CoCoA » CoCoALib

13 Feb 2014

12:02 Bug #413: OrdvArith: use of a single buffer is NOT THREADSAFE

Christof could you make some further tests after modifying slightly the OpenMP version of the @OrdvArith@ code I sent... John Abbott

12 Feb 2014

16:30 Feature #342: Remove denominators: QQ[x] -> ZZ[x] (and PushBack(coeff, PP))

My non-authorative solution is... Winfried Bruns

11 Feb 2014

20:10 Feature #440: Port RealRoots to C++

Make the code prefer "binary rationals" as much as possible; exploit this fact when evaluating the polynomial. In an... John Abbott

20:04 Feature #440 (New): Port RealRoots to C++

Port the @RealRoots@ code to C++ so that it is accessible to CoCoALib users!
I hope that a well-written port might... John Abbott

09 Feb 2014

16:52 Support #438: Polynomial multiplication (product of RingElem)

I am sorry for having raised a non-issue.
Regarding NumTerms: is it a data field of the polynomial or is it compu... Winfried Bruns

16:08 Support #438: Polynomial multiplication (product of RingElem)

Winfried Bruns wrote:
> I think it is a question that needs careful testing before any changes should be done. As ob... Anna Maria Bigatti

12:48 Support #438: Polynomial multiplication (product of RingElem)

I think it is a question that needs careful testing before any changes should be done. As observed by John and mentio... Winfried Bruns

09:50 Support #438 (In Progress): Polynomial multiplication (product of RingElem)

In @SparsePolyRing.C -- myMul@ there is this code... Anna Maria Bigatti

08 Feb 2014

15:40 Support #438: Polynomial multiplication (product of RingElem)

[from an email sent to Bruns] I do recall that there are complications for inserting the terms in the correct order; ... John Abbott

15:33 Support #438 (In Progress): Polynomial multiplication (product of RingElem)

Bruns asks whether @f*g@ and @g*f@ take the same time or whether one way round is better than the other. He mentions... John Abbott

15:18 Feature #437 (In Progress): MemPool: order free blocks?

Computation speed in C5 often degrades over time; I think this may be due to loss of data locality in the RAM. In an... John Abbott

15:14 Feature #142 (In Progress): Improve threadsafety

John Abbott

10:43 Feature #342: Remove denominators: QQ[x] -> ZZ[x] (and PushBack(coeff, PP))

This function seems to be optimal.
What is the authorative function for the converse transformation ZZ[X] --> QQ[X] ? Winfried Bruns

07 Feb 2014

00:47 Feature #436 (New): Improve UPoly eval at BigRat for binary rational

It should be possible to make @EvalUPoly@ faster for binary rationals (_i.e._ whose denom is a power of 2)
John Abbott

04 Feb 2014

19:42 Bug #413: OrdvArith: use of a single buffer is NOT THREADSAFE

On our compute server with 20 threads the performance was about 5% better with the OpenMP adaptation for some medium ... Christof Soeger

03 Feb 2014

18:53 Feature #222: Printing polynomials - spaces between terms

I notice in @SparsePolyRing.C:551@ that there is a check via @IsPrintAtom@. The documentation says _true iff arg doe... John Abbott

18:24 Feature #222: Printing polynomials - spaces between terms

In note 6 I unwittingly overlooked some (important?) points: for instance I did not consider compound coefficients in... John Abbott

17:37 Bug #413: OrdvArith: use of a single buffer is NOT THREADSAFE

CVS
time nmzIntegrate -E CondEffPlurSymm-100
nmzIntegrate -E CondEffPlurSymm-100 218,09s user 0,69s system 385% cpu... Christof Soeger

12:43 Bug #413: OrdvArith: use of a single buffer is NOT THREADSAFE

OK, I'll try to rewrite it -- maybe even this afternoon while invigilating!
John Abbott

11:47 Bug #413: OrdvArith: use of a single buffer is NOT THREADSAFE

I did some tests and the timings were exactly the same. So I checked again. I only have a version which still include... Christof Soeger

11:07 Bug #413: OrdvArith: use of a single buffer is NOT THREADSAFE

Yes, I still have it and will give it a try. Christof Soeger

15:27 Design #432: Semantics of IsPrintedWithMinus

Anna rightly asked how should TwinFloats be handled.
Right now, I'm undecided. I'm quite sure that any polynomial... John Abbott

12:49 Design #432 (In Progress): Semantics of IsPrintedWithMinus

If we follow the suggestion in note 6 of issue #222 then we need to define @IsPrintedWithMinus@ only for numerical ri... John Abbott

01 Feb 2014

10:20 Feature #222: Printing polynomials - spaces between terms

I would prefer a symmetric appearance, either no space around the + sign or a blank on bothsides.
But it is a mat... Winfried Bruns

31 Jan 2014

20:32 Feature #222: Printing polynomials - spaces between terms

This issue has been sitting idle for a year. We should decide, and then implement!
*Addendum:* JAA thinks that a le... John Abbott

20:29 Design #432 (In Progress): Semantics of IsPrintedWithMinus

We should clarify what the semantics of @IsPrintedWithMinus@ are.
I suggest that if @IsPrintedWithMinus(x)@ gives ... John Abbott

20:25 Design #429 (In Progress): factorization: public data fields, or mem fns allowing the fields to be updated.

John and Anna talked about Christof's proposal, and both believe it is a good idea! :-)
John will look at existing... John Abbott

20:17 Bug #413 (Feedback): OrdvArith: use of a single buffer is NOT THREADSAFE

@Christof, I have unclosed the issue because Anna proposed to redo your long computation using impl *(B)*, just to co... John Abbott

19:47 Feature #431 (Resolved): NewPolyRing_DMPI has no constructor with a predefined PPM

Wow I wrote something that actually works?? :-)
Bonus extra 10% for this issue :-)
Now I'll have to write a test,... John Abbott

11:16 Feature #431 (In Progress): NewPolyRing_DMPI has no constructor with a predefined PPM

John implemented it. Works for me. Christof Soeger

30 Jan 2014

18:59 Bug #413 (Closed): OrdvArith: use of a single buffer is NOT THREADSAFE

Several test were successful. I consider it as done. Christof Soeger

17:18 Feature #342: Remove denominators: QQ[x] -> ZZ[x] (and PushBack(coeff, PP))

Anna pointed out that is already a fn *@ClearDenom@* which does almost what is wanted: it multiplies the polynomial b... John Abbott

17:02 Feature #209 (Feedback): ReadExpr: input polynomials in CoCoALib

Official part of CoCoALib now.
Even with example in
example/ex-PolyInput2.C
(still missing: documentation)
Anna Maria Bigatti

14:10 Feature #209 (Resolved): ReadExpr: input polynomials in CoCoALib

Example program is done and working fully:
reads any kind expression with + - * / ^ ()
They key was John Abbott's... Anna Maria Bigatti

16:36 Feature #431 (Closed): NewPolyRing_DMPI has no constructor with a predefined PPM

We would like to do ... Christof Soeger

16:30 Feature #407: RingElem ctor from mpz_t (from Bruns)

To summarize, there are now the following constructors:... Christof Soeger

14:01 Feature #430: Use -fopenmp flag when compiling CoCoALib with libnormaliz

For me it now works fine. With and without libnormaliz. Christof Soeger

10:38 Feature #430: Use -fopenmp flag when compiling CoCoALib with libnormaliz

without --with-libnormaliz it still tries to check and get
nm: '': No such file Christof Soeger

29 Jan 2014

19:50 Design #254 (Resolved): How to return a Hilbert Series in CoCoALib

Now all essentially set.
One thing to decide: factorization of denom is now arbitrary:
Normaliz uses a compact form... Anna Maria Bigatti

17:48 Feature #218 (In Progress): CoCoALib normaliz interface

added HilbertSeries and multiplicity functions Christof Soeger

16:47 Feature #430 (Feedback): Use -fopenmp flag when compiling CoCoALib with libnormaliz

Resolved the outstanding problem -- the flag had to passed correctly to @qmake@.
JAA has fixed the @make-c5makefile.... John Abbott

16:08 Feature #430 (In Progress): Use -fopenmp flag when compiling CoCoALib with libnormaliz

JAA has written a new script, and added a suitable call to @configure@.
Christof confirms that it works for him, exc... John Abbott

15:38 Feature #430 (Closed): Use -fopenmp flag when compiling CoCoALib with libnormaliz

Right now libnormaliz can be included to CoCoALib only without using OpenMP. We did it this way to have an easier sta... Christof Soeger

14:48 Feature #342: Remove denominators: QQ[x] -> ZZ[x] (and PushBack(coeff, PP))

Regarding the conversion @QQ[x]@ to @ZZ[x]@ clearing denominators, JAA thinks the following is a reasonable approach:... John Abbott

13:55 Feature #407 (Feedback): RingElem ctor from mpz_t (from Bruns)

After discussing with Anna and Christof...
* We *reject the idea of using a "marker"* because it is cumbersome, also... John Abbott

12:27 Bug #149: Makefile dependencies

JAA believes the correct approach is to have separate dependencies for *internal* dependencies and for *external* dep... John Abbott

28 Jan 2014

18:18 Design #429 (Closed): factorization: public data fields, or mem fns allowing the fields to be updated.

Christof was surprised to see that the data fields in a @factorization@ are publicly modifiable. He suggested having... John Abbott

18:04 Design #254: How to return a Hilbert Series in CoCoALib

Anna Maria Bigatti wrote:
> Constructors:
> (RingElem num, factorization<RingElem> DenFactors) -- general
> (vecto... Christof Soeger

18:04 Bug #413 (Feedback): OrdvArith: use of a single buffer is NOT THREADSAFE

Somewhat unexpectedly Christof's tests shows that approach *(C)* is as fast as approach *(B)* but it is also simpler ... John Abbott

18:00 Bug #428 (Feedback): PPMonoidOv is not threadsafe

Christof's tests have all gone well.
The change was simple (but a bit tedious/ugly).
I've even checked-in the code ... John Abbott

12:12 Bug #428: PPMonoidOv is not threadsafe

JAA has modified the code to use local buffers if @CoCoA_THREADSAFE_HACK@ has a positive value.
*15:00* Christof say... John Abbott

12:09 Bug #428 (Closed): PPMonoidOv is not threadsafe

@PPMonoidOv@ uses a "global" buffer much as @OrdvArith@ did.
This is not threadsafe -- Christof has observed crashes... John Abbott

17:55 Support #195 (Feedback): OrdvArith documentation needs rewriting

The doc should now be essentially complete.
Also greatly improved doc for @PPOrdering@.
John Abbott

15:06 Support #195: OrdvArith documentation needs rewriting

A first version of the doc has been written! It even includes comments about thread-safety! John Abbott

14:59 Support #234 (Feedback): Update/Improve geobucket documentation

Anna Maria Bigatti

14:48 Feature #233 (Feedback): AsINT and AsRAT -- ConvertTo<BigInt> and ConvertTo<BigRat>

Final decision is for ConvertTo<BigInt> and ConvertTo<BigRat> Anna Maria Bigatti

13:59 Design #427: Error names and error messages (current design)

Many errors now are called *Not...* (e.g. *NotNonZero*) but that's ambiguous/confusing.
Suggestion: rename them *Need... Anna Maria Bigatti

10:39 Design #427 (In Progress): Error names and error messages (current design)

As a sub-issue of re-designing errors here we note the suggested names. Anna Maria Bigatti

12:13 Feature #248 (Feedback): IsDivisible for RingElem with nice interface

All done (I think) Anna Maria Bigatti

27 Jan 2014

16:11 Feature #426 (Closed): ConvertTo with optional arg to give more helpful error message

The *@ConvertTo@* fn is used extensively in the interface between the C5 interpreter and the CoCoALin fns which actua... John Abbott

12:13 Feature #209 (In Progress): ReadExpr: input polynomials in CoCoALib

Anna, Christof & John propose the following for an initial impl:
* input polys are terminated by newline
* the whole ... John Abbott

25 Jan 2014

22:10 Bug #413 (In Progress): OrdvArith: use of a single buffer is NOT THREADSAFE

We should discuss this in Osnabrueck next week.
*(A)* the solution using local buffers in every fn that needs one is... John Abbott

24 Jan 2014

20:40 Feature #313: Elim(vector<long>) as PPOrderingCtor

Anna told me she has an impl (of exactly what I don't recall).
Let's make it available -- I'd like it to be availabl... John Abbott

16 Jan 2014

16:57 Feature #209: ReadExpr: input polynomials in CoCoALib

Would it make sense to discuss this when we go to Osnabrueck? (in 10 days' time)
*Note: remember this KISS philosophy!* John Abbott