MonotonicLogCubicNaturalSpline doesn't preserve monotonicity

[Jag]

I am using MonotonicLogCubicNaturalSpline with extrapolation set to TRUE, and it seems to lose monotonicity between the first & second X-Value.

The input values are all strictly decreasing. But the output is decreasing until it reaches the second X-Value[183], at which point it increases from 0.999045 to 0.999354. The output is not even matching at the first X Value. Input for X=91 is 0.999657 but, the interpolation is sending back 0.999357.
Please see the input & output below.

Any help will be much appreciated!


Input:
X Value [91] -  Y Value [0.999657]
X Value [183] - Y Value [0.999354]
X Value [274] - Y Value [0.999045]
X Value [365] - Y Value [0.998726]
X Value [457] - Y Value [0.99837]
X Value [548] - Y Value [0.997918]
X Value [639] - Y Value [0.997302]
X Value [731] - Y Value [0.996525]
X Value [1096] - Y Value [0.991699]
X Value [1461] - Y Value [0.982502]
X Value [1826] - Y Value [0.968062]
X Value [2192] - Y Value [0.94841]
X Value [2557] - Y Value [0.92509]
X Value [2922] - Y Value [0.899574]
X Value [3287] - Y Value [0.872604]
X Value [3653] - Y Value [0.844616]
X Value [4383] - Y Value [0.78806]
X Value [5479] - Y Value [0.708377]
X Value [7305] - Y Value [0.59849]
X Value [9131] - Y Value [0.509214]
X Value [10958] - Y Value [0.43323]
X Value [14610] - Y Value [0.325498]
X Value [18263] - Y Value [0.258686]


Output:
[1] - 0.999653
[2] - 0.99965
[3] - 0.999647
[4] - 0.999644
[5] - 0.99964
[6] - 0.999637
[7] - 0.999634
[8] - 0.999631
[9] - 0.999627
.
.
.
[87] - 0.99937
[88] - 0.999367
[89] - 0.999364
[90] - 0.99936
[91] - 0.999357
[92] - 0.999354
.
.
.
.
[180] - 0.999053
[181] - 0.999049
[182] - 0.999045
[183] - 0.999354
[184] - 0.99935
[185] - 0.999347
[186] - 0.999344
[187] - 0.99934

[Ferdinando M. Ametrano-3]

Hi Jag,

you must be doing something wrong, last but not least providing too little information to really help you.

please find attached an Excel workbook providing monotonic interpolation evidence on your data. Use any recent QLXL with it

ciao -- Nando

On Fri, Jan 18, 2013 at 12:31 PM, Jag <[hidden email]> wrote:

I am using MonotonicLogCubicNaturalSpline with extrapolation set to TRUE, and
it seems to lose monotonicity between the first & second X-Value.

The input values are all strictly decreasing. But the output is decreasing
until it reaches the second X-Value[183], at which point it increases from
0.999045 to 0.999354. The output is not even matching at the first X Value.
Input for X=91 is 0.999657 but, the interpolation is sending back 0.999357.
Please see the input & output below.

Any help will be much appreciated!


Input:
X Value [91] -  Y Value [0.999657]
X Value [183] - Y Value [0.999354]
X Value [274] - Y Value [0.999045]
X Value [365] - Y Value [0.998726]
X Value [457] - Y Value [0.99837]
X Value [548] - Y Value [0.997918]
X Value [639] - Y Value [0.997302]
X Value [731] - Y Value [0.996525]
X Value [1096] - Y Value [0.991699]
X Value [1461] - Y Value [0.982502]
X Value [1826] - Y Value [0.968062]
X Value [2192] - Y Value [0.94841]
X Value [2557] - Y Value [0.92509]
X Value [2922] - Y Value [0.899574]
X Value [3287] - Y Value [0.872604]
X Value [3653] - Y Value [0.844616]
X Value [4383] - Y Value [0.78806]
X Value [5479] - Y Value [0.708377]
X Value [7305] - Y Value [0.59849]
X Value [9131] - Y Value [0.509214]
X Value [10958] - Y Value [0.43323]
X Value [14610] - Y Value [0.325498]
X Value [18263] - Y Value [0.258686]


Output:
[1] - 0.999653
[2] - 0.99965
[3] - 0.999647
[4] - 0.999644
[5] - 0.99964
[6] - 0.999637
[7] - 0.999634
[8] - 0.999631
[9] - 0.999627
.
.
.
[87] - 0.99937
[88] - 0.999367
[89] - 0.999364
[90] - 0.99936
[91] - 0.999357
[92] - 0.999354
.
.
.
.
[180] - 0.999053
[181] - 0.999049
[182] - 0.999045
[183] - 0.999354
[184] - 0.99935
[185] - 0.999347
[186] - 0.999344
[187] - 0.99934



--
View this message in context: http://quantlib.10058.n7.nabble.com/MonotonicLogCubicNaturalSpline-doesn-t-preserve-monotonicity-tp13926.html
Sent from the quantlib-users mailing list archive at Nabble.com.

------------------------------------------------------------------------------
Master HTML5, CSS3, ASP.NET, MVC, AJAX, Knockout.js, Web API and
much more. Get web development skills now with LearnDevNow -
350+ hours of step-by-step video tutorials by Microsoft MVPs and experts.
SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122812
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

------------------------------------------------------------------------------
Master HTML5, CSS3, ASP.NET, MVC, AJAX, Knockout.js, Web API and
much more. Get web development skills now with LearnDevNow -
350+ hours of step-by-step video tutorials by Microsoft MVPs and experts.
SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122812
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users