more robust) than the one
       included in spice sources. We should include this one. 
       A friend of mine here in Rome (Pierpaolo Pierozzi) is working 
       on it.
  
   b) Jacobian: In the calculation of the jacobian, spice3f makes an
      unnecessary matrix/vector multiplication ,where a simple 
       matrix + delta update should be more performant. We should
       investigate in this direction.

   c) LTE: This calculus is buggy. It works because of bugs. The constant
      TRTOL (tuned to 7) conflicts with theory. I have not understood this 
      point, if someone is interested I (or him/her) can write Vincentelli 
      about it. 

   d) Charge:  Spice does not conserve the charge. The problem relies on
               capacitors modeling. This will be a great improvement.   

4) Documentation:
   Search the Vincentelli's home page at Berkeley, it should be
   http://www.eecs.berkeley.ed/~alberto, and download E219A lectures,
   they are very interesting and explain how a circuit simulator
   should be built. 

That's all folks

Paolo


Partial thread listing: