Define the RMS Response Calculation Method of Random Vibration

I am currently using msc nastran for random vibration analysis.
With RANDPS, TABRND1(log) and paran,rmsint,loglog,
I calculate the rms response by Log-Log interpolation.

The manual explains RMSINT as follows.

RMSINT
Default = LINEAR for the trapezoidal approximation.
RMSINT specifies the interpolation method for numerical integration when computing both RMS (Root
Mean Square) and N0 (Number of Zero Crossings or Mean Frequency) from PSDF (Power Spectral Density
Function). RMSINT = LINEAR requests the trapezoidal approximation, which is the existing Nastran
approach. RMSINT = LOGLOG requests the Log-Log interpolation

I set the conditions in optistruct to get the same result,
but all responses show linear results.
(I can't find the card called rmsint in optistruct)

Generally, since the load is applied in the log scale in the test, I think it is right to calculate it with log-log interpolation.

Is there a card that can reproduce this similarly?

I want to do all the analysis with optistruct.

If you have experience, please help me.

Thank you