Subclassing (the RateHelper) in Python
Locked
 
|
Hi,
I am trying to subclass the RateHelper using python, so that the derived class (Inflation Linked Bond) can be used within the bootstrapper. I am struggling my way through c++, swig and the wrapper code, and so far I failed miserably with the error: NotImplementedError: No matching function for overloaded 'new_PiecewiseFlatForward' This error normally relates to wrong type of constructor arguments, but in this case it seems more complex and I gave up trying to understand the wrapper code. I then tried the "easier" task of subclassing (in python) the SwapRateHelper, overriding some of the methods for testing purposes. I have no problem using derived classes in other python code, but when instances of such classes are consumed within QL, the overridden methods are not called, the ones of the base class are used instead :-( E.g. in the bootstrapper PiecewiseFlatForward::performCalculations(), the overridden discountGuess() is never called. This is a bit of a show stopper. Can this type of subclassing be done in Python or am I wasting my time? If yes, how? Or should I write c++ code and then use swig? Or some other python extension technique? What have others done so far? |
Re: Subclassing (the RateHelper) in Python
|
|
On 10/14/2005 11:24:10 AM, ago wrote:
> I am trying to subclass the RateHelper using python, so that the > derived class (Inflation Linked Bond) can be used within the > bootstrapper. I am struggling my way through c++, swig and the > wrapper code, and so far I failed miserably with the error: > > NotImplementedError: No matching function for overloaded > 'new_PiecewiseFlatForward' I guess the wrappers didn't recognize instances of your class as rate helpers. What was the interface you exported? Can you post the relevant bit of the .i file? > I then tried the "easier" task of subclassing (in python) the > SwapRateHelper, overriding some of the methods for testing purposes. That won't work. SWIG must be told to generate special code for that (see the SWIG documentation for "directors".) However, I'm under the impression that it would not work because we're exporting smart pointers rather than the actual classes. Later, Luigi ---------------------------------------- To err is human -- to blame it on a computer is even more so. -- unknown |
Re: Subclassing (the RateHelper) in Python
|
|
Hi Luigi, I am porting the cpp testsuite In Java. I am looking for covering the entire testsuite and the work is progressing rapidly. Luigi, Do you think it will have a value in putting this in the CVS base sometime. A significant part of the test suite passes already in Java :) I need a help from a SWIG Interface person to help me get through. In fact, I don't find the interface for the following cases : - the capfloor ImpliedVolatility method in the wrappers. Capfloor has been split into its sub-classes, Cap, Floor and Collar but no Capfloor where ImpliedVolatility id defined. - In CompoundForward class, there is no methods corresponding to the C++ method : Rate compoundForward(const Date& d1, Integer f, bool extrapolate = false) const; Rate compoundForward(Time t1, Integer f, bool extrapolate = false) const; - The Class GapPayoff : public StrikedTypePayoff is not interfaced in SWIG - The Class CapletConstantVolatility and its super class CapletVolatilitystructure are not interfaced. - The Class InarrearIndexedCoupon and Itssuper class IndexedCoupon are not interfaced. Nabil On 10/14/05, Luigi Ballabio <[hidden email]> wrote:
|
ParCoupon and Calendar problem
|
|
Hello, all.
I have the following funny situation which I don't know how to fix. I am building a par curve using a combination of depos and swaps. For the swaps, I am using a convention of Unadjusted on both the fixed and the floating legs. However, this causes a problem when the swap maturity date is on a weekend. I stepped through the code and found the following in parcoupon.cpp in the ParCoupon::amount() method: Date temp = index_->calendar().advance(accrualEndDate_, -fixingDays_, Days); DiscountFactor endDiscount = termStructure->discount(index_->calendar().advance( temp, index_->settlementDays(), Days)); The problem is that the calendar().advance() takes a default dayCountConvention of Following. The result is that my end date (temp) becomes the following Monday. This causes an exception to be thrown when executing YieldTermStructure::discount(const Date& d, bool extrapolate) due to the checkRange() method failing. I'd like to know what's the best way to fix this. thanks in advance Sid __________________________________ Yahoo! Music Unlimited Access over 1 million songs. Try it free. http://music.yahoo.com/unlimited/ |
Re: ParCoupon and Calendar problem
|
|
On 10/14/2005 05:38:40 PM, Siddharth Sharma wrote:
> Hello, all. > I have the following funny situation which I don't > know how to fix. I am building a par curve using a > combination of depos and swaps. For the swaps, I am > using a convention of Unadjusted on both the fixed and > the floating legs. However, this causes a problem when > the swap maturity date is on a weekend. > ... > I'd like to know what's the best way to fix this. Siddhart, I guess you could modify the ParCoupon so that it takes a business-day convention and uses it for adjustment. Later, Luigi ---------------------------------------- There is no opinion so absurd that some philosopher will not express it. -- Marcus Tullius Cicero, "Ad familiares" |
«
Return to quantlib-users
|
1 view|%1 views
| Free forum by Nabble | Edit this page |