Mein Vater wäre morgen 72 geworden. Ich habe mich schon lange daran gewöhnt, daß er nicht da ist, dennoch habe ich ruhige Momente, wo ich mir wünschte ich könnte ihm erzählen was ich so mache.
Er würde es verstehen.
Er sagte mir einmal "Ich hätte es geschafft dich in die Medizin zu bugsieren, wenn ich dort eine Zukunft gesehen hätte." Ich glaube er sah im ganzen medizinischen Komplex keine Zukunft, mit Sicherheit auch im finanziellen Aspekt. Oder er sah einfach keine Perspektive darin, für mich. Auf alle Fälle stand er meinen mathematischen Ambitionen sehr aufgeschlossen gegenüber. Er kaufte mir das Buch das ich immer noch sehr verehre, "A fizika kultúrtörténete" (die Kulturgeschichte der Physik). Ich habe es vor kurzem meinem Sohn auf englisch gekauft.
Das aus meinen Ambitionen nicht wirklich viel wurde, lag wahrscheinlich daran, daß ich in die Informatik gegangen bin ;-)
Aber egal, es macht mir ja viel Spaß, und die beiden Gebiete verzahnen sich ja zusehends. Es gibt einen signifikanten Trend, die Grundlagen der Mathematik auf das Fundament der Typentheorie zu stellen, und somit die Mengenlehre praktisch zu entthrohnen, nach mehr als 100 Jahren Herrschaft.
In einem kleinen Winkel dieses Kriegsschauplatzes mache ich mir selber gedanken. Gestern Abend fuhr ich nach Ansbach, um meine Mutter zu besuchen, und nahm auch Hamster Daniela mit. Ich war wirklich müde geworden nach dem Abendessen, las noch ein Paar Seiten Kategorientheorie und schlief ein. Aber zwischen 2 und 3 wachte ich auf, und ich denke in dem Moment verstand ich, daß meine Idee von Typen als Flächen sich mit den Opetopen vereinbaren lassen, wenn man sich nur in die Kodimension versetzt. Das Stichwort Poincaré-Dualität kommt da auf, was alles erklären könnte.
Ich nahm den Hamster mit runter, wir setzten uns vor die Terrassentür, und der Schein des Vollmondes fiel auf uns. Es war mystisch.
Außerdem arbeite ich an einer Kodierung, so daß etwas gleichwertiges zum Lambda-Kalkül mit Opetopen kodiert werden kann. Mal schauen wie das alles zusammenkommt.
Ich denke mein Vater wäre der einzige aus meinem Umkreis der für solche Gedanken ein wirklich offenes Ohr gehabt hätte, selbst wenn er das meiste gar nicht verstanden hätte. Aber er wäre auf mich manchmal stolz gewesen.
Tuesday, August 20, 2013
Monday, August 19, 2013
Compiling GHC on RHEL6
Compared to building GHC on RHEL5 this is a breeze.
I built
But... it still did not build :-(
As always, Stack Overflow for the win!
Specifying
did the trick.
For good measure I also added
to the configure line, but this might be redundant?
(Another thought, find seems to reveal some
So, finally,
The part that involves building was the pleasure part:
Then I dared running the test suite:
This step takes its time, as it is not parallelizable.
There is only one drop of bitterness, namely the
nothing unexpected here.
Have fun with GHC on RHEL6!
git
is in the installation — checkperl boot
— check./configure
— breaks, linker does not find libgmp.so
I built
gmp
from sources, and installed it into my home directory.But... it still did not build :-(
As always, Stack Overflow for the win!
Specifying
setenv LIBRARY_PATH $HOME/lib
did the trick.
For good measure I also added
--with-gmp-libraries=
$HOME/lib
--with-gmp-includes=
$HOME
/include
to the configure line, but this might be redundant?
(Another thought, find seems to reveal some
libgmp.so
under /usr
, I might point it there.)So, finally,
./configure ...
— checkThe part that involves building was the pleasure part:
nice make -j14
— checkThen I dared running the test suite:
make test
— checkThis step takes its time, as it is not parallelizable.
There is only one drop of bitterness, namely the
stage2
compiler does not run on RHEL5. But that is something I did not expect, actually.make install
— checknothing unexpected here.
Have fun with GHC on RHEL6!
Saturday, August 10, 2013
Proudly presenting the »nopetope«!
(This is is jotted down, raw posting, that may never get finished. I am publishing it anyway, as it pretty much reflects my current mood.)
I am about to do some research with coverings of trees, and it was only natural to look at Baez-Dolan metatrees and the corresponding notion of opetopes. The paper contains a famous 5-minute definition and it is a brain teaser worth reading. They introduce trees and nestings (actually two sides of the same coin) and their superposition, called a constellation, which has to follow some rules, but drawing spheres is a creative process.
Then there are zooms which connect constellations as long as the left nesting and the right tree are morally identical.
My lambda graphs are basically search trees and a nesting would add the operational notion of evaluation. Since we can freely choose our evaluation strategy (confluence?), the latter corresponds to a nesting. It remains to find out what the degenerate zooms are under this aspect.
I am just reading the "Polynomial functors and opetopes" paper (http://arxiv.org/pdf/0706.1033.pdf) and it asserts that it's starting constellation is a one leafed tree:
But I wonder if this is fundamental, and since leaves are stripped by zooms, any number will do.
So I'll suggest starting with zero leaves, and calling the resulting zoom complex(es) the nopetopes. This might be the first mathematical term I have coined :-)
For the mathematically-challenged, a nopetope is just a lollipop wrapped in cellophane, while an opetope is the two-stick version thereof.
It is a funny coincidence that "one" and "ope-" contain the same vocals. Going on, we could also have twopetopes and thropetopes, fouretopes etc. But I doubt these are significant in any way.
And now back to the paper and then to an Ωmega implementation of nopetopes...
PS: while writing this my imagination went though... Are trees and nestings compatible with the famous correspondences
energy-mass, wave-particle of physics? Looks like I have to start some more research.
I am about to do some research with coverings of trees, and it was only natural to look at Baez-Dolan metatrees and the corresponding notion of opetopes. The paper contains a famous 5-minute definition and it is a brain teaser worth reading. They introduce trees and nestings (actually two sides of the same coin) and their superposition, called a constellation, which has to follow some rules, but drawing spheres is a creative process.
Then there are zooms which connect constellations as long as the left nesting and the right tree are morally identical.
My lambda graphs are basically search trees and a nesting would add the operational notion of evaluation. Since we can freely choose our evaluation strategy (confluence?), the latter corresponds to a nesting. It remains to find out what the degenerate zooms are under this aspect.
I am just reading the "Polynomial functors and opetopes" paper (http://arxiv.org/pdf/0706.1033.pdf) and it asserts that it's starting constellation is a one leafed tree:
Subscribe to:
Posts (Atom)