grandi@kpno.UUCP (11/29/83)
Since net.astro.expert has just magically appeared this morning, I
thought I would try to get an interesting discussion going. Given
that the medium of the discussion is a computer network, my topic may
seem rather obvious -- the state of computing in professional
astronomy. Let me pose two questions and make a few comments about
each one--
Should a national astronomy supercomputing center be set up?
Supercomputing seems to have become a "hot" topic lately, and the NSF
is responding by holding back some of its money (which caused KPNO's budget
to be cut slightly) for such purposes. What priority should such a
program have compared to the dessimination of VAX class computers (for both
data reduction/analysis and theoretical work) to university
departments and research groups? I'm inclined to support more VAXes,
but I am no doubt biased being both an observer and a KPNO computer bureaucrat.
Are astronomy graduate students getting a proper grounding in
software engineering? In my previous incarnation as an astronomy
professor at UCLA, I was sometimes almost shocked at the awful programming
practices used by some of the graduate students. Should we make it
our business to teach students about coding standards, structured
programming and software tools or do we "let them pick it up on their
own" much as we expect many to pick up electronics or machine shop
skills? Have you ever had to use software on an instrument that was
written by a new grad student who was shown an 11/23, handed a FORTH
writeup and told "go to it!"? If we do not teach software skills,
how about numerical techniques? use of the Fourier Transform?
I hope these questions get the discussion off to a good start--
steve grandi, kitt peak national observatory, 950 n. cherry ave, tucson, az
(602) 325-9228
{arizona,decvax,hao,ihnp4,seismo,unc,utastronomy,ut-sally} !kpno!grandiwls@astrovax.UUCP (William L. Sebok) (11/29/83)
> Since net.astro.expert has just magically appeared this morning ... For me it didn't just just magically appear but took more than a month of work to bring it into existence. Whew... Now it is up to the rest of you. Enjoy. > Should a national astronomy supercomputing center be set up? > ..... What priority should such a > program have compared to the dessimination of VAX class computers (for both > data reduction/analysis and theoretical work) to university > departments and research groups? ... I definitely think that Vax class machines should be available wherever they are needed. Beyond that I am not sure. I know of a number of projects under- way here that could use heavy computing power. There are many hydrodynamic codes and such running that could easily go to a finer grid. This poor VAX 750 knows what it is like to constantly have several of these programs constantly running in the background. > Are astronomy graduate students getting a proper grounding in > software engineering? ... Should we make it > our business to teach students about coding standards, structured > programming and software tools or do we "let them pick it up on their > own" much as we expect many to pick up electronics or machine shop > skills? I could have used some training on electronics or machine shop. I guess one always has the problem that instruction time is limited and adding more emphasis on one thing means taking away emphasis on something else. I definitely think that undergraduates should be required to take one or more computer courses. Some times I feel that a few professors could use it too. -- Bill Sebok Princeton University, Astrophysics {allegra,akgua,burl,cbosgd,decvax,ihnp4,kpno,princeton}!astrovax!wls
woods@hao.UUCP (Greg Woods) (11/29/83)
First of all, from my understanding of these two groups, this topic does
not belong in the expert subgroup, which was intended for technical
discussions *directly* related to astronomy. This is sort of a peripheral
topic (albeit an important one!), and so belongs in the more general group
net.astro . I have posted this article to both groups in an attempt to move
the discussion to where I think it belongs.
Woods' first law (forgive me if someone has already claimed this) of
scientific computing is, whatever computing power is available will eventually
become saturated, i.e. it is never enough. We have two VAX 11/750's, a PDP
11/70, and a share of two CRAY-1As, and we still manage to just about
saturate everything (well, the second VAX isn't yet ready for general use;
it is being used to bring up 4.2BSD). Of course, we have to share the CRAYs
with the rest of NCAR, including the large 3-D cloud models. We are also
responsible for a good share of the SMM (Solar Maximum Mission) data analysis,
a share which is likely to increase after the scheduled repair of the SMM
satellite by the space shuttle astronauts (I think it is STS-13, but I'm not
positive). [This is one reason we just bought the two new VAXen!]. I'm willing
to bet that any scientific institution reading this would report a similar
situation at their facility.
GREG
--
{ucbvax!hplabs | allegra!nbires | decvax!brl-bmd | harpo!seismo | ihnp4!kpno}
!hao!woodsriddle@ut-sally.UUCP (Prentiss Riddle) (12/06/83)
Hmm. And I was under the impression that out-of-work astronomy PhD's were usually highly employable as software engineers and the like.