In all cases the analytic approximations for the prices have been
used. The Libor factorloadings are based on all combinations of volatility surfaces
(CONST, JR, M) and correlation types (JR, CS) and have been generated by the method
LiborFactorLoading::sample
which provides default values for most
parameters.
The prices are intended to be used for cross calibration experiments (between differently configured LMMs).
LiborCalibrator
expects.
Here Caplet(i)
is a caplet on the interval [T_i,T_{i+1}]
and Swaption(p,q)
exercises at time T_p into a swap on the interval [T_p,T_q].
The files contain the prices of all caplets Caplet(i)
and coterminal swaptions
Swaption(i,n)
which exercise into a swap terminating at the horizon T_n where
i=1,...,n-1.
n | dimension (number of Libor accrual intervals) |
Instrument | "Caplets", "Swaptions" |
LmmType | "DL" (DriftlessLMM), "PC" (PredictorCorrectorLMM) |
VolSurfaceType | "CONST", "JR", "M" (see class VolSurface) |
CorrelationType | "JR", "CS" (see class Correlations) |
Thus for example CapletsIn-DL-dim40-JR-CS.txt
is the file containing
the caplet prices computed in a DriftlessLMM
of dimension 40
(40 accrual periods) with Jaeckel-Rebonato (JR) volatility surface and
Coffee-Shoenmakers (CS) correlations.
The calibration routine expects such a file and the corresponding file with swaption prices
(in the same dimension -- not checked!) in the directory
$DEV_ROOT/src/martingale
and then writes corresponding files
CapletsOut, SwaptionsOut
with the relative calibration errors in percent.