failure in SanityCheck (trac #407)

cloudy-bot commented 6 years ago

reported by: peter

The attached case is a simplified script sent by Jacopo Fritz. It needs to be combined with a modified version of continuum_mesh.ini (also attached). When run with the current head of the trunk, it fails very quickly in !SanityCheck due to an opacity being zero. Jacopo Fritz noted the same problem in c17.00, so it very likely also exists on c17_branch.

Migrated from https://www.nublado.org/ticket/407

{
    "status": "closed",
    "changetime": "2019-02-26T14:38:03Z",
    "_ts": "1551191883534378",
    "description": "The attached case is a simplified script sent by Jacopo Fritz. It needs to be combined with a modified version of continuum_mesh.ini (also attached). When run with the current head of the trunk, it fails very quickly in !SanityCheck due to an opacity being zero. Jacopo Fritz noted the same problem in c17.00, so it very likely also exists on c17_branch.",
    "reporter": "peter",
    "cc": "",
    "resolution": "fixed",
    "time": "2018-01-22T11:39:59Z",
    "component": "atomic/molecular data base",
    "summary": "failure in SanityCheck",
    "priority": "major",
    "keywords": "",
    "version": "trunk",
    "milestone": "c17.02",
    "owner": "nobody",
    "type": "defect - failed assert"
}

cloudy-bot commented 6 years ago

input script Attachment: jacopo.in

cloudy-bot commented 6 years ago

modified mesh definition Attachment: continuum_mesh.ini

cloudy-bot commented 6 years ago

@CloudyLex commented:

Milestone renamed

@CloudyLex changed milestone from "c17.01 release" to "C17.01 release"

cloudy-bot commented 6 years ago

@CloudyLex commented:

Milestone renamed

@CloudyLex changed milestone from "C17.01 release" to "C17.01"

cloudy-bot commented 6 years ago

@CloudyLex changed milestone from "C17.01" to "C17.02"

cloudy-bot commented 6 years ago

@peter-van-hoof-noaccount commented:

Milestone renamed

@peter-van-hoof-noaccount changed milestone from "C17.02" to "c17.02"

cloudy-bot commented 6 years ago

@peter-van-hoof-noaccount commented:

The code complains about the photoionization opacity being zero near the threshold. These are evaluated by Dima's ADfA routine phfit(). The ADfA data internally use eV as the energy unit. When the threshold is being extracted from the table using the routine ph1(), the values are converted from eV to Ryd by multiplying with a factor 0.9998787/EVRYD. The source of the numerical constant is unclear. When supplying a photon energy to phfit() however, the constant 0.9998787 is missing and we use only rfield.anu(ip)*EVRYD. This is inconsistent, leading to zero opacities near threshold.

cloudy-bot commented 6 years ago

@peter-van-hoof-noaccount commented:

The constant 0.9998787 seems inappropriate and has been removed on the trunk in r12609.

cloudy-bot commented 6 years ago

@peter-van-hoof-noaccount commented:

The changed wavelengths for the RRC "lines" were fixed in the test suite on the trunk in r12610.

The changes were merged to c17_branch in r12611.

@peter-van-hoof-noaccount changed resolution from "" to "fixed"
@peter-van-hoof-noaccount changed status from "new" to "closed"

cloudy-bot commented 6 years ago

@peter-van-hoof-noaccount commented:

The fix on c17_branch was partially reversed in r12613. In that commit the code was modified to generate the same wavelengths as in c17.01 for RRC entries on the line stack. This does not hamper the bug fix, and avoids breaking existing scripts. The changes to the test suite scripts were also reverted as they are no longer needed.

cloudy-astrophysics / bug-tracker-migration-test

failure in SanityCheck (trac #407) #409