Pass `actualstep` to `b4step2`

[mandli]

When playing with some debugging runs I found that actualstep is not being passed to b4step correctly. It used to be passed to stepgrid and then on to b4step, but moving b4step to advanc makes that not work anymore. I think I have a fix, but this does mean acutalstep is no longer available in stepgrid.

[mandli]

I also noticed that prepbigstep calls stepgrid with actualstep = .false.. I feel like b4step should also be called here to better mimic what is needed here, but just wanted to check.

[mjberger]

I don't think the actualstep thing is problem. b4step call might be. The only thing b4step2 does if actualstep is true is update topo if the aux arrays aren't yet finalized. This does not happen if actualstep is false. So the value of actualstep isn't an issue. The call to b4step2 might be an issue though if the other things it does still need to be done, such as setting storm fields? Is this needed, or is it created in prepbigstep already by calling setaux and then coarsening? The other stuff, like checking for nans and speed limit checking might not be relevant for 1 step of prepbigstep taken for error estimation? I would put a note in prepbigstep that says to think about putting in a call to b4step2 before calling stepgrid if it's actually needed, but it doesn't appear to be.

…

On Aug 27, 2025, at 4:38 PM, Kyle Mandli ***@***.***> wrote: mandli left a comment (clawpack/geoclaw#667) <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3229673985&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=gWKGdCt2S9mkfoZPekCciYcg51z9yB5rHRaEQLqhji7xxvUA3OukA5HAQvF-eYBV&s=_gDWSkD_JXL8-rTg4xrEPrJshlVbEXMNlgwBjxFx2rw&e=> I also noticed that prepbigstep calls stepgrid with actualstep = .false.. I feel like b4step should also be called here to better mimic what is needed here, but just wanted to check. — Reply to this email directly, view it on GitHub <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3229673985&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=gWKGdCt2S9mkfoZPekCciYcg51z9yB5rHRaEQLqhji7xxvUA3OukA5HAQvF-eYBV&s=_gDWSkD_JXL8-rTg4xrEPrJshlVbEXMNlgwBjxFx2rw&e=>, or unsubscribe <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AAGUGC4FFIKXGIQY7R32KCT3PYJL7AVCNFSM6AAAAACE7KSCBCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEMRZGY3TGOJYGU&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=gWKGdCt2S9mkfoZPekCciYcg51z9yB5rHRaEQLqhji7xxvUA3OukA5HAQvF-eYBV&s=xCDEiOrckZLPWTy9gYcALT7zxWjBb4hTrpE8BtBpWgY&e=>. You are receiving this because you were assigned.

[rjleveque]

It seems like a problem that the call to b4step2 in advanc has the wrong number of parameters, since it's missing actualstep. Shouldn't actualstep = .true. in this call since this is done jub above where stepgrid is called with actualstep = .true.?

@mandli: Why do you say actualstep won't be available in stepgrid since it still seems to be in the call to stepgrid from advanc and prepbigstep?

[mjberger]

sorry I missed that. i was looking at why we need it at all from prepbigstep.

…

On Sep 3, 2025, at 7:05 PM, Randall J. LeVeque ***@***.***> wrote: rjleveque left a comment (clawpack/geoclaw#667) <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3251075374&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=kJt677wgDSsdQ5uTOjFaYSQxe78lOECWL5s9CT5Ub5P2nuu_cw5d3Pe3fSr1p0gR&s=FJBgs8-b939TM9fq5_bfBKAX4JKejlQoWi7UxzRhNag&e=> It seems like a problem that the call to b4step2 in advanc has the wrong number of parameters, since it's missing actualstep. Shouldn't actualstep = .true. in this call since this is done jub above where stepgrid is called with actualstep = .true.? @mandli <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_mandli&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=kJt677wgDSsdQ5uTOjFaYSQxe78lOECWL5s9CT5Ub5P2nuu_cw5d3Pe3fSr1p0gR&s=FUott39tiDVWh1-bbbpo4AMfo_qylwbApCyd504wy_A&e=>: Why do you say actualstep won't be available in stepgrid since it still seems to be in the call to stepgrid from advanc and prepbigstep? — Reply to this email directly, view it on GitHub <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3251075374&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=kJt677wgDSsdQ5uTOjFaYSQxe78lOECWL5s9CT5Ub5P2nuu_cw5d3Pe3fSr1p0gR&s=FJBgs8-b939TM9fq5_bfBKAX4JKejlQoWi7UxzRhNag&e=>, or unsubscribe <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AAGUGC5R5C5IYKCHPYACTD33Q5X3HAVCNFSM6AAAAACE7KSCBCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTENJRGA3TKMZXGQ&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=kJt677wgDSsdQ5uTOjFaYSQxe78lOECWL5s9CT5Ub5P2nuu_cw5d3Pe3fSr1p0gR&s=7xzya7ZH6VsVAsRfqko6UBHa8MtjKsQXXgvAssMCzHc&e=>. You are receiving this because you were assigned.

[mandli]

It seems like a problem that the call to b4step2 in advanc has the wrong number of parameters, since it's missing actualstep. Shouldn't actualstep = .true. in this call since this is done jub above where stepgrid is called with actualstep = .true.?

Yup, actualstep == .true. when called from stepgrid so from advanc I belive it should have actualstep == .true. and when called from prepbigstep actualstep == .false..

@mandli: Why do you say actualstep won't be available in stepgrid since it still seems to be in the call to stepgrid from advanc and prepbigstep?

stepgrid does not use actualstep save for calling b4step2 so I figured it was best removed, but this is a small point.

The call to b4step2 might be an issue though if the other things it does still need to be done, such as setting storm fields? Is this needed, or is it created in prepbigstep already by calling setaux and then coarsening? The other stuff, like checking for nans and speed limit checking might not be relevant for 1 step of prepbigstep taken for error estimation?

There may be extraneous work done by calling it here, but I do not think it would cause an issue with the storm based aux array filling. We could turn it off if actualstep == .false. though.

[mjberger]

still not clear on whether prepbigstep needs to call b4step2 with actualstep false. But confirming that it is safe to remove it from stepgrid, or add the argument to advanc.

…

On Sep 4, 2025, at 10:36 AM, Kyle Mandli ***@***.***> wrote: mandli left a comment (clawpack/geoclaw#667) <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3254039441&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=PwOtOb99SqM5Wzt4dDpjMMOqyRyYIZNhK2uoYEqQkah5-mBbYNqPVAx4ZmLEB6t5&s=TRBtlCLxWFHWefu2yBlVIeqQRvWW7OMZdGsInB16z8U&e=> It seems like a problem that the call to b4step2 in advanc has the wrong number of parameters, since it's missing actualstep. Shouldn't actualstep = .true. in this call since this is done jub above where stepgrid is called with actualstep = .true.? Yup, actualstep == .true. when called from stepgrid so from advanc I belive it should have actualstep == .true. and when called from prepbigstep actualstep == .false.. @mandli <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_mandli&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=PwOtOb99SqM5Wzt4dDpjMMOqyRyYIZNhK2uoYEqQkah5-mBbYNqPVAx4ZmLEB6t5&s=gsTLEycFHNPz05wK9CTqspzET3v_y2y8vjiGbji4Kh4&e=>: Why do you say actualstep won't be available in stepgrid since it still seems to be in the call to stepgrid from advanc and prepbigstep? stepgrid does not use actualstep save for calling b4step2 so I figured it was best removed, but this is a small point. The call to b4step2 might be an issue though if the other things it does still need to be done, such as setting storm fields? Is this needed, or is it created in prepbigstep already by calling setaux and then coarsening? The other stuff, like checking for nans and speed limit checking might not be relevant for 1 step of prepbigstep taken for error estimation? There may be extraneous work done by calling it here, but I do not think it would cause an issue with the storm based aux array filling. We could turn it off if actualstep == .false. though. — Reply to this email directly, view it on GitHub <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3254039441&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=PwOtOb99SqM5Wzt4dDpjMMOqyRyYIZNhK2uoYEqQkah5-mBbYNqPVAx4ZmLEB6t5&s=TRBtlCLxWFHWefu2yBlVIeqQRvWW7OMZdGsInB16z8U&e=>, or unsubscribe <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AAGUGCZ3QGMW2IIQI5QOIAD3RBE5HAVCNFSM6AAAAACE7KSCBCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTENJUGAZTSNBUGE&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=PwOtOb99SqM5Wzt4dDpjMMOqyRyYIZNhK2uoYEqQkah5-mBbYNqPVAx4ZmLEB6t5&s=neWXJkdorR8RA2eo1quTAgNCjA755mwfMfJsgVCSUNk&e=>. You are receiving this because you were assigned.

[mandli]

Yeah, not sure that I completely get why it needs to be called either, but need to look into it more thoroughly.

[mjberger]

it needs to be called for time dependent storm fields. Otherwise only setaux is called. So if you need it, add the call to b4step2 to prepbigstep.

…

On Sep 4, 2025, at 10:43 AM, Kyle Mandli ***@***.***> wrote: mandli left a comment (clawpack/geoclaw#667) <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3254072979&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=fmCVNVvogRGiUYuCGnLG0XBlFOIvDNUU-tf-pUfFBPXq_64RTU5ceAmlI77tO2sP&s=osqBm3IwM7HdLAFV7jINvjN43KEwOFL4ZgwJs0-nF1w&e=> Yeah, not sure that I completely get why it needs to be called either, but need to look into it more thoroughly. — Reply to this email directly, view it on GitHub <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_clawpack_geoclaw_issues_667-23issuecomment-2D3254072979&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=fmCVNVvogRGiUYuCGnLG0XBlFOIvDNUU-tf-pUfFBPXq_64RTU5ceAmlI77tO2sP&s=osqBm3IwM7HdLAFV7jINvjN43KEwOFL4ZgwJs0-nF1w&e=>, or unsubscribe <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AAGUGC5CH4BEMMZWBZU7BZL3RBF2TAVCNFSM6AAAAACE7KSCBCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTENJUGA3TEOJXHE&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=4PEQeJ-eTbHJv3d0pLWzLQ&m=fmCVNVvogRGiUYuCGnLG0XBlFOIvDNUU-tf-pUfFBPXq_64RTU5ceAmlI77tO2sP&s=7n6jMYQwWC8tu3ZVx4VEuWfaE8PW-TbPUZk2-uQCYIw&e=>. You are receiving this because you were assigned.

[rjleveque]

In prepbigstep.f there are two calls to setaux, one on the fine grid and one on the coarse. So does b4step2 need to be called twice, once before each of these, with actualstep = .false. in both cases?

We should fix this soon and do a 5.13.1 release, since currently geoclaw code dies. Not sure how that happened, I think I merged in PR#660 at the last minute without checking that things ran properly.

[rjleveque]

I think this is resolved by #670.