quantlib-dev — Discussion of QuantLib development

Re: [Quantlib-dev] CVS root change

[eric e. <eri...@gm...>]

Hello,

On 5/16/06, Luigi Ballabio <lui...@gm...> wrote:
> I haven't tried the anonymous CVS yet; reports are welcome.

I've tried anonymous access from behind a firewall without success.

Under the old infrastructure, for those of us behind firewalls which
block traffic on the standard ports, sourceforge provided alternate
servers/ports for anonymous (pserver) and development (ssh) access:

cvs-pserver.sourceforge.net 443
cvs-ssh.sourceforge.net 443

Under the new infrastructure, for developer access from behind a
firewall, the updated docs say to use the standard server with an
alternate port

quantlib.cvs.sourceforge.net 443

and that works fine for me but as far as I can see the new docs say
nothing at all about anonymous access from behind a firewall.

I raised a sourceforge support request but in the meantime I'd be
grateful for any ideas anyone may have.

Regards,
Eric

[Quantlib-dev] CVS root change

[Luigi B. <lui...@gm...>]

Hi all,
	the CVS repository has a new address; see =20
<http://quantlib.org/cvs.shtml> for details. I haven't tried the =20
anonymous CVS yet; reports are welcome.

Luigi


----------------------------------------

A debugged program is one for which you have not yet found the
conditions that make it fail.
-- Jerry Ogdin

[Quantlib-dev] Heads up---SourceForge CVS

[Luigi B. <lui...@gm...>]

Hi all,
	you might have noticed the CVS problems we had in the past few =20
days. SourceForge is about to switch to new servers which should be =20
online later today. Unfortunately, this means that we might have to =20
check out new working copies as the address of the underlying CVS =20
server will change [*]. In that case, if you have local modifications =20
you were waiting to commit, you'll have to merge them into the newly =20
checked-out working copy and commit from there. Please do this with =20
care; I suggest you take a diff between your working copy and the =20
repository as of yesterday, May 11th (this can be done by passing the =20
'-D "12 May 2006"' flag to "cvs checkout") and apply the resulting =20
patch carefully so that you don't accidentally overwrite other changes.

Thanks,
	Luigi

[*] This might not be true if cvs.quantlib.org is made to point to the =20
new server. Nando, can you look into it?


----------------------------------------

standards, n.:
         The principles we use to reject other people's code.

Re: [Quantlib-dev] Re: ConvertibleBond framework

[<TB...@ao...>]

Hi toyin,
 
I have not looked at the interest rate monte carlo work. But I need it for  a 
Mortgage Backed Security pricing engine I am supposed to be working  on.
 
 
But I will download latest cvs version and look at it.
 
 
Regards
 
Theo
 

[Quantlib-dev] Re:[Quantlib-users] HullWhiteProcess class usage...

[Toyin A. <toy...@ho...>]

Hi,

Thankyou for your response.

Okay for Caps/Floors you use the forward measure.

Can you give me an example of an instrument that is naturally applicable to 
the risk neutral measure?

Again, time permitting, it would be useful to view examples of using the 
four processes.

Toy out.


>From: "mar...@li..." <mar...@li...>
>CC: "quantlib-dev" <qua...@li...>,"quantlib-users" 
><qua...@li...>,"luigi.ballabio" 
><lui...@gm...>
>Subject: Re:[Quantlib-users] HullWhiteProcess class usage...
>Date: Thu, 11 May 2006 12:09:13 +0200
>
>toyin sent me the files and now I know that they are in the CVS.
>
>I think there is not a simple way to modify the files you mentioned: the 
>pricing formulas implemented in the code are referred to a forward measure.
>The idea is: I write the pricing formula in a chosen measure and then 
>simulate in that measure.
>So, the most natural measure (at least for caps/floors and many other 
>instruments) is the forward measure.
>If you want to simulate the dynamics in another measure, than you have to 
>write the pricing formula in that measure (and not always it is possible, 
>for eaxample I do not know analytical formulas for pricing caps in the risk 
>neutral framework). Then it is just a matter of implementing it: remember 
>that the endDiscount_ in nmHullWhiteEngine is the actual discount (ie zero 
>coupon price) and its use is justified by forward measure pricing, but you 
>cannot use it in the case of risk neutral pricing.
>The discount factor is a random variable and it can be substitued with the 
>price only in case of the relative forward measure.
>The idea is similar to the one of bgm pricing.
>
>regards,
>Marco
>
>
>
>
>-------------------------------------------------------
>Using Tomcat but need to do more? Need to support web services, security?
>Get stuff done quickly with pre-integrated technology to make your job 
>easier
>Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>http://sel.as-us.falkag.net/sel?cmd=lnk&kid0709&bid&3057&dat1642
>_______________________________________________
>Quantlib-users mailing list
>Qua...@li...
>https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Quantlib-dev] Re:[Quantlib-users] HullWhiteProcess class usage...

[marco\.tarenghi\@libero\.it <mar...@li...>]

toyin sent me the files and now I know that they are in the CVS.

I think there is not a simple way to modify the files you mentioned: the =
pricing formulas implemented in the code are referred to a forward measur=
e.
The idea is: I write the pricing formula in a chosen measure and then sim=
ulate in that measure.
So, the most natural measure (at least for caps/floors and many other ins=
truments) is the forward measure.
If you want to simulate the dynamics in another measure, than you have to=
 write the pricing formula in that measure (and not always it is possible=
, for eaxample I do not know analytical formulas for pricing caps in the =
risk neutral framework). Then it is just a matter of implementing it: rem=
ember that the endDiscount_ in nmHullWhiteEngine is the actual discount (=
ie zero coupon price) and its use is justified by forward measure pricing=
, but you cannot use it in the case of risk neutral pricing.
The discount factor is a random variable and it can be substitued with th=
e price only in case of the relative forward measure.
The idea is similar to the one of bgm pricing.

regards,
Marco

[Quantlib-dev] Re:[Quantlib-users] HullWhiteProcess class usage...

[Toyin A. <toy...@ho...>]

Hi,

Firstly excellent work (you and your team) on the new framework, I've been 
playing with it for a
while.

Actually Luigi contributed this code within the dev version of CVS (no
examples though) and with the current problems that sourceforge is having
you cannot directly obtain this code.

Okay, basically I'm looking for a class similar to mchullwhiteengine that
uses the non-forward process class.

Basically I want to simply compute the payoff and discount. I do not want to
use some sort of forwarddate notion. I suspect that the non-forward
HullWhite process class provides this for me.
However the problem is how does one discount the payoff (line #79 of the
mchullwhiteengine.cpp file) if I use this class.

From the current implementation within the mchullwhiteengine.cpp file, an
accural factor is computed (upto the forward date) and then the payoff
discounted by the deterministic discount factor of the forward date.

As the non-forward HullWhite process has no notion of the forward time, I'm
not too sure how one modifies the code within the mchullwhiteengine.cpp
(line #79) file in order to discount correctly (ie, do you simply discount
each payoff by the deterministic discount factor of the paymant date and
forget about the multiplication of the accuralfactor?)

Let's put this another way (in case the logic above is incorrect)...

Basically I'm looking for examples of using all of the four interest rate 
process
objects within the pricing of an instrument under a montecarlo setting.

Once I have this "bigger picture", I'm sure I'll be able to price the way 
that I wish too.

Best Reagrds,
Toyin Akin.

>From: "mar...@li..." <mar...@li...>
>CC: "quantlib-dev" <qua...@li...>,"quantlib-users" 
><qua...@li...>,"luigi.ballabio" 
><lui...@gm...>
>Subject: Re:[Quantlib-users] HullWhiteProcess class usage...
>Date: Thu, 11 May 2006 11:55:17 +0200
>
>Hi Toyin, my name is Marco and I am one of the persons who contributed the 
>forward processes files to QuantLib.
>Actually I have to say that we cannot find the files you mentioned in your 
>mail...
>We have the 0.3.12 release, but we do not find the forward processes 
>neither the mcHullWhite pricing engine class.
>Where did you find them?
>Anyway I cannot clearly understand your problem: I mean, the two processes 
>are nearly the same so you should be able to instantiate them in the same 
>way and you can avoid to set the ForwardMeasureTime.
>But probably if I could see the files I would better understand.
>
>On the other side you are also right: at the moment you have to instantiate 
>only simple StochasticProcess1D objects if you need to use the 
>stocahsticProcessArray. Actually, when we developed the forward processes 
>classes, we focused on the pricing side of the problem. When the process is 
>in the risk neutral measure, you cannot price the instrument (e.g. a cap) 
>via Monte Carlo.
>This problem was much more evident in the g2Process, where passing to the 
>forward measure also the correlation matrix changes: we spent a lot of time 
>on this silly fact...
>I do not know what you are really interested in, but if you need a process 
>containing two other process, I sugget you to follow the g2Process idea, 
>that is creating a new stochasticProcess containing two simple processes, 
>and these processes could be either forward processes or not. Just be 
>careful when considering the measure in which you're doing the pricing!!
>
>I hope I've been clear, please feel free to contact me for any other 
>question.
>
>Regards,
>Marco
>
>
>---------- Initial Header -----------
>
>From      : qua...@li...
>To          : qua...@li..., 
>qua...@li...
>Cc          : lui...@gm...
>Date      : Thu, 11 May 2006 03:16:02 +0100
>Subject : [Quantlib-users] HullWhiteProcess class usage...
>
>
>
>
>
>
>
> >
> > Hi All,
> >
> > I am looking for an example of how one would use the HullWhiteProcess 
>class.
> > There is already an example of how one would use the 
>HullWhiteForwardProcess
> > class within the
> > mcHullWhite Pricing Engine class.
> >
> > The key issue for me are (because we do not need to use the
> > setForwardMeasureTime() function)
> > what adjustments needs to be made to a copy of the mcHullWhite  Pricing
> > Engine class (within the PricingEngines/CapFloor folder) in order to 
>make
> > use of the HullWhiteProcess object instead
> > (for example - the accrualFactor variable, which discount factors to use 
>for
> > discounting the payoff at each period, stochastic, deterministic 
>etc...).
> >
> > In addition, because these classes are derived from the 
>StochasticProcess1D
> > class, will they be processed correctly if used within the
> > StochasicProcessArray class. Let's say 2 HullWhiteProcess objects, an FX
> > process object along with a correlation matrix.
> >
> > I know that person(s) from "Banca Profilo S.p.A" initially contributed 
>this
> > code so if they or anyone else has an idea of how to the said class, it
> > would be great.
> >
> > Best Regards,
> > Toyin Akin.
> >
> > >From: Luigi Ballabio <lui...@gm...>
> > >To: Toyin Akin <toy...@ho...>
> > >Subject: Re: mcHullWhite Pricing Engine...
> > >Date: Fri, 05 May 2006 17:10:46 +0200
> > >
> > >
> > >On 05/02/2006 07:58:27 AM, Toyin Akin wrote:
> > >>Another question regarding lines #62 and #63 within the mcHullWhite
> > >>pricing engine.cpp file.
> > >>
> > >>You use +1 and +2 as indexes into the path array.
> > >>
> > >>I am making the assumption that the path[] array stores the short  
>rates
> > >>and path[0] is the short rate on the fixing date of the first
> > >>caplet/floorlet (also the fixing rate to be used).
> > >
> > >No, path[0] is the current time (t = 0).
> > >
> > >As for your other questions, you might want to post on the mailing list 
>  so
> > >that you can reach the contributor of the processes. I haven't  looked 
>very
> > >hard into the implementation yet...
> > >
> > >Later,
> > >	Luigi
> > >
> > >
> > >----------------------------------------
> > >
> > >All generalizations are false, including this one.
> > >-- Mark Twain
> >
> >
> >
> >
> > -------------------------------------------------------
> > Using Tomcat but need to do more? Need to support web services, 
>security?
> > Get stuff done quickly with pre-integrated technology to make your job 
>easier
> > Download IBM WebSphere Application Server v.1.0.1 based on Apache 
>Geronimo
> > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> > _______________________________________________
> > Quantlib-users mailing list
> > Qua...@li...
> > https://lists.sourceforge.net/lists/listinfo/quantlib-users
> >
>
>
>
>-------------------------------------------------------
>Using Tomcat but need to do more? Need to support web services, security?
>Get stuff done quickly with pre-integrated technology to make your job 
>easier
>Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>http://sel.as-us.falkag.net/sel?cmd=lnk&kid0709&bid&3057&dat1642
>_______________________________________________
>Quantlib-users mailing list
>Qua...@li...
>https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Quantlib-dev] Re:[Quantlib-users] HullWhiteProcess class usage...

[marco\.tarenghi\@libero\.it <mar...@li...>]

Hi Toyin, my name is Marco and I am one of the persons who contributed th=
e forward processes files to QuantLib.
Actually I have to say that we cannot find the files you mentioned in you=
r mail...
We have the 0.3.12 release, but we do not find the forward processes neit=
her the mcHullWhite pricing engine class.
Where did you find them?
Anyway I cannot clearly understand your problem: I mean, the two processe=
s are nearly the same so you should be able to instantiate them in the sa=
me way and you can avoid to set the ForwardMeasureTime.
But probably if I could see the files I would better understand.

On the other side you are also right: at the moment you have to instantia=
te only simple StochasticProcess1D objects if you need to use the stocahs=
ticProcessArray. Actually, when we developed the forward processes classe=
s, we focused on the pricing side of the problem. When the process is in =
the risk neutral measure, you cannot price the instrument (e.g. a cap) vi=
a Monte Carlo. 
This problem was much more evident in the g2Process, where passing to the=
 forward measure also the correlation matrix changes: we spent a lot of t=
ime on this silly fact...
I do not know what you are really interested in, but if you need a proces=
s containing two other process, I sugget you to follow the g2Process idea=
, that is creating a new stochasticProcess containing two simple processe=
s, and these processes could be either forward processes or not. Just be =
careful when considering the measure in which you're doing the pricing!!

I hope I've been clear, please feel free to contact me for any other ques=
tion.

Regards,
Marco


---------- Initial Header -----------

From      : qua...@li...
To          : qua...@li..., qua...@li...=
urceforge.net
Cc          : lui...@gm...
Date      : Thu, 11 May 2006 03:16:02 +0100
Subject : [Quantlib-users] HullWhiteProcess class usage...







> 
> Hi All,
> 
> I am looking for an example of how one would use the HullWhiteProcess c=
lass.
> There is already an example of how one would use the HullWhiteForwardPr=
ocess 
> class within the
> mcHullWhite Pricing Engine class.
> 
> The key issue for me are (because we do not need to use the 
> setForwardMeasureTime() function)
> what adjustments needs to be made to a copy of the mcHullWhite  Pricing=
 
> Engine class (within the PricingEngines/CapFloor folder) in order to ma=
ke 
> use of the HullWhiteProcess object instead
> (for example - the accrualFactor variable, which discount factors to us=
e for 
> discounting the payoff at each period, stochastic, deterministic etc...=
).
> 
> In addition, because these classes are derived from the StochasticProce=
ss1D 
> class, will they be processed correctly if used within the 
> StochasicProcessArray class. Let's say 2 HullWhiteProcess objects, an F=
X 
> process object along with a correlation matrix.
> 
> I know that person(s) from "Banca Profilo S.p.A" initially contributed =
this 
> code so if they or anyone else has an idea of how to the said class, it=
 
> would be great.
> 
> Best Regards,
> Toyin Akin.
> 
> >From: Luigi Ballabio <lui...@gm...>
> >To: Toyin Akin <toy...@ho...>
> >Subject: Re: mcHullWhite Pricing Engine...
> >Date: Fri, 05 May 2006 17:10:46 +0200
> >
> >
> >On 05/02/2006 07:58:27 AM, Toyin Akin wrote:
> >>Another question regarding lines #62 and #63 within the mcHullWhite  =

> >>pricing engine.cpp file.
> >>
> >>You use +1 and +2 as indexes into the path array.
> >>
> >>I am making the assumption that the path[] array stores the short  ra=
tes 
> >>and path[0] is the short rate on the fixing date of the first  
> >>caplet/floorlet (also the fixing rate to be used).
> >
> >No, path[0] is the current time (t =3D 0).
> >
> >As for your other questions, you might want to post on the mailing lis=
t  so 
> >that you can reach the contributor of the processes. I haven't  looked=
 very 
> >hard into the implementation yet...
> >
> >Later,
> >	Luigi
> >
> >
> >----------------------------------------
> >
> >All generalizations are false, including this one.
> >-- Mark Twain
> 
> 
> 
> 
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, securit=
y?
> Get stuff done quickly with pre-integrated technology to make your job =
easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geron=
imo
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=
=3D121642
> _______________________________________________
> Quantlib-users mailing list
> Qua...@li...
> https://lists.sourceforge.net/lists/listinfo/quantlib-users
> 

RE: [Quantlib-dev] Re: ConvertibleBond framework

[Toyin A. <toy...@ho...>]

Okay,

Nice work by the way...

Have you had a look at the HullWhite, G2 process classes yet (I know that 
the code is currently within the dev CVS)?

I believe that you too were looking for an interest rate montecarlo 
framework.

Toy out.

>From: TB...@ao...
>To: toy...@ho...
>CC: lui...@gm..., qua...@li...
>Subject: [Quantlib-dev] Re: ConvertibleBond framework
>Date: Thu, 11 May 2006 04:52:18 EDT
>
>Hi toyin,
>
>Luigi can correct me if I am wrong but so far the implementation is  
>correct.
>
>The posting you saw was regarding discrete dividends which is not part of
>the current implementation as what we had in developement was incorrect.
>
>I intend to look at the discrete dividends bit.
>
>
>Regards
>
>Theo
>

[Quantlib-dev] Re: ConvertibleBond framework

[<TB...@ao...>]

Hi toyin,
 
Luigi can correct me if I am wrong but so far the implementation is  correct.
 
The posting you saw was regarding discrete dividends which is not part of  
the current implementation as what we had in developement was incorrect.
 
I intend to look at the discrete dividends bit.
 
 
Regards
 
Theo
 

[Quantlib-dev] Re: ConvertibleBond framework

[Luigi B. <lui...@gm...>]

On 05/11/2006 06:31:51 AM, Toyin Akin wrote:
> Is the ConvertibleBond framework operating correctly within quantlib?

To the best of my knowledge, the released version is correct.

Later,
	Luigi


----------------------------------------

The first rule of intelligent tinkering is to save all the parts.
-- Paul Erlich

[Quantlib-dev] ConvertibleBond framework

[Toyin A. <toy...@ho...>]

Hi Theo,

Is the ConvertibleBond framework operating correctly within quantlib?

I seem to remember previous postings suggesting that the implementation was 
incorrect.
Is this still the case?

Best Regards,
Toyin Akin.

[Quantlib-dev] HullWhiteProcess class usage...

[Toyin A. <toy...@ho...>]

Hi All,

I am looking for an example of how one would use the HullWhiteProcess class.
There is already an example of how one would use the HullWhiteForwardProcess 
class within the
mcHullWhite Pricing Engine class.

The key issue for me are (because we do not need to use the 
setForwardMeasureTime() function)
what adjustments needs to be made to a copy of the mcHullWhite  Pricing 
Engine class (within the PricingEngines/CapFloor folder) in order to make 
use of the HullWhiteProcess object instead
(for example - the accrualFactor variable, which discount factors to use for 
discounting the payoff at each period, stochastic, deterministic etc...).

In addition, because these classes are derived from the StochasticProcess1D 
class, will they be processed correctly if used within the 
StochasicProcessArray class. Let's say 2 HullWhiteProcess objects, an FX 
process object along with a correlation matrix.

I know that person(s) from "Banca Profilo S.p.A" initially contributed this 
code so if they or anyone else has an idea of how to the said class, it 
would be great.

Best Regards,
Toyin Akin.

>From: Luigi Ballabio <lui...@gm...>
>To: Toyin Akin <toy...@ho...>
>Subject: Re: mcHullWhite Pricing Engine...
>Date: Fri, 05 May 2006 17:10:46 +0200
>
>
>On 05/02/2006 07:58:27 AM, Toyin Akin wrote:
>>Another question regarding lines #62 and #63 within the mcHullWhite  
>>pricing engine.cpp file.
>>
>>You use +1 and +2 as indexes into the path array.
>>
>>I am making the assumption that the path[] array stores the short  rates 
>>and path[0] is the short rate on the fixing date of the first  
>>caplet/floorlet (also the fixing rate to be used).
>
>No, path[0] is the current time (t = 0).
>
>As for your other questions, you might want to post on the mailing list  so 
>that you can reach the contributor of the processes. I haven't  looked very 
>hard into the implementation yet...
>
>Later,
>	Luigi
>
>
>----------------------------------------
>
>All generalizations are false, including this one.
>-- Mark Twain

[Quantlib-dev] HullWhiteProcess class usage...

[Toyin A. <toy...@ho...>]

Hi All,

I am looking for an example of how one would use the HullWhiteProcess class.
There is already an example of how one would use the HullWhiteForwardProcess 
class within the
mcHullWhite Pricing Engine class.

The key issue for me are (because we do not need to use the 
setForwardMeasureTime() function)
what adjustments needs to be made to a copy of the mcHullWhite  Pricing 
Engine class (within the PricingEngines/CapFloor folder) in order to make 
use of the HullWhiteProcess object instead
(for example - the accrualFactor variable, which discount factors to use for 
discounting the payoff at each period, stochastic, deterministic etc...).

In addition, because these classes are derived from the StochasticProcess1D 
class, will they be processed correctly if used within the 
StochasicProcessArray class. Let's say 2 HullWhiteProcess objects, an FX 
process object along with a correlation matrix.

I know that person(s) from "Banca Profilo S.p.A" initially contributed this 
code so if they or anyone else has an idea of how to the said class, it 
would be great.

Best Regards,
Toyin Akin.

>From: Luigi Ballabio <lui...@gm...>
>To: Toyin Akin <toy...@ho...>
>Subject: Re: mcHullWhite Pricing Engine...
>Date: Fri, 05 May 2006 17:10:46 +0200
>
>
>On 05/02/2006 07:58:27 AM, Toyin Akin wrote:
>>Another question regarding lines #62 and #63 within the mcHullWhite  
>>pricing engine.cpp file.
>>
>>You use +1 and +2 as indexes into the path array.
>>
>>I am making the assumption that the path[] array stores the short  rates 
>>and path[0] is the short rate on the fixing date of the first  
>>caplet/floorlet (also the fixing rate to be used).
>
>No, path[0] is the current time (t = 0).
>
>As for your other questions, you might want to post on the mailing list  so 
>that you can reach the contributor of the processes. I haven't  looked very 
>hard into the implementation yet...
>
>Later,
>	Luigi
>
>
>----------------------------------------
>
>All generalizations are false, including this one.
>-- Mark Twain

Re: [Quantlib-dev] getting involved in Quantlib

[Luigi B. <lui...@gm...>]

On 05/10/2006 05:38:41 PM, Anne-Marie HUGUES wrote:
> I am Anne-Marie HUGUES fron Nice Sophia Antipolis University.
> I am responsible of  a Master degree in Computer science maths and
> Finance.
> My students have a 4 months project to work on from december to march
> I wonder if they could help you in coding some classes of the =20
> quantlib library?

Hello,
	I'm delighted to see that our library might be used in the =20
university---that was one of the objectives. Thanks for the bit of =20
news.  As for what to choose, I'm afraid I'll have to ask you some =20
questions about your students. At what programming level are they? What =20
is the syllabus of the course? Also, will they all work on the same =20
project or will you need different projects for different groups?

Later,
	Luigi


----------------------------------------

There are two ways to write error-free programs; only the third one =20
works.
-- unknown

[Quantlib-dev] getting involved in Quantlib

[Anne-Marie H. <am...@wa...>]

Hello
I am Anne-Marie HUGUES fron Nice Sophia Antipolis University.
I am responsible of  a Master degree in Computer science maths and =
Finance.
My students have a 4 months project to work on from december to march
I wonder if they could help you in coding some classes of the quantlib =
library?
Last year a group worked on your classes to see if they could be used by =
RFS together with their software Kondor +, th project was proposed by =
RFS
I find it could be more valuable for my students to actually contribute =
by developing some classes.
I have seen the todo list, how do we choose?

THanks=20
AM HUGUES
hu...@es...

[Quantlib-dev] QuantLib:0.3.12 bug and polishes

[Eugene S. <sh...@tb...>]

Hello All!

1. [Bug] I've got the error like "Tamas R Sashalmi"=20

Testing analytic Heston engine against Black formula...
quantlib/0.3.12/src/test-suite/hestonmodel.cpp(296): fatal error in =
"HestonModelTest::testAnalyticVsBlack": failed to reproduce Black price
calculated: 0.80067
expected: 0.80067
error: 1.071473e-007

It was on Windows only, it is ok on my 64-bit Linux.
I use: AMD Athlon(tm) 64 X2 Dual=20
         with: Microsoft Windows XP Pro (v.2002) SP2
          and: Visual Studio 2005

2. I use some extra compiler's options: -std=3Dgnu99 -pedantic -Wall =
-ansi
    So I get many errors during compilation:

2.1. C++ style comment in C sources.
      In files ql/RandomNumbers/primitivepolynomials.[ch]
      there are "// ... " comments used instead of "/* ... */".

2.2. Non ASCII symbols
      Also in files ql/RandomNumbers/primitivepolynomials.[ch]
      I've got many errors like this:
         error: stray '\169' in program
      It is because there are name in comment with strange symbol:
        "Monte Carlo Methods in Finance" by Peter J?kel

Thanks!

Sincerely yours,
Eugene Shevkoplyas

Trading Bricks
     =20
     =20
 

RE: [Quantlib-dev] Suggestion for sobol generator

[Richard G. <Ric...@ma...>]

Hi Luigi

I wrote the attached harness to perform exactly this test (should have sent=
 it with my code). I've tested skipping the first 1000 in the sequence up t=
o dimension 50 with no problems.

Regards
Richard


-----Original Message-----
From: Luigi Ballabio [mailto:lui...@gm...]
Sent: Thursday, 4 May 2006 3:51 PM
To: Richard Gould
Cc: qua...@li...
Subject: Re: [Quantlib-dev] Suggestion for sobol generator


On 2/6/06, Richard Gould <Ric...@ma...> wrote:
> I've been looking at using the Sobol number generator class in QuantLib f=
or
> running parallel monte-carlo simulations across many CPUs. I've modified =
the
> constructor to take an extra argument to the constructor to skip ahead in
> the sequence.

Richard,
    I finally got around to merge your code into the repository. It
compiles cleanly, but I had problems testing it. I expected that if I
skipped N samples and extracted 10 more, such 10 samples should be
equal to those obtained by initializing a generator without skipping,
extracting and discarding N samples, and extracting 10 more. This
should also hold in the degenerate case when N=3D0. However, this didn't
seem to be the case; the samples I obtained were somewhat different.
Can you check that your implementation passes such test and send me a
patch if it does?

Thanks,
    Luigi


NOTICE=20
The information contained in this email is confidential. If you are not the=
 intended recipient, you must not disclose or use the information in this e=
mail in any way. If you received it in error, please tell us immediately by=
 return email and delete the document.  We do not guarantee the integrity o=
f any e-mails or attached files and are not responsible for any changes mad=
e to them by any other person.

[Quantlib-dev] Fwd: Re: R-SWIG-Quantlib status report

[Joseph W. <jo...@gn...>]

Apologies if this is a duplicate, but the original message doesn't seem to=
=20
have made it to the sourceforge list.

=2D---------  =E8=BD=AC=E5=8F=91=E7=9A=84=E4=BF=A1=E4=BB=B6  ----------

Subject: Re: R-SWIG-Quantlib status report
Date: Wednesday 03 May 2006 13:56
=46rom: Joseph Wang <jo...@gn...>
To: William S Fulton <ws...@fu...>
Cc: Duncan Temple Lang <du...@wa...>, Dirk Eddelbuettel=20
<ed...@de...>, Luigi Ballabio <lui...@gm...>, swig-devel=20
<swi...@li...>

Just an update.

I got RSwig integrated with the test-suite, and after about two weeks of
 work, it currently compiles about 85% of the test cases.  The test cases
 which it doesn't compile seem not to be too difficult to fix.  They involve
 things like the fact that R doesn't like "in" as a variable name or has
 difficulties with functions that start with a _.

I was wondering what remains to be done in order to get R-Swig integrated
 with the mainline SWIG tree.  The current version in my tree seems to have
 all of the features that were in Prof. Temple Lang's version except for the
 copy to and from R functionality and that was excluded because it wasn't
 obvious how this would work with complex C++ data structures.  Also,
 assuming that there are not any major design issues, then I think that it
 would be useful to have R-Swig available in the SWIG distribution so that
 there is a wider audience for big fixing.  Assuming that there are no desi=
gn
 issues, I think R-SWIG is also at a point at which what is broken can be
 tracked with a bug tracking system.

If I could get a TODO list of the things that need to be done and design
issues that need to be resolved before R-Swig is integrated into the main
source tree, I can start working on those.

=2D------------------------------------------------------

Re: [Quantlib-dev] Re: Quantlib-dev digest, Vol 1 #441 - 2 msgs

[Luigi B. <lui...@gm...>]

On 04/20/2006 10:54:02 PM, TB...@ao... wrote:
>=20
>> My preferred approach  is to model the stock price less the PV of =20
>> the dividends and then add dividends  in. This is described in my =20
>> book. It has the advantage of being consistent with  the way =20
>> European options are valued using Black-Scholes and is widely used =20
>> in  practice.
>> John  Hull
>=20
> This means that what we had previously during early days of =20
> development of Convertible Bonds engine was correct.

Theo,
	apologies for the delay. Yes, the idea was correct---but I =20
think the implementation was not. However, go ahead and draft an =20
implementation.

Later,
	Luigi


----------------------------------------

I'd never join any club that would have the likes of me as a member.
-- Groucho Marx

[Quantlib-dev] [ quantlib-Feature Requests-1398900 ] libboost_unit_test_framework-vc80-mt-gd-1_33_1.lib

[SourceForge.net <no...@so...>]

Feature Requests item #1398900, was opened at 2006-01-07 00:09
Message generated for change (Settings changed) made by lballabio
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=362740&aid=1398900&group_id=12740

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
>Status: Closed
Priority: 5
Submitted By: pravinster (pravinster)
Assigned to: Nobody/Anonymous (nobody)
Summary: libboost_unit_test_framework-vc80-mt-gd-1_33_1.lib

Initial Comment:
Hi,

I am relatively new to Quantlib and Microsoft C++ IDEs.
So, kindly bear with me.

I built the Boost libraries and included the libraries
at proper places within the VC-2005 IDE. However, when
I try to compile the test-suite project, I receive an
error that,

1>LINK : fatal error LNK1104: cannot open file
'libboost_unit_test_framework-vc80-mt-gd-1_33_1.lib'

Apparently, this file was not built when I built boost
using bjam, since this file is nowhere on my computer.
I cannot find any documentation regarding this type of
error from Boost.org website. Does anyone have this
file? Or, do you know how to build it, or where to get
it from? I would really appreciate any advice/help. Thanks.

-Pravin

PS: If anyone would be so kind, you could e-mail this
file to me at pc...@ho.... Thanks.


----------------------------------------------------------------------

Comment By: Ferdinando Ametrano (nando)
Date: 2006-01-07 10:54

Message:
Logged In: YES 
user_id=34616

You have to download and properly "install" the Microsoft
platform SDK along with VC-2005. Then build the Boost
libraries again and
libboost_unit_test_framework-vc80-mt-gd-1_33_1.lib will be
built.

hope it helps

ciao -- Nando

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=362740&aid=1398900&group_id=12740

[Quantlib-dev] [ quantlib-Feature Requests-1469488 ] Unadjusted month end business day convention

[SourceForge.net <no...@so...>]

Feature Requests item #1469488, was opened at 2006-04-12 22:12
Message generated for change (Comment added) made by lballabio
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=362740&aid=1469488&group_id=12740

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
>Status: Closed
Priority: 5
Submitted By: Nobody/Anonymous (nobody)
>Assigned to: Luigi Ballabio (lballabio)
Summary: Unadjusted month end business day convention

Initial Comment:
There is no business day convention available that 
conforms to the convention of the US 2 year treasury 
note - which is the last day of the month unadjusted 
for weekends or holidays.

I corrected this for myself by adding 
UnadjustedMonthEnd to the BusinessDayConvention and 
changing the Calendar::adjust function by replacing:

  if (c == Unadjusted)
    return d;

with:

  if(c == UnadjustedMonthEnd && origin != Date() &&   
isEndOfMonth(origin))
    return Date::endOfMonth(d);

  if (c == Unadjusted || c == UnadjustedMonthEnd)
    return d;


----------------------------------------------------------------------

Comment By: Luigi Ballabio (lballabio)
Date: 2006-05-04 14:48

Message:
Logged In: YES 
user_id=75450

The feature is now implemented in CVS and will be available in the 
next library release.


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=362740&aid=1469488&group_id=12740

Re: [Quantlib-dev] Suggestion for sobol generator

[Luigi B. <lui...@gm...>]

On 2/6/06, Richard Gould <Ric...@ma...> wrote:
> I've been looking at using the Sobol number generator class in QuantLib f=
or
> running parallel monte-carlo simulations across many CPUs. I've modified =
the
> constructor to take an extra argument to the constructor to skip ahead in
> the sequence.

Richard,
    I finally got around to merge your code into the repository. It
compiles cleanly, but I had problems testing it. I expected that if I
skipped N samples and extracted 10 more, such 10 samples should be
equal to those obtained by initializing a generator without skipping,
extracting and discarding N samples, and extracting 10 more. This
should also hold in the degenerate case when N=3D0. However, this didn't
seem to be the case; the samples I obtained were somewhat different.
Can you check that your implementation passes such test and send me a
patch if it does?

Thanks,
    Luigi

Re: [Quantlib-dev] QuantLib developement

[Toyin A. <toy...@ho...>]

Hi,

Sorry, silly request, G2 is a 2-factor model !!

Toy out.


>From: "Toyin Akin" <toy...@ho...>
>To: lui...@gm..., TB...@ao...
>CC: qua...@li...
>Subject: Re: [Quantlib-dev] QuantLib developement
>Date: Fri, 28 Apr 2006 15:14:41 +0100
>
>
>Hi Luigi,
>
>Is it possible to derive the G2Process and G2ForwardProcess classes from 
>the XXX1D base class as you have done with the HullWhite versions?
>
>Toy out.
>
>>From: "Luigi Ballabio" <lui...@gm...>
>>To: "TB...@ao..." <TB...@ao...>
>>CC: qua...@li...,"Toyin Akin" 
>><toy...@ho...>
>>Subject: Re: [Quantlib-dev] QuantLib developement
>>Date: Fri, 28 Apr 2006 13:11:02 +0200
>>
>>On 4/11/06, Luigi Ballabio <lui...@gm...> wrote:
>>>I'll be committing shortly a contribution I received. It provides
>>>processes based on Hull-White that can be used in a Monte Carlo model.
>>>I'll let you know when they're available.
>>
>>Done. There are a few new processes in ql/Processes and a new engine
>>in ql/PricingEngines/CapFloor that shows how to use them.
>>
>>Luigi
>>
>>
>>-------------------------------------------------------
>>Using Tomcat but need to do more? Need to support web services, security?
>>Get stuff done quickly with pre-integrated technology to make your job 
>>easier
>>Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>>http://sel.as-us.falkag.net/sel?cmd=lnk&kid0709&bid&3057&dat1642
>>_______________________________________________
>>Quantlib-dev mailing list
>>Qua...@li...
>>https://lists.sourceforge.net/lists/listinfo/quantlib-dev
>
>
>
>
>-------------------------------------------------------
>Using Tomcat but need to do more? Need to support web services, security?
>Get stuff done quickly with pre-integrated technology to make your job 
>easier
>Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
>_______________________________________________
>Quantlib-dev mailing list
>Qua...@li...
>https://lists.sourceforge.net/lists/listinfo/quantlib-dev

Re: [Quantlib-dev] Implementing high-low volatility model

[Joseph W. <jo...@co...>]

On Wednesday 26 April 2006 02:47=EF=BC=8Cyou wrote=EF=BC=9A

> while I don't have any special contribution on your design I just
> wonder why you are considering the effort of adding this framework to
> QuantLib. Isn't R a better alternative for this kind of work?

=46unny that you should mention this since I am using R for this=20
work. :-) :-) :-)

I have my R-SWIG interface to QuantLib working well enough so that I can=20
switch transparently between C++ and R.  For example, it has a type convert=
er=20
so that an R vector of doubles gets automatically converted to a vector of=
=20
doubles which the C++ engine runs on, and then the result gets automaticall=
y=20
converted back to a data.frame which can then be plotted.

Originally, I was planning to do the coding for GARCH and those models in R=
 or=20
use some package like Rmetrics, but the thing that I found when I started=20
doing that is that once you have more than about 20 lines of code, you run=
=20
into code management and reusability issues.  The good and bad thing with R=
=20
is that it is very flexible, by





> QuantLib focus has always been on derivatives, and while econometric
> estimations are often used as input in derivatives models, the two
> domains are almost well defined...
>
> ciao -- Nando
>
> On 4/25/06, Joseph Wang <jo...@gn...> wrote:
> > I'm in the process of trying to implement a class that does a
> > Garman-Klass estimation of volatility based on high-low data.
> >
> > Right now what I have in mind is a class called IntervalQuote(?) that
> > contains open, close, high, low information, and this will be used with
> > the TimeSeries template to produce
> >
> > TimeSeries<IntervalQuote>
> >
> > which will be the input into the calculation classes.  There will be a
> > helper class that creates TimeSeries<IntervalQuote> from a vector of
> > dates, open, close, high, and low data.
> >
> > It seems that one should break up the VolatilityModel into two parts.=20
> > One part is LocalEstimator (?), the second part combinings the daily
> > estimation into a time series using constant combining or GARCH, which
> > would be subclasses of EstimationCombiner (?)
> >
> > Since the LocalEstimator will need different types of inputs, there may
> > be a need to create a trait.
> >
> > Thoughts?  I'm especially interested in feedback as to getting the nami=
ng
> > conventions right.
> >
> >
> >
> >
> >
> > -------------------------------------------------------
> > Using Tomcat but need to do more? Need to support web services, securit=
y?
> > Get stuff done quickly with pre-integrated technology to make your job
> > easier Download IBM WebSphere Application Server v.1.0.1 based on Apache
> > Geronimo
> > http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=
=3D121642
> > _______________________________________________
> > Quantlib-dev mailing list
> > Qua...@li...
> > https://lists.sourceforge.net/lists/listinfo/quantlib-dev