FDTD tips

When using BPML, external wall should be treated as PECs for simplicity no matter how many the maximum sigma is. The best way to do that is setting initial field value on external wall to be zero, and then don't update field value of any point on the wall.

That is to say, it could be written as



Saving a few of computing resource.

 

Injecting one module into another in Python

Here is a simple code snippet to do that. Saving it to an individual file and import it in other module to use it.



Using exec statement is also good, but need to pass globals() or locals() dictionary to the function executing exec statement.

Doing dirty things in Python is really not easy...:/

 

Assign initial value to a pointer

Today I suddenly got an idea. Most of time when we use pointer, we create a variable, and assign initial value to it, and then take its address become the value of a new pointer for succeeding use. But we create a redundant identifier here.

In fact, we could always create a fundamental type pointer and assign initial value to it through dereference operator like this snippet.



and no redundant variable here.

But when we define a simple struct like this.



We should assign initial value through arrow operator like third and fourth line in below snippet. Using brace to assign value as the same as first snippet is wrong syntax.



With a complex struct definition, this way increases many redundant code. Using brace to assignment initial value without creating redundant identifier could be done like this.



But I consider writing (*aPtr) = { 1, 2 }; may be the most intuitive way and it doesn't conflict with other syntax in C.

 

Library for Numerical Computing

Briefly summarize some numerical library here. It looks there are too many useful tools. Either for parallel computing or for matrix, numerical calculating and plotting.

PLOT:

MatPlotLib: it seems to be the most awesome plotting tool among this list

PGPlot: C library, also have binding in Perl, Python,Ruby

PLPlot: cross-platform plot library, also have binding in Perl

Scientific Python

MRPlot


PARALLEL:

MPI Ruby: MPI Ruby binding

PyMPI: MPI integrated Python interpreter

MPI for Python Python MPI binding

Simple Remote Python : SrPy

Parallel Python


CALCULATION:

Perl Data Language: difficult to be categorized. Including calculating, plotting together. include FFTW, PGPlot, PLPlot, and many data processing tool such as HDF. Awesome Environment!

MPMath: Multi-Precision math functions collection.

RNUM

NArray

LibFFTW, also have binding in Perl, Ruby

Ruby DCL

LAPACK

SciPy

NumPy

GNU Scientific Library: C library, also have bindings in Ruby named Ruby/GSL and Ruby-GSL. and Python

RSRuby: bridge between Ruby and R

ARTICLES:

Ruby for Science

 

Discussion about anonymous function in Perl and Ruby

Just for fun, tonight I write a methods invoking two anonymous functions to iterate a Range with specified condition in Ruby. It looks like

class Range
    def each_satisfy()
    end
end

And I want to invoke it as

(1..10).each_satisfy(condition_lambda, callback_lambda)

(1..10).each_satsify { |n| ...condition statment... } do |n|
    ...callback statement...
end

When I try to define its prototype as

def each_satisfy(&condition, &callback)
end

Ruby Interpreter broke with syntax error. So that, finally I write it as

def each_satisfy(condition, &callback)
    self.each do |n|
        yield n if condition.call n
    end
end

and it should be used as

(1..10).each_satsify lambda { |n| ...condition statment... } do |n|
    ...callback statement...
end

It reminded me that Perl also has the same problem. Even when we declare

sub each_satisfy(&&) {

}

Still only first sub keyword could be omitted. That means we should invoke it as

each_satisfy { ... condition ... } sub {
    my ($iter) = @_;
    ... callback ...
}

The only difference is Perl omitted the first (sub) and Ruby omitted the last (lambda).

I do want to figure out WHY they could not be designed to accept arbitrary quantity of anonymous functions.

 

LWP, cURL, OpenSSL and Posterous

The fun of programming is there is always subtle mechanism in implementation. It costs time to discover but sometime is deserving. for example.

$ curl --basic -u <user> http://example.com/api/

$ echo -n "user:pass" | openssl base64 -e

and



to



 

Posterous API in Perl

Posterous.com is a new (micro)-blog mesh-up system.

You can use it mesh up almost all your (micro)-blog. Post once, Publish Everywhere. And there is still many funny features.

It looks pretty good, so I translated its API to Perl. You can get it from CPAN. Still work in progress, but would be productive soon.

 

Can't locate Git.pm in @INC

Well, the installation of Github::Import worked well on Linux boxs but always reported "can't locate Git.pm in @INC" on Macintosh boxs. It is just a little trick and the answer is here.

Git.pm comes with git instead of being a part of CPAN. Building a copy of git-core or copying from other box solve it quickly.

 

the roadmap of upgrowth

和長輩們聊天總是很有趣的。即使那是一句老生常談，從長輩提攜的口吻中說出，依然受用。

6/26 的謝師宴，和賴暎杰老師坐在一起，我是這麼問的：「研究遇到瓶頸時，是怎麼面對的。」

「每天都多學會一件事，就不用擔心。」

yep...

 

Data::Model

Data::Model is a new ORM created by yappo, Its usage style is similar to DataMapper and Jifty::DBI.

Different from most famous equivalent, Data::Model handles multi-database. In order to do that, every model was appointed its database and table name in addition to its schema specification.

Let's see a simple example:



This is a simplest model in Data::Model, Only two DDL in it.

The first, base_driver( $driver ) specify which database the model would connect to. $driver is a Data::Model::Driver::DBI object. Write those diffusion code in every model is really not perlish, so we extract all possible drivers to MyApp::DB::driver().



The second, pass the table name and schema to install_model(), this step is the same as above two ORM system. and then, all thing done. By the way, columns() is the best syntax sugar provided by Data::Model::Schema.

Wait... Because Data::Model is too young to do auto_migration, we simply write MyApp::DB::make_schema() to do that. Finally, we could do simple CRUD in our application. as below.



Though Data::Model is too young to have some important feature such as validator and hook, even has no ability to handle relation between two table. The prototype is really exciting.

[Chinese Version]

 

.