Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
overview paper co-author invitation
#11
(06/15/2009, 12:33 AM)Kouznetsov Wrote: I am not successful with the public key. I copypast my intent:
...
D:~/BITBU>ssh-keygen -t public
unknown key type public
--------------------------
Why it does not want to geberate the public key?
What other type of key can be generated instead?
------------------------
The answer is RTFM (Read The Fucking Manual!).
Here an excerpt from "man ssh-keygen":
Code:
-t type
             Specifies the type of key to create.  The possible values are
             ``rsa1'' for protocol version 1 and ``rsa'' or ``dsa'' for proto-
             col version 2.
So perhaps best to use is dsa:
Code:
ssh-keygen -t dsa
It asks you where to store the file and the passphrase to use with that file.
It creates:
id_dsa
id_dsa.pub

Roughly the first file contains your private key (dont give it to anyone), the second file contains your public key (give it to everyone).

The procedure from scratch

You need:
  • Installed ssh
  • Installed python
  • Installed mercurial
  • ssh public key "~/.ssh/id_dsa.pub". You can create one with "ssh-keygen -t dsa".
  • bitbucket account. (You dont need an own repository.)

Then you do at bitbucket
  • Upload your public ssh-key to your bitbucket account: Once logged in, go to Account and paste the contents of ~/.ssh/id_dsa.pub into the corresponding field and add key.
  • Send me an e-mail: Go to Inbox
  • Follow the project (to be notified about changes): be logged in, go to the project, click following.

On your local computer:
Go to some folder where you want the project folder to be created, then do
Code:
hg clone https://username@bitbucket.org/bo198214/bunch/
(replace "username" by your bitbucket username.) This should create the project folder "bunch".

A normal editing cycle would be as follows (the given commands assume you are in the project folder (or subfolders of it)):
  1. Code:
    hg pull
    gets the latest version from the bitbucket repository.
  2. Edit some files in the project folder
  3. Code:
    hg commit
    this stores the new version in the repository *locally on your computer*. You can continue with the previous item, or the next step.
  4. Code:
    hg push
    this pushes your last commit to the bitbucket repository. Continue with the first step.

If you create new files instead of just editing existing ones you have to do an
Code:
hg add <filename>
before commit.

Quote:I have created the folder
http://bitbucket.org/kouznetsov/cauchi-integral/
Can you access this folder?
Should I upload there the latex files, C++ codes, pics?
No own repository is needed.
Reply
#12
I have downloaded mercurial and the bunch/ directory from bo's directory, and have written some of the "intuitive" section. Now I am wondering how I would upload it or "push" it back to the server?
Reply
#13
(07/31/2009, 08:04 AM)andydude Wrote: I have downloaded mercurial and the bunch/ directory from bo's directory, and have written some of the "intuitive" section. Now I am wondering how I would upload it or "push" it back to the server?

Thats great!
Just in this moment I restarted to work on the paper.
My aim is to finish it in August.

Please post/send me your public ssh-key.
I will add it to the repository and a "hg push" should be no problem then (or are there other error-messages?).
Reply
#14
(07/31/2009, 08:43 AM)bo198214 Wrote: Please post/send me your public ssh-key.
I will add it to the repository and a "hg push" should be no problem then (or are there other error-messages?).

I added you as a writer to the repository.
(I see I dont need your public ssh-key.)

Please push as soon as possible as I want to merge my change (and take off that burden from your shoulders Smile )
Reply
#15
I think it worked! Smile Now we can work together hopefully... also, I was wondering if I could change "super function" -> "superfunction" is that OK?
Reply
#16
(08/02/2009, 01:38 AM)andydude Wrote: I think it worked! Smile

Indeed! I merged your changes with my local changes, you can see the branch/merge graph here.

Quote:also, I was wondering if I could change "super function" -> "superfunction" is that OK?

Yes I did it already, previously. I use the following conventions:
superfunction (no hyphen) and super-exponential (with hyphen).
This pattern is extensible e.g. super-factorial = superfunction of the factorial

However our widely used super-logarithm does not fit in this pattern,
thatswhy I followed Dmitriis suggestion: arcsuper-exponential

This is the inverse of the super-exponential and hence what we previously called super-logarithm. arcsuper-exponential means also Abel function of the exponential, so we *dont* introduce an extra arcsuperfunction.

I have the convention to put a TODO at every place where I know it needs to be rewritten. At finishing one simply needs to look that there are no more "TODO"s in the text. In our collaborative case I suggest a name suffix, e.g. TODO[henryk] means that it remains to do for henryk.

If we just repeat a proposition from a book or established theory (to use it in an own derivation) please use the "wellknown" environment. E.g.
Code:
\begin{wellknown}
This is a well known proposition
\end{wellknown}
For common (i.e. really common, not just in the tetration community) or standard definitions, please use e.g.:
Code:
\begin{sdef}
The factorial is defined as ...
\end{sdef}
Please have also a look at the decl.tex file to not reinvent abbreviations, e.g. \mathbb{C} is just \C.

I suggest we dont use the e-mail system of bitbucket, because its errornous. However perhaps we can use the issue tracker, lets see.

@Andrew: Instead of with matrices I would like you to start with the equations, because the matrices are not really necessary to explain the approach. Something roughly like: From the Abel equation ... powerseries ... infinite equation system ... (represented with the Carleman matrix as ...) ambiguity of solutions .... intuitive solution ... etc.
Reply
#17
(08/03/2009, 12:41 AM)andydude Wrote: What's wrong with "super-logarithm"?

Well I thought this is obvious: The prefix "super" for a function f describes a function F with
F(x+1)=f(F(x)). But super logarithm is not a superfunction of the logarithm, but an inverse superfunction of the exponential.

Quote:Hmm... I have no idea how I would describe it without matrices...

Ah I see now, it is because of the composition, for which you need the Carleman matrix. I had in mind to use the direct formula for powerseries composition (which is also in the paper already)



If you apply this to the Abel equation you get an infinite linear equation system in which expresses exactly
. Which is a row vector equation; if you transpose you get the usual column vector equation.

Ya, think about it. I let it to you. The goal should be a motivated and understanding reader Smile

Quote:However, it would be good to generalize a little, because the definition of the Abel matrix doesn't apply to doing intuitive iteration of (x -> a x) because you have to use different truncations for this case.

Of course it does not work out of the box for f(x)=ax because it has a fixed point at 0 (and the method does not work at fixed points).
You have to (linearly) conjugate f such that the fixed point is somewhere else. It is an open question whether the intuitive Abel function is indeed log_a (and we should put that into the paper as conjecture. If it is true it would boost the significance of this method.). I followed this idea here. I merely could (non-publicly) achieve that the solution of the infinite equation system does not depend on the fixed point determined through the conjugation.
Reply
#18
(08/03/2009, 08:37 AM)bo198214 Wrote: Well I thought this is obvious: The prefix "super" for a function f describes a function F with F(x+1)=f(F(x)). But super logarithm is not a superfunction of the logarithm, but an inverse superfunction of the exponential.

You're going in circles! The "super" terminology has always meant: a rank-4 function that is analogous to a rank-3 function. It was by analogy to this usage that you decided to use the "superfunction" terminology. It seems almost like a backronym to change the meaning of "super" to fit the usage in "superfunction", which I do not prefer (remember, I liked "iterational function"). I have a feeling that the push for consistent terminology will leave the corpus of writings on this forum in a state of complete inconsistency. I vote for "superlogarithm" or "Abel function of exponential". No "arcsuper".

(08/03/2009, 08:37 AM)bo198214 Wrote: Ya, think about it. I let it to you. The goal should be a motivated and understanding reader Smile
Oh, I see. Ok, I could definitely use in place of the Carleman matrix, that would certainly make things easier to discuss without inventing the notation. I will give it a try.

Andrew Robbins
Reply
#19
Now that I work more with mercurial I see that the work flow described above is not completely correct for pulling changes from the bitbucket repository.

To keep your local repository consistent with the remote bitbucket repository you have to do the following:

Code:
> hg pull
This pulls changes from the remote repository

> hg merge
This merges these changes into your current repository.

If there are conflicts, i.e. someone else changed the same part of the file that you changed too, you have to resolve conflicts. These are marked with <<<<< and >>>>>> in the file.
If you have done this, do a
> hg resolve -m
> hg commit
After that you can work as usual on your files, i.e. make some changes, make some commits, and if you reached some official state you can push it to the bitbucket repository. And everybody else can pull your changes.
Reply
#20
Now it gets really interesting. 3 people (me, Andrew, Ansus) made commits and pushs. This is my first experience in distributed writing of an article. But it seems to work well and I enjoy that.

As a general rule: make commits often. As this would result in fewer conflicts as it can better track the changes made to the file. Also make a push whenever you finished something.

And remember that you can do a push usually only after you pulled and merged the changes from bitbucket. The graph of the branches and merges can be seen on the repository page at the menu item "Changesets".
Reply


Possibly Related Threads...
Thread Author Replies Views Last Post
  Old Research Paper andydude 3 2,562 01/24/2018, 04:12 AM
Last Post: andydude
  changing terminology (was: overview paper co-author invitation) andydude 49 56,251 11/05/2009, 02:06 AM
Last Post: Base-Acid Tetration
  Paper with iteration intro andydude 4 5,897 05/13/2009, 07:53 AM
Last Post: andydude



Users browsing this thread: 1 Guest(s)