[Users] Meeting Minutes 2021-01-07

Thu Jan 7 13:33:00 CST 2021

Present: Maria BH, Bill G, Roland H, Ken S, Yosef Z, David B, Atul K,
Peter D, Zach E, Steve B, Frank L

Chair Maria, Minutes Bill

* SPEC benchmark contribution 
[https://bitbucket.org/einsteintoolkit/tickets/issues/2470 task] [RH, SB]

RH, SPEC group replied and happy to have our contribution.
Same person assigned to us that worked with Erik last time.  They want a
different code than last time.  RH, could use the newest version of
Maclachlan does derivs in steps, does not do them in one go, or different
formulation of BSSN (CZZ4, Z4C), or not use it at all, but use Baikal
code.  Which is fine they want un-optimized code to test the
compilers optimization.  SPEC runs the inner kernel, not a full run.
PD, Maclachlan is mostly Erik.  Whatever is easiest?  ZE,
Baikal, we could do something.  Maybe more future directed sense;
Baikal is based on NRPy and designed to output Einstein Eqns on any
coord system.  When Black Holes at Home is off the ground, had in
mind submitting to SPEC, solving Einstein eqns in curvilinear coordinates.
SPEC is only performed on desktops and laptop computers.

RH, deadline for fully done is February (25 Feb 2021 PST time) and
needs to be a production code.  Would like it to the ET "sticker" on
it.  Alternate forms of BSSN Z4 with constrained damping in various
ways and two flavors to do that, one maybe Daniella Alic at AEI
(https://arxiv.org/abs/1106.2254 ) and other David Hildich
(https://doi.org/10.1103/PhysRevD.81.084003 and
https://arxiv.org/abs/1107.5539 ).  CZZ4 is in Maclachlan and is the
more formal; Z4C is there but is a branch on the ET.  They are very
similar to each other.  Could use that and version of Kranc that
generates the derivatives in two steps.

Non-vectorized Machlachlan code, which may or may not be there. PD, is
using the non-vectorized version of Machlachlan and it is working.

Must strip away ancillary code and look at the "kernel," the one or two
routines to do the updates.  There is a mini-driver that exists.

Roland will make a list and some notes and will discuss/decide next week.

* search for ET release team [RH]

RH, next release is end of May.  Need one or two persons that will be
release chair.  Last few times it was Steve or Roland.  Do have a list
of the tasks involved...not a lot of work but not negligible. Zach E
volunteers as the release chair and Bill G will be a "junior lieutenant."
Keep looking at and refining the "check-list,"

See Release Details: 
https://docs.einsteintoolkit.org/et-docs/Release_Details
and Release Process: 
https://docs.einsteintoolkit.org/et-docs/Release_Process

* simfactory usage survey, 
[https://docs.google.com/forms/d/e/1FAIpQLSf3RzB_o8Y8fIdwZOPoWJyKp9jvV2tIvoml7z2bO0lyOPfvUA/viewform?usp=sf_link 
draft] [RH]

RH, promoting Simfactory as the major tool to use.  But we get a lot
of questions on the mailing list.  Also it is complicated.  And the 
original
maintainers are gone.  To simplify doing something with it, we want to
know how are people using it and which parts are used?  Maybe gives
an idea of which parts can be cut.  RH and SB in the process of porting
to Python 3.

This group should comment on the survey form.  Is a bit long and does
not cover all topics.

* BBH gallery inclusion of LIGO data [RH], see 
[http://lists.einsteintoolkit.org/pipermail/users/2020-September/007590.html 
Minutes for 2020-09-03], the proposed 
[https://docs.einsteintoolkit.org/et-docs/images/d/d8/ET_2020_05_strain-NR-HA.pdf 
figure]

Peter Schaffarcyzk has a new figure for the GW150914 gallery example.
When Peter did the 2019 testing he overplotted the LIGO data on the
ET calculated strain.  We agreed to do this but questions about details
of plot.

Ask Peter S: is the Hanford the best data?  Is it whitened, if so it should
say so?  Power B 2020 needs to be run longer, green tail around t=0.44.
X-axis just label "time" (though it is in units of seconds and not mass?)?

MBH, Roland will contact Peter and ask about getting the scripts.

* Summer Student projects (ad hoc discussion)

Maria and Zach, good to have projects for students to take on.
ZE, Undergraduates ask about summer projects and Google Summer
of Code.  Frank L, You need a sponsor, and continued acceptance
can be fickle.  Have to take students and make sure the work is
done on time.  SB, needs to be a list of undergraduate projects;
should post a list.  RH, There is a list on the Wiki, but not all are
suitable for a summer student.  More computer science-y than
physics-y, easier or more concrete tasks.

ET Wiki list at 
https://docs.einsteintoolkit.org/et-docs/Summer_student_projects .

* call for projects to be included in the next release [RH]

** Simfactory Python3 support [SB, RH]

SB, have a port that is mostly working.  Does the things he wants
to do.  There are features in the survey that Steve has never used...some
features were never working.  One snarly issue with Python3, has a
tendency to require your language environment setup correctly or
will die in funny ways.  Less robust/user friendly than Python2, it seems.

** POWER 
[https://bitbucket.org/einsteintoolkit/tickets/issues/2413/include-power-code-to-extrapolate 
#2413] [RH, MH]

RH, no update but can be cleaned up and have in a state ready for
review by end of the month.  SB, is it alternative to Mathematica
script?  RH, that is part of it.  Does a couple of methods for evolving
Psi4 to Scri-I+.  Toward having more things as pure Python and not
Mathematica.  Helps on clusters.  Has been converted to Python3
and works.  SB, might want something for the trajectory and then
the gallery example plots can be done without Mathematica.  YZ,
what requires Mathematica?  RH, The gallery example has example
Mathematica notebooks, but likely all could be moved to non-Mathematica.

** con2prim updates for GRHydro [https://arxiv.org/abs/1712.07538 paper] 
[https://bitbucket.org/einsteintoolkit/tickets/issues/2414/include-con2prim-framework-by-siegel 
#2414] [RH]

RH, topic seems a misnomer.  Update method in GRHydro, Dan Siegel
and students at Columbia have done this.  It turns out that Arizona
is working on this, but working with Illinois GRMHD and not sure how
close that is to the one in ET.  If they are willing to contribute their
code would like to have it.  Need to hear from the authors if they want
and will contribute it to ET.  Roland will invite one of them to the
meeting to give us a status report.

** updates to Selfforce1D [PD]

PD, no update.

** Canuda updates and new features 
[https://bitbucket.org/einsteintoolkit/tickets/issues/2485/include-complex-and-real-scalar-evolution 
#2485] [MZ, HW]

RH, Miguel Zilhao and Helvi Witek, they have extra functionality for
complex and scalar fields, seems ready and used for publication. Helvi
is busy with a proposal until the end of January.  There is a code there
and we need / would like that someone wants to use it for their research.
Seems the closest to ready for the next ET release.

[https://www.einsteintoolkit.org/tools/unanswered.php unanswered 
question on mailing list]

None.  Script tagged the Dec 17th minutes, not an unanswered post.

[https://bitbucket.org/einsteintoolkit/tickets/issues?status=open&status=new&sort=-updated_on 
open tickets sorted by update time]

#2486, the gallery example uses Z-symmetry but allows the 3
components of the spin when only the z-component can be specified.
Would like someone other than Roland to look at it and fix it. Need a
volunteer!

Ticket #2486, https://bitbucket.org/einsteintoolkit/tickets/issues/2486/

[https://bitbucket.org/einsteintoolkit/tickets/issues?status=new&status=open&sort=-updated_on&q=Please%20review 
tickets ready for review]

* AOB

* New chair and minute taker:

Jan 14-chair Peter D, minutes Maria BH

Jan 21-

* Maria BH, was trying to run at WVU and was given a warning..."no open 
file
bricks connection scheme reported" (?) and the simulation ran fine.  ZE, 
low
level network file structure that MPI is interfacing with.  Not a 
problem, an
error.

* Zach E, update on LaTeX Parser, almost finished BSSN and starting to
transition from Parser to working on GPU kernel generation with NRPy.
Ken is writing example CUDA codes and then will code NRPy to do that.
Everything parsed (BSSN) within a week or two.

-- 
==============================
William Gabella
Research Assistant Professor
Department of Physics and Astronomy
Nashville, TN USA
(o) 615-343-2713