GammaLib

• GCTAResponse::irf
• GCTAResponse::irf_radial
• GCTAResponse::irf_elliptical
• GCTAResponse::irf_diffuse

The GCTAResponse::irf should use the following code for average response handling:

    // Retrieve average response references
    const GCTAExposure& exposure = ... (where ever we get the exposure from)
    const GCTAMeanPsf&  psf = ... (where ever we get the PSF from)

    // Get photon attributes
    const GSkyDir& srcDir  = photon.dir();
    const GEnergy& srcEng  = photon.energy();
    const GTime&   srcTime = photon.time();

    // Compute only if we're sufficiently close to PSF
    if (delta <= delta_max) {

        // Get effective area component
        irf = exposure(srcDir, srcEng);

        // Multiply-in PSF
        if (irf > 0.0) {

            // Get PSF component
            irf *= psf(srcDir, delta, srcEng);

            // Apply deadtime correction
            irf *= obs.deadc(srcTime);

        } // endif: exposure was non-zero

    } // endif: we were sufficiently close to PSF

The GCTAResponse::irf_radial, GCTAResponse::irf_elliptical and GCTAResponse::irf_diffuse methods need a more fundamental re-organisation as the integrations are performed in a system centred on the pointing, hence all coordinate transformations are pointing dependent.

Maybe the best thing would be to introduce a new class, called for example GCTACubeResponse, that handles the cube-style response. It would then be upon reading (e.g. GCTAObservation::read) where the response type would be defined.

• GCTAResponseIrf (which is the old GCTAResponse)
• GCTAResponseCube (which is the new cube-style IRF)

I’m working now on refactoring the code ...