GlassBR: Move code-oriented QDefs forced into SRS to Choices as hand-wired QDefs.#4664
GlassBR: Move code-oriented QDefs forced into SRS to Choices as hand-wired QDefs.#4664
Conversation
hand-wired QDefs.
balacij
force-pushed
the
mvHandWiredGlassBRDefs
branch
from
76aa7e6 to
883c78f
Compare
|
Another PR with too much Drasil-speak in its description. And a weird side-effect of changing the order of output in the code. And I don't understand the need to move some of the code around. But, at the higher level: if my guess of what this is doing is correct, then I think this is a change in a good direction. |
|
Piggybacking on #4663 (comment), this PR removes the DD+IM hack to get hand-written variable calculation solutions into the code generator by moving them to the options set we have for the code generator. |
…nto mvHandWiredGlassBRDefs
balacij
force-pushed
the
mvHandWiredGlassBRDefs
branch
from
dfb7999 to
834e77d
Compare
|
Regarding the re-ordering, that just looks like it's purely by chance at the moment. The solution execution order is partially ordered about dependencies but not sorted about something else. I tried moving around where the I think that is an issue for another PR? |
|
I think interpY and interpZ should eventually be codified as theories. Interpolation isn't part of the SRS for GlassBR, so that the SRS can remain abstract. Interpolation is part of the "how" for the SRS's "what". I believe we've talked before about the idea of refining the abstract theories of the SRS into concrete code. The interpolation is an intermediate step in this refinement. I could imagine us eventually generating documentation around these "numerical method" theories. I think the documentation for an interpolation library, would like similar to the documentation of the SRS. It would just be a matter of a viewpoint change so that interpolation becomes the "what" and the "how" would be the generated code. |
…nto mvHandWiredGlassBRDefs
balacij
requested review from
JacquesCarette,
samm82 and
smiths
as code owners
| @@ -93,8 +93,6 @@ \subsection{Table of Symbols} | |||
| \\ | |||
| $\mathit{interpY}$ & InterpY & -- | |||
There was a problem hiding this comment.
Something is weird here: interpY is still in the table of symbols, but interpZ is not. This seems inconsistent: either both should be, or none should be.
There was a problem hiding this comment.
This is now dealt with in 195f1bb. I missed an IM that also referenced interpY.
| @@ -1237,9 +1232,7 @@ <h3>Data Definitions</h3> | |||
| </tr> | |||
| <tr> | |||
| <th>RefBy</th> | |||
There was a problem hiding this comment.
This RefBy change seems not good? The "calculation of demand" is still a thing, but with this change, it has gone?
| </tr> | ||
| <tr> | ||
| <th>RefBy</th> | ||
| <td> |
There was a problem hiding this comment.
And here, even more content has disappeared. This content might have been "hooked in" in a sub-optimal way before -- but it shouldn't entirely disappear. It needs to been hooked in some other way.
| |\\(h\\) |Minimum thickness |[DD:minThick](./SecDDs.md#DD:minThick) |\\({\text{m}}\\) | | ||
| |\\(\mathit{isSafeLR}\\) |3 second load equivalent resistance safety requirement|[IM:isSafeLR](./SecIMs.md#IM:isSafeLR) |-- | | ||
| |\\(\mathit{isSafePb}\\) |Probability of glass breakage safety requirement |[IM:isSafePb](./SecIMs.md#IM:isSafePb) |-- | | ||
| |\\(J\\) |Stress distribution factor (Function) |[IM:stressDistFac](./SecIMs.md#IM:stressDistFac)|-- | |
There was a problem hiding this comment.
I'm not entirely sure yet, but it is definitely not of type: Function!
This is the code we generate for it.
/** \brief Calculates stress distribution factor (Function)
\param inParams structure holding the input values
\param q_hat dimensionless load
\return stress distribution factor (Function)
*/
public static double func_J(InputParameters inParams, double q_hat) {
StreamWriter outfile;
outfile = new StreamWriter("log.txt", true);
outfile.WriteLine("function func_J called with inputs: {");
outfile.Write(" inParams = ");
outfile.Write("Instance of InputParameters object");
outfile.WriteLine(", ");
outfile.Write(" q_hat = ");
outfile.WriteLine(q_hat);
outfile.WriteLine(" }");
outfile.Close();
return Interpolation.interpZ("SDF.txt", inParams.AR, q_hat);
}| @@ -19,31 +19,12 @@ The goal [GS:Predict-Glass-Withstands-Explosion](./SecGoalStmt.md#willBreakGS) i | |||
| |Output Constraints| | | |||
| |Equation |\\[B=\frac{k}{\left(a\\,b\right)^{m-1}}\\,\left(E\\,h^{2}\right)^{m}\\,\mathit{LDF}\\,e^{J}\\] | | |||
| |Description |<ul><li>\\(B\\) is the risk of failure (Unitless)</li><li>\\(k\\) is the surface flaw parameter (\\(\frac{\text{m}^{12}}{\text{N}^{7}}\\))</li><li>\\(a\\) is the plate length (long dimension) (\\({\text{m}}\\))</li><li>\\(b\\) is the plate width (short dimension) (\\({\text{m}}\\))</li><li>\\(m\\) is the surface flaw parameter (\\(\frac{\text{m}^{12}}{\text{N}^{7}}\\))</li><li>\\(E\\) is the modulus of elasticity of glass (\\({\text{Pa}}\\))</li><li>\\(h\\) is the minimum thickness (\\({\text{m}}\\))</li><li>\\(\mathit{LDF}\\) is the load duration factor (Unitless)</li><li>\\(J\\) is the stress distribution factor (Function) (Unitless)</li></ul>| | |||
There was a problem hiding this comment.
Look here, J is in the description, but no longer in the notes nor does it appear as an IM. This leaves the SRS in an inconsistent state.
| |Output Constraints| | | ||
| |Equation |\\[\hat{q}=\frac{q\\,\left(a\\,b\right)^{2}}{E\\,h^{4}\\,\mathit{GTF}}\\] | | ||
| |Description |<ul><li>\\(\hat{q}\\) is the dimensionless load (Unitless)</li><li>\\(q\\) is the applied load (demand) (\\({\text{Pa}}\\))</li><li>\\(a\\) is the plate length (long dimension) (\\({\text{m}}\\))</li><li>\\(b\\) is the plate width (short dimension) (\\({\text{m}}\\))</li><li>\\(E\\) is the modulus of elasticity of glass (\\({\text{Pa}}\\))</li><li>\\(h\\) is the minimum thickness (\\({\text{m}}\\))</li><li>\\(\mathit{GTF}\\) is the glass type factor (Unitless)</li></ul>| | ||
| |Notes |<ul><li>\\(q\\) is the 3 second duration equivalent pressure, as given in [DD:calofDemand](./SecDDs.md#DD:calofDemand).</li><li>\\(a\\) and \\(b\\) are the dimensions of the plate, where (\\(a\geq{}b\\)).</li><li>\\(E\\) comes from [A:standardValues](./SecAssumps.md#assumpSV).</li><li>\\(h\\) is defined in [DD:minThick](./SecDDs.md#DD:minThick) and is based on the nominal thicknesses.</li><li>\\(\mathit{GTF}\\) is defined in [DD:gTF](./SecDDs.md#DD:gTF).</li></ul> | |
There was a problem hiding this comment.
Same here: q is used in the Description above, but now has gone from here.
|
With this PR, I removed the IMs and the one DD involving Any thoughts on how we should fix them? They currently directly deal with symbols that directly go into code (which they shouldn't be doing). Should we give them slightly more "abstract" definitions? e.g., |
I agree that we don't want to remove the idea of interpolation.
I'm not sure we should do that right now though. I feel like we need to nail down exactly what we mean by theory before we start adding more theories. |
3 participants
The IM and DD involving interpY and interpZ appear to be forced into the SRS as a result of not having any way to read in the quantities directly (#3355). We should be able to force them out of the SRS by moving to a list of hand-wired QDefs provided through
Choices.This PR is slightly dubious and I was a bit rough with the removal of the DD and the IM, which is why I'm leaving this PR as a draft. But feedback would be appreciated on the rationale @JacquesCarette @smiths . In the future, these models (DD/IM) can be moved towards an SDS if we don't already have a better mechanism in place for input parameters with default values AND support for matrices/lists.