Comments for [Scientific & Financial Computation ] -- Shawn's blog[Scientific & Financial Computation ] -- Shawn's blog

Comment on Constraints on Arguments in Python by Shawn Zhang

Shawn Zhang — Thu, 24 Jan 2013 05:51:31 +0000

Hi, Trevor
I think the decorator will be good solution if the programming contains 100 functions which need constraints on their arguments. if the number of functions remains a small amount , your suggestion is more simple and elegant .

Comment on Constraints on Arguments in Python by Trevor Rowland

Trevor Rowland — Thu, 24 Jan 2013 05:40:53 +0000

Not as pretty as the decorator, but you can use isinstance() for example:-
def addsomething(x,y):
....if not isinstance(x, int):
........raise
I have used this on the odd occasion to implement overloaded function calls when other people are calling my code and I'm not sure what they are sending me

Cheers
Trevor

Comment on Constraints on Arguments in Python by Shawn Zhang

Shawn Zhang — Tue, 22 Jan 2013 10:49:34 +0000

Hi,Mike Thanks for your comment , that's a thoughtful thinking and I recently having a look at Ocaml(which is a strongly type language), and I can't agree more with you. That's a lot cooler if using a strongly typed language ， not only saving developing workload but also enjoy a far more executing speed than Python. I've been told that Ocaml is 70% elegance as Python and 70% speed performance of C. that's why I'm learning this language, it can served as a general-purpose language and strongly typed(with great type inference system)

Comment on Constraints on Arguments in Python by Mike Meyer

Mike Meyer — Tue, 22 Jan 2013 10:30:54 +0000

You can get that workload savings you talk about in languages with statically typed variables - *if* the language has a good type inferencing engine. For instance, you can write:

add x y = x + 1

to define your add function, and just like python, it'll work for any type for which the + operator is defined. However, because the variables have types, the compiler (not the run-time system) will catch the error if you try and pass in a pair of variables which can't be added, saving even *more* time.

Given that the compiler already has a notion of types, then all you need is a notation to tell it which types you want to allow, and it can provide the same kind of restrictions that your decorator does. Again, caught at compile time instead of run time.

Now all I need is a language for general-purpose programming that I as much as I like Python that has a type system of that quality.

Comment on Generic Newton-Raphson Method in R by lose weight naturally

lose weight naturally — Sun, 16 Dec 2012 08:01:30 +0000

Thank you ever so for you blog article.Really thank you! Keep writing.

Comment on Generic Newton-Raphson Method in R by weight loss fast

weight loss fast — Sun, 16 Dec 2012 07:57:32 +0000

I think this is a real great post.Really thank you! Want more.

Comment on Market Data Extraction from LMAX API using Regular Expression by Shawn

Shawn — Tue, 20 Nov 2012 01:46:56 +0000

hi, Julien ,sorry for late reply.

I'm not tested it yet .
but this regrex practise provide a clear code ,but less OO.
if you want to get prices of bid /ask , you have to loop the getAskPrices() getBidPrices(),that will be a overhead cost .

Comment on Market Data Extraction from LMAX API using Regular Expression by julien

julien — Mon, 08 Oct 2012 13:38:19 +0000

Hey this looks great! Does this runs faster than orderBookEvent's built i methods?
Also I was wondering if you know a good way to retrieve the instrument name?

I've recently started to look at the LMAX API seriously and so far I like it very much. Hope we can exchange some implementation ideas !