Decorrelated Linear Power Spectrum of PSCz 0.6 Jy with high-latitude mask
(Hamilton A. J. S., Tegmark M., 2000, MNRAS, in press, astro-ph/0008392).
http://casa.colorado.edu/~ajsh/pscz/

When fitting to theoretical models at linear scales, this decorrelated
power spectrum is to be preferred over the correlated power spectrum,
since the decorrelated estimates can be treated as uncorrelated.

It is not possible to decorrelate the nonlinear power spectrum.
If you want a power spectrum with uncorrelated error bars over
the full range of wavenumbers, you might consider using the prewhitened
power spectrum, which is less correlated than the power spectrum itself.

The definition of power follows the Peebles (1980) convention
P(k) = int e^{i k.r} xi(r) d^3 r
where xi(r) is the correlation function.

k is the median wavenumber (in h/Mpc) of the band-power window.
k- and k+ are the wavenumbers (in h/Mpc) where the band-power window
falls to half its maximum.  The median and half-maximum points
are those of the scaled and discretized band-power windows as defined in
Hamilton A. J. S., Tegmark M., 2000, MNRAS, 312, 285 (astro-ph/9905192).
P(k) is the estimated power (in h^-3 Mpc^3) in the band-power,
and DeltaP(k) (in h^-3 Mpc^3) its 1-sigma uncertainty.

    k     k-     k+     P(k) DeltaP(k)
.0137   .0097   .0171   133000. 920000.
.0175   .0130   .0219   20200.  54200.
.0214   .0165   .0264   -11100. 21300.
.0249   .0200   .0297   36600.  21400.
.0280   .0232   .0330   36600.  16600.
.0319   .0268   .0376   5580.   13200.
.0366   .0308   .0434   8250.   10800.
.0422   .0365   .0492   11700.  9100.
.0485   .0423   .0561   19400.  7600.
.0560   .0491   .0635   10400.  6000.
.0646   .0569   .0731   4680.   4550.
.0747   .0670   .0833   10600.  3400.
.0863   .0783   .0947   6490.   2520.
.0998   .0902   .110    4630.   1750.
.115    .106    .126    5930.   1270.
.133    .123    .144    2400.   970.
.154    .143    .165    2990.   750.
.178    .166    .190    2980.   570.
.205    .192    .219    1650.   410.
.237    .221    .254    963.    266.
.274    .257    .292    929.    211.
.316    .298    .335    927.    189.