SERVING THE QUANTITATIVE FINANCE COMMUNITY

  • 1
  • 2
  • 3
  • 4
  • 5
  • 7
 
User avatar
Cuchulainn
Posts: 63263
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Ideas about new open source quanfin project

October 3rd, 2011, 2:01 pm

QuoteOriginally posted by: daveangelQuoteOriginally posted by: CuchulainnQuoteA good thing would be (at the final end of your project) to have also a graphic interface to make your code a complete black blox with a nice user interface. UI is possible, but the main concern now is the programming interface between the different modules.when I first came into this business I had this great idea that we could have a PDE solver and the user specifies boundary values and initial conditions from a UI similar to how a finite element analysis system is designed to work.You mean some kind of language (STEP-like/EBNF) that gets parsed? It took ne 4 weeks to discretise a CRT as preprocessor to my FEM system but these days Boost might able to do it. Are you suggesting a kind of payoff language that get parsed into C++/F#/C# code?
Chips chips chips Du du du du du Ci bum ci bum bum Du du du du du Ci bum ci bum bum Du du du du du
http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Cuchulainn
Posts: 63263
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Ideas about new open source quanfin project

October 3rd, 2011, 2:06 pm

QuoteOriginally posted by: outrunQuoteOriginally posted by: SierpinskyJanitorIf we wanna juicy cool GUIs just give Serp back his credentials please...Also, I am volunteering for Payoff processing in FpML ( ie Contract Serialization )SerpThis Author list of going away today, I know it's a bit confusing.. We are working hard on setting this up correct with good intentions for everyone. We are looking at an open community approach like boost, * anyone can ask the community "is X a good idea?" * if *someone* thinks so, then work with the (sub)community that hooks on to code their idea in the sandbox (or maybe you do it alone, or have done it in the past), the barrier is low* have sensible quality guidelines* put your lib up for review, have the community review it, (here the barrier is high)* and finally accepted a lib in release based on the feedback of the whole community.What's the purpose/added value of a GUI at this stage?IMO GUI stuff is not on the project critical path. It is a *view* of the underlying model. Projects with a premature focus on GUI...My suggestion would be to come back with hard arguments on high-risk/high priority project activities.
Last edited by Cuchulainn on October 2nd, 2011, 10:00 pm, edited 1 time in total.
Chips chips chips Du du du du du Ci bum ci bum bum Du du du du du Ci bum ci bum bum Du du du du du
http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
SierpinskyJanitor
Posts: 1069
Joined: March 29th, 2005, 12:55 pm

Ideas about new open source quanfin project

October 3rd, 2011, 2:09 pm

QuoteThere is an industry XML standard for specifying contract details, can't remember the name ... r u kidding me? I have personally spawned 2 threads on Payoff spec languages, also you seemed to have removed "Serp" from this list:Quotedefine a group of Authors and Owners. Currently we have Paul (Wilmott), Cuchulainn (Daniel Duffy), Outrun (Thijs van den Berg), Renorm and Semanticum (Dominik).care to explain why are we discriminating and ignoring Janitors, Mr. Outrun? I have nothing against discriminating myself, in fact I do it constantly, but at least there should be some kind of rationale behind it.
 
User avatar
SierpinskyJanitor
Posts: 1069
Joined: March 29th, 2005, 12:55 pm

Ideas about new open source quanfin project

October 3rd, 2011, 2:10 pm

QuoteWhat's the purpose/added value of a GUI at this stage?the same as locking ourselves up to Excel - none.
 
User avatar
Cuchulainn
Posts: 63263
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Ideas about new open source quanfin project

October 3rd, 2011, 2:19 pm

QuoteOriginally posted by: SierpinskyJanitorQuoteWhat's the purpose/added value of a GUI at this stage?the same as locking ourselves up to Excel - none.I think you have missed my point, completely. It's a design issue, then the UI will come later. That's what I meant. And no one said anything about being tied into Excel?? BTW Excel is imporatnt. //IMO GUI stuff is not on the project critical path. It is a *view* of the underlying model. Projects with a premature focus on GUI tend to be very UNSTABLEMy suggestion would be to come back with hard arguments on high-risk/high priority project activities.
Last edited by Cuchulainn on October 2nd, 2011, 10:00 pm, edited 1 time in total.
Chips chips chips Du du du du du Ci bum ci bum bum Du du du du du Ci bum ci bum bum Du du du du du
http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
SierpinskyJanitor
Posts: 1069
Joined: March 29th, 2005, 12:55 pm

Ideas about new open source quanfin project

October 3rd, 2011, 2:26 pm

the proto-SPEC clearly refers to Excel. All I am suggesting is keeping an open-mind rgd this.I would never undermine the value and importance of Excel, but locking our spec up to this as the only interface is a major deal-breaker. Many potential clients would simply ignore this LIB under such constraints. Let me tell you about one of them: me.** yes, I don´t like Excel.
 
User avatar
SierpinskyJanitor
Posts: 1069
Joined: March 29th, 2005, 12:55 pm

Ideas about new open source quanfin project

October 3rd, 2011, 2:47 pm

pardon? In zee spec:QuoteExcel (no GUI planned yet (do we need GUI beyond Excel?). move to Interfaces)All I am saying is: YES we need/should go beyond Excel, also, I am voluteering for that as well. I am also adding, as a potential client of any such LIB/tool, that I wouldn´t even bother benchmarking it were it focused solely on XLLs for interfacing and/or integration.
 
User avatar
Cuchulainn
Posts: 63263
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Ideas about new open source quanfin project

October 3rd, 2011, 3:56 pm

QuoteOriginally posted by: outrunOur goal for the next days is to quickie layout an open community like boost; one that works and focuses on deliverables and healthy socially. I personally know Paul and Cuch and I help setup some tools so that we can quickly move forward and get going. The discriminating aspect righr now during thus IT stuff would be that we know each, met face to face, worked together before, drank beer, wrote equations on each others blackboards, things like that.We'll be writing down the bit about process and community I wrote in a couple of posts back on the project page in more detail. That's going to be a version 1 community guideline. Thing like that need to be clear, and we borrow from the sucessfull boost. We want to get going and layout a good environment this week.Objectively speaking, track record is important and if you see all the real useful posts by @rnorm (for example) here, then you know what I mean. And haviing a 'product' is a boon.Personally, I am not interested in bla bla but design patterns (new style), boost, FDM framework and linking up to outrun's exact option calculator in the short term ([6,8] months. Software forum is for more easy-going banter.
Last edited by Cuchulainn on October 2nd, 2011, 10:00 pm, edited 1 time in total.
Chips chips chips Du du du du du Ci bum ci bum bum Du du du du du Ci bum ci bum bum Du du du du du
http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
rmax
Posts: 6080
Joined: December 8th, 2005, 9:31 am

Ideas about new open source quanfin project

October 3rd, 2011, 4:28 pm

QuoteOriginally posted by: CuchulainnQuoteOriginally posted by: outrunOur goal for the next days is to quickie layout an open community like boost; one that works and focuses on deliverables and healthy socially. I personally know Paul and Cuch and I help setup some tools so that we can quickly move forward and get going. The discriminating aspect righr now during thus IT stuff would be that we know each, met face to face, worked together before, drank beer, wrote equations on each others blackboards, things like that.We'll be writing down the bit about process and community I wrote in a couple of posts back on the project page in more detail. That's going to be a version 1 community guideline. Thing like that need to be clear, and we borrow from the sucessfull boost. We want to get going and layout a good environment this week.Objectively speaking, track record is important and if you see all the real useful posts by @rnorm (for example) here, then you know what I mean. And haviing a 'product' is a boon.Personally, I am not interested in bla bla but design patterns (new style), boost, FDM framework and linking up to outrun's exact option calculator in the short term ([6,8] months. Software forum is for more easy-going banter.I can't access Renorm's doc from where I am situated at present, but:Looking at the posts to date there are some pretty enthusiastic individuals, and some pretty enthusiastic and outspoken people (as you would expect in finance), and this is going to run into problems if you don't have a common understanding as to what you want to DO. Probably a lot of the OpenSource projects start as a bubble, and then can't get beyond the easy bit when problems rear their head as no-one as a frame of reference for resolution.1. Is this an educational exercise to teach people how to solve QF problems (not a bad place to start when looking at Linux) and talk about architecture etc2. Is it expected to be used in smaller shops where they cannot afford good solid libraries and frameworks3. A working framework where people can add and remove libraries at will (I am not sure everyone in the market is going to be happy to use the same methods to boot strapping YC etc).You are going to need principles?. If this is all in renorm's doc then I apologise!
 
User avatar
Cuchulainn
Posts: 63263
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Ideas about new open source quanfin project

October 3rd, 2011, 4:47 pm

QuoteOriginally posted by: rmaxQuoteOriginally posted by: CuchulainnQuoteOriginally posted by: outrunOur goal for the next days is to quickie layout an open community like boost; one that works and focuses on deliverables and healthy socially. I personally know Paul and Cuch and I help setup some tools so that we can quickly move forward and get going. The discriminating aspect righr now during thus IT stuff would be that we know each, met face to face, worked together before, drank beer, wrote equations on each others blackboards, things like that.We'll be writing down the bit about process and community I wrote in a couple of posts back on the project page in more detail. That's going to be a version 1 community guideline. Thing like that need to be clear, and we borrow from the sucessfull boost. We want to get going and layout a good environment this week.Objectively speaking, track record is important and if you see all the real useful posts by @rnorm (for example) here, then you know what I mean. And haviing a 'product' is a boon.Personally, I am not interested in bla bla but design patterns (new style), boost, FDM framework and linking up to outrun's exact option calculator in the short term ([6,8] months. Software forum is for more easy-going banter.I can't access Renorm's doc from where I am situated at present, but:Looking at the posts to date there are some pretty enthusiastic individuals, and some pretty enthusiastic and outspoken people (as you would expect in finance), and this is going to run into problems if you don't have a common understanding as to what you want to DO. Probably a lot of the OpenSource projects start as a bubble, and then can't get beyond the easy bit when problems rear their head as no-one as a frame of reference for resolution.1. Is this an educational exercise to teach people how to solve QF problems (not a bad place to start when looking at Linux) and talk about architecture etc2. Is it expected to be used in smaller shops where they cannot afford good solid libraries and frameworks3. A working framework where people can add and remove libraries at will (I am not sure everyone in the market is going to be happy to use the same methods to boot strapping YC etc).You are going to need principles?. If this is all in renorm's doc then I apologise!Agree. Things happen quickly so we will catch up. Points 3 (and of course 1,2) are high priority. I keep ranting about interface because they are needed for point 3. Some interfaces (e.g. bootstrap) can be left unfilled and this will not step on anyone's toes whereas the comon-or-garden stuff will have implemenations. At least that's the way I see it.
Last edited by Cuchulainn on October 2nd, 2011, 10:00 pm, edited 1 time in total.
Chips chips chips Du du du du du Ci bum ci bum bum Du du du du du Ci bum ci bum bum Du du du du du
http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
rmax
Posts: 6080
Joined: December 8th, 2005, 9:31 am

Ideas about new open source quanfin project

October 3rd, 2011, 5:01 pm

QuoteCuch said:Agree. Things happen quickly so we will catch up. Points 3 (and of course 1,2) are high priority. I keep ranting about interface because they are needed for point 3. Some interfaces (e.g. bootstrap) can be left unfilled and this will not step on anyone's toes whereas the comon-or-garden stuff will have implemenations. At least that's the way I see it.makes sense. I can see a good majority of people will use it if they can snap-in what they need and it avoids a lot of the tedious non prop stuff. Agree in terms of interfaces: I am all for open source; interfaces can be contributed by anyone, but need to be tightly controlled. Would design that bit first before getting carried away in terms of the more exotic stuff.Also think in terms of licensing it should be very clear as to what other 3rd party products are being used and the legal ramifications of that need to be thought through. People who are going to want to use this will not want to be exposed to a heap of operational risk due to uncontrolled open source having to be included in the build. The license will have to take that into account as well...
Last edited by rmax on October 2nd, 2011, 10:00 pm, edited 1 time in total.
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On