Serving the Quantitative Finance Community

 
User avatar
bearish
Posts: 6521
Joined: February 3rd, 2011, 2:19 pm

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

November 7th, 2020, 2:58 pm

I predict that in five years time the title of this thread will be consistent with the calendar.
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

November 7th, 2020, 10:14 pm

I predict that in five years time the title of this thread will be consistent with the calendar.
I suppose you will be looking for a recount?
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

December 3rd, 2020, 8:09 pm

On the contrary, data scientists are making math cool again. Now everybody wants to learn linear algebra, calculus, and beyond.
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 27th, 2021, 12:44 pm

What is interesting  is to try and predict the next 5 years. Some possible scenarios are:

1. Prototyping in Python replaced/augmented with C++ production systems.
2. Evolving mathematical maturity (e.g. SGD gives way to RKHS etc.)
3. Specialised libraries, no 'one size fits all'.
4. Python code maintenance becoming 'great balls of mud'

Maybe
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
bearish
Posts: 6521
Joined: February 3rd, 2011, 2:19 pm

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 27th, 2021, 8:40 pm

I definitely agree that the state of affairs in 2025 is more interesting than in 2525, if for no other reason that I (and most of us around here) can plausibly hope to be alive and kicking in the former. I think both Python and C++ are probably here to stay, but Julia does fill in a space between them, with ease of use that exceeds the former and runtime speed that approaches the latter. She also plays well with others, and is represented by the “Ju” in Jupyter. And I know we have discussed this before, but a lot of commercial value is found in great balls of mud. Until it all goes terribly pear shaped, but that is usually “later”.
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 9:00 am

Purchase a product from 3rd parties under 2 conditions.

1. It is very expensive.
2. It doesn't satisfy the requirements.
3. Hire two experienced consultants to tweak the parameters.
Last edited by Cuchulainn on February 28th, 2021, 9:41 am, edited 1 time in total.
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
JohnLeM
Posts: 515
Joined: September 16th, 2008, 7:15 pm

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 9:33 am

I think both Python and C++ are probably here to stay, but Julia does fill in a space between them, with ease of use that exceeds the former and runtime speed that approaches the latter. She also plays well with others, and is represented by the “Ju” in Jupyter. And I know we have discussed this before, but a lot of commercial value is found in great balls of mud. Until it all goes terribly pear shaped, but that is usually “later”.
I am currently refounding my C++ code with python interface, and you are telling me that I should have chosen Julia ?
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 9:43 am

I think both Python and C++ are probably here to stay, but Julia does fill in a space between them, with ease of use that exceeds the former and runtime speed that approaches the latter. She also plays well with others, and is represented by the “Ju” in Jupyter. And I know we have discussed this before, but a lot of commercial value is found in great balls of mud. Until it all goes terribly pear shaped, but that is usually “later”.
I am currently refounding my C++ code with python interface, and you are telling me that I should have chosen Julia ?
No; bearish is saying it fills a gap. What that 'gap is, he doesn't say.
Supporting C++, Python _and_ Julia is probably a non-starter,

// Julia is a dynamic programming language; who wants that?
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
JohnLeM
Posts: 515
Joined: September 16th, 2008, 7:15 pm

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 9:58 am

I think both Python and C++ are probably here to stay, but Julia does fill in a space between them, with ease of use that exceeds the former and runtime speed that approaches the latter. She also plays well with others, and is represented by the “Ju” in Jupyter. And I know we have discussed this before, but a lot of commercial value is found in great balls of mud. Until it all goes terribly pear shaped, but that is usually “later”.
I am currently refounding my C++ code with python interface, and you are telling me that I should have chosen Julia ?
No; bearish is saying it fills a gap. What that 'gap is, he doesn't say.
I am having a closer look to this langage, it seems nice: template specialization, dynamic / static typing. C++ bindings seems a little bit painful (only through extern C ? maybe this project is mature enough ? but his thread seems to point it as obsolete). Nice mechanism of multple dispatch (this is very handy to design simple interface)
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 10:55 am

Yes, all nice cute features. And multiple dispatch is easy in C++ .. it's called Visitor. I discovered it independently in 1992 for CAD applications. It's not really an interface thing, it has a different purpose (separation of data from I/O) ... use std::function if you want pluggable components, or std::visit.
ROI?

30 years later C++11 supports visitors

https://en.cppreference.com/w/cpp/utility/variant/visit

This is a good way to support them multimethods.
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 11:11 am

... and ... Julia is #29 in the TIOBE top 10 hot list

https://www.tiobe.com/tiobe-index/
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
bearish
Posts: 6521
Joined: February 3rd, 2011, 2:19 pm

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 3:54 pm

One is reminded of the phrase “he filled a much needed gap.”

Some of these language discussions tend to drift into variants of “What is a good drone to perform visual inspection of remote fields? ... No, you really want a sturdy tractor, and this is the best one!”

Python obviously has a much larger user community, talent pool, and existing libraries. It is, for better and for worse, a general purpose language, whereas Julia was created for the primary purpose of doing math. Or so I’ve been led to believe. And the VS Code environment is pretty nice.
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 8:40 pm

One is reminded of the phrase “he filled a much needed gap.”

Some of these language discussions tend to drift into variants of “What is a good drone to perform visual inspection of remote fields? ...  No, you really want a sturdy tractor, and this is the best one!”
Image
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Alan
Posts: 10652
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 8:47 pm

I switched my latest book project coding from Mathematica to Python about a month ago, and am so glad I did. While I am a Python novice, I found it was easy to port my Mathematica functionality and more very quickly. 

But, what I would like to see by 2025 is some kind of, let's say, Python Virtual Assistant where you could just say things like:

Me: Acquire these files from the CBOE
Assist: Negotiating ... Cost will be $X. Agreed?
Me: Agree. 
Assist: Data acquired.
Me: Now split them into one file for each option expiration and underlying root symbol.
Assist. Coding ... Done
Me: Wait, back-up. When you split them, drop options with zero bids or expiring on the trade date.
Assist. Re-coding ... Done
Me: Take a look at equation (4.1) in chapter2.tex and code that.
Assist: Ambiguous. First, here are some questions ...
Me: (Answers questions)
Assist: Coding ... Done
etc, etc

Would be nice. If we can teach a computer to drive a car, it ought to be able to do this.  :D      
 
User avatar
Cuchulainn
Topic Author
Posts: 64669
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: FIVE hundred years from now: can you predict the Software Landscape in anno 2525?

February 28th, 2021, 9:28 pm

Let me put it this way, Mr. Alan. The 9000 series is the most reliable computer ever made. No 9000 computer has ever made a mistake or distorted information. We are all, by any practical definition of the words, foolproof and incapable of error.
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl