corrections/review of "nchomolog.lib?"

[Oleksandr]

1. lots global procedures without help (some of them are even in the header)

2. lots of theory explanations are missing,
Thus, it's not clear whether everything is same as in commutative case
or there are some differences...

Moreover, it seems that commutative analogous procedures give
somewhat different results in commutative tests,
(e.g. take a look at the results in end of example for ncExt_R)

Please show that they are the same or fix something...

3. lib' name is too long - please rename or it may be worth to merge
this lib into the commutative "homolog.lib" if there are not much
differences (see 1.)

4. there are very few comments concerning algorithms not to mention
step-by-step algorithm explanations

Due to the 1. absence of theory explanations and step-by-step
algorithms comments,  2. usually different results in commutative
tests (compared to analogous from homolog.lib)
it means to me that this LIB is still experimental...

ps: as usual my comments/corrections/suggestions are in the attachment (TODO + uppercase).

[Oleksandr]

my corrections / suggestions for "homolog.lib"

[dreyer]

From singular-team:

Hi,

for this lib more work has to be done and hence I postpone its entrance into release!

Singular wrote:

#56: review of "homolog.lib"

---

Reporter: motsak | Owner: levandov Type: task | Status: new Priority: major | Milestone: Release 3-1-0

Component: singular-libs | Version: Keywords: | ---------------------------+------------------------------------------------

1. lots global procedures without help (some of them are even in the
header)

ok, will be fixed.

2. lots of theory explanations are missing, Thus, it's not clear whether everything is same as in commutative case or there are some differences...

Some combinatorics like maps for Hom and tensors are the same. They appear under a bit different names for the pedagogical reason, since this lib was used twice in teaching - once by Gerhard at ICTP and once by me at Santiago-de-Compostella. Hence, for compatibility with the lecture notes I propose to keep these procedures as they are.

Most real module stuff is VERY different. Even a sneak peek on ncExt_R etc. shows that we work in a very different manner. I think merging of nchomolog.lib with homolog.lib requires too much time & work, and I propose this lib to be autonomous.

Moreover, it seems that commutative analogous procedures give somewhat different results in commutative tests, (e.g. take a look at the results in end of example for ncExt_R)

Please show that they are the same or fix something...

ok, to be fixed.

3. lib' name is too long - please rename or it may be worth to merge this lib into the commutative "homolog.lib" if there are not much differences (see 1.)

rename - yes. are "nchom.lib", "nchomo.lib", "nchomalg.lib" better than "nchomolog.lib"? merge - no, see above.

4. there are very few comments concerning algorithms not to mention step-by-step algorithm explanations

ARE WE OBLIGED TO COMMENT ON EVERY STEP? this is new to me. which Singular libs feature this at the moment? of course some explanations will be added.

Due to the 1. absence of theory explanations and step-by-step algorithms comments, 2. usually different results in commutative tests (compared to analogous from homolog.lib) it means to me that this LIB is still experimental... }}}

agree.

ps: as usual my comments/corrections/suggestions are in the attachment (TODO + uppercase).

Thanks for analyzing the lib. We will go back to the discussion in 2009. With best regards, Viktor

[Oleksandr]

VL:

nchomolog (name contest is still open, "nchomalg" sounds good) DO NOT enter the current release.