Failed test in CVS

[Joseph Wang]

I'm getting this error when I run the test suite.  Any ideas as to where
to start
looking for a fix?

digitaloption.cpp(735): fatal error in
"DigitalOptionTest::testMCCashAtHit": American put option with
cash-or-nothing payoff:
    spot value: 105.000000
    strike:           100.000000
    dividend yield:   20.00000%
    risk-free rate:   10.00000%
    reference date:   November 1st, 2004
    maturity:         April 30th, 2005
    volatility:       20.00000%

    expected   value: 12.271500
    calculated value: 12.280527
    error:            0.009027
    tolerance:        0.001000

Tests completed in 907.46 s

[Luigi Ballabio-2]

At 18:47 01/11/2004, Joseph Wang wrote:

>I'm getting this error when I run the test suite.  Any ideas as to where
>to start looking for a fix?
>
>digitaloption.cpp(735): fatal error in
>"DigitalOptionTest::testMCCashAtHit": American put option with
>cash-or-nothing payoff:
>    spot value: 105.000000
>    strike:           100.000000
>    dividend yield:   20.00000%
>    risk-free rate:   10.00000%
>    reference date:   November 1st, 2004
>    maturity:         April 30th, 2005
>    volatility:       20.00000%
>
>    expected   value: 12.271500
>    calculated value: 12.280527
>    error:            0.009027
>    tolerance:        0.001000
>
>Tests completed in 907.46 s

... [show rest of quote]

Joseph,
         what QuantLib version would this be?

Later,
         Luigi

[Joseph Wang]

It's the version in CVS.

Luigi Ballabio wrote:

> At 18:47 01/11/2004, Joseph Wang wrote:
>
>> I'm getting this error when I run the test suite.  Any ideas as to
>> where to start looking for a fix?
>>
>> digitaloption.cpp(735): fatal error in
>> "DigitalOptionTest::testMCCashAtHit": American put option with
>> cash-or-nothing payoff:
>>    spot value: 105.000000
>>    strike:           100.000000
>>    dividend yield:   20.00000%
>>    risk-free rate:   10.00000%
>>    reference date:   November 1st, 2004
>>    maturity:         April 30th, 2005
>>    volatility:       20.00000%
>>
>>    expected   value: 12.271500
>>    calculated value: 12.280527
>>    error:            0.009027
>>    tolerance:        0.001000
>>
>> Tests completed in 907.46 s
>
>
>
> Joseph,
>         what QuantLib version would this be?
>
> Later,
>         Luigi
>

... [show rest of quote]

[Luigi Ballabio-2]

At 23:23 02/11/2004, Joseph Wang wrote:
>It's the version in CVS.

Joseph,
         I'm afraid I can't reproduce the failure. May you give me a few
more info, namely,

a) is it still failing? (Although unlikely, it might have been a November 1st
    thing)
b) What platform are you working

[Luigi Ballabio-2]

Ouch, wrong key. Here we go again:

At 23:23 02/11/2004, Joseph Wang wrote:
>It's the version in CVS.

Joseph,
         I'm afraid I can't reproduce the failure. May you give me a few
more info, namely,

a) is it still failing? (Although unlikely, it might have been a November 1st
    thing)
b) What platform are you working on? What compiler?

Later,
         Luigi

[Luigi Ballabio-2]

At 12:50 03/11/2004, Luigi Ballabio wrote:
>At 23:23 02/11/2004, Joseph Wang wrote:
>>It's the version in CVS.
>
>Joseph,
>         I'm afraid I can't reproduce the failure.

Joseph,
         I take it back---I serendipitously managed to reproduce it by
changing the mti member of the Mersenne Twister generator from static to
mutable. Did you do anything of the sort?

Nando, Neil (there's your name on top of the related files: )
a) where do the test data come from? With what accuracy?
b) was the internal sequence generator in the McDigitalEngine supposed to
    interact with the main generator through mti? I sure hope not...

Later,
         Luigi

[Joseph Wang]

Actually no.  I'm compiling this out of the box on a Linux
dell laptop using Mandrake cooker and gcc-c++ 3.4.
I'll look at the random number generator.

BTW, I've uploaded an autogen.sh file onto the sourceforge
patches that automatically does the configure.

Luigi Ballabio wrote:

> At 12:50 03/11/2004, Luigi Ballabio wrote:
>
>> At 23:23 02/11/2004, Joseph Wang wrote:
>>
>>> It's the version in CVS.
>>
>>
>> Joseph,
>>         I'm afraid I can't reproduce the failure.
>
>
> Joseph,
>         I take it back---I serendipitously managed to reproduce it by
> changing the mti member of the Mersenne Twister generator from static
> to mutable. Did you do anything of the sort?
>
> Nando, Neil (there's your name on top of the related files: )
> a) where do the test data come from? With what accuracy?
> b) was the internal sequence generator in the McDigitalEngine supposed to
>    interact with the main generator through mti? I sure hope not...
>
> Later,
>         Luigi

... [show rest of quote]

[Joseph Wang]

Luigi Ballabio wrote:

> Oh, it was you then. I had to reject the patch because it was
> anonymous and  it looked like it might have been copyrighted (it had
> references to libxml,  so I suspected it was taken from another project.)


Actually I did copy it from libxml, but libxml is copyrighted under the MIT
license, which compatible with the BSD license, so using that file for
quantlib
is o.k.

>
> Also, we have a bootstrap script in the repository that invokes the  
> autotools. Asking out of my ignorance, is it common practice to call
> such a  script autogen.sh? Does it usually include the configure part?
>
Yes I believe this is the case.

> Later,
>     Luigi
>
>

[Neil Firth]

Hi,

It is a bit strange that it has just started failing, but as it is a MC
method for a difficult test case, and not that far out I wouldn't worry
too much. Try incresing the number of simulations and see whether it
converges, or reducing the tolerance. Let me know how you get on.

Neil



Joseph Wang wrote:

> I'm getting this error when I run the test suite.  Any ideas as to
> where to start
> looking for a fix?
>
> digitaloption.cpp(735): fatal error in
> "DigitalOptionTest::testMCCashAtHit": American put option with
> cash-or-nothing payoff:
>    spot value: 105.000000
>    strike:           100.000000
>    dividend yield:   20.00000%
>    risk-free rate:   10.00000%
>    reference date:   November 1st, 2004
>    maturity:         April 30th, 2005
>    volatility:       20.00000%
>
>    expected   value: 12.271500
>    calculated value: 12.280527
>    error:            0.009027
>    tolerance:        0.001000
>
> Tests completed in 907.46 s
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Sybase ASE Linux Express Edition - download now for FREE
> LinuxWorld Reader's Choice Award Winner for best database on Linux.
> http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
> _______________________________________________
> Quantlib-users mailing list
> [hidden email]
> https://lists.sourceforge.net/lists/listinfo/quantlib-users
>

... [show rest of quote]