Analyses

In Dyad, an analysis is "the things you can do with a model". A typical workflow in Dyad is to create a set of components, and then create an analysis that runs on the entry point to that set of components. It produces a solution object that provides information about the model, and can be used to build various visualizations.

Analyses can perform a wide variety of model analytics, such as:

• Simulate the model over a given time span

• Analyze the stability of a controller

• Run a parameter sweep

• Build an FMU

• Calibrate the model to data

and much more. All analyses have a common interface which take in the model, run an analysis, and give back a pre-defined set of artifacts, such as plots of the solution or generated binaries.

Using the Julia-Level Analysis Interface (Dyad Studio)

Dyad Studio includes a lower level interface to define and interact with analyses at the Julia level, with more flexibility than what is exposed in the GUI.

Example: TransientAnalysis of a Simple Block

Using the Dyad-Level Analysis Interface (Dyad Studio)

A TransientAnalysis for the SecondOrderSystemTest component is written in Dyad as:

analysis SecondOrderSystemTransient
  extends TransientAnalysis(stop=5)
  model = DyadExampleComponents.SecondOrderSystemTest()
end

The Dyad compiler will generate the SecondOrderSystemTransient function that will run the analysis.

result = SecondOrderSystemTransient()

Transient Analysis Solution for TransientAnalysis
retcode: Success
Interpolation: 3rd order Hermite
t: 91-element Vector{Float64}:
 0.0
 9.999999999999999e-5
 0.0010999999999999998
 0.008325009772234456
 0.020994176283791393
 0.0364343358670688
 0.05674523514445098
 0.08286989819734197
 0.11221638932307687
 0.14642347979835246
 ⋮
 4.531377144933996
 4.600395661887067
 4.658489156486032
 4.717039575802435
 4.797097378344955
 4.853010164393907
 4.913798996999781
 4.986904078860146
 5.0
u: 91-element Vector{Vector{Float64}}:
 [0.0, 0.0]
 [1.570795034867126e-7, 5.235985172127158e-12]
 [1.9004744117386418e-5, 6.96868358226933e-9]
 [0.0010824613825442051, 3.010699496947032e-6]
 [0.006676004188171907, 4.740719263845825e-5]
 [0.018672260214054187, 0.00023715849108202821]
 [0.03852585131045319, 0.0008157098034480075]
 [0.05916280060463827, 0.0021185079283594587]
 [0.06134638493866648, 0.003951352196806262]
 [0.03540013662961773, 0.005667636390935834]
 ⋮
 [0.04952985663058974, 0.14529342452097024]
 [0.000127935674026706, 0.14663964118144118]
 [0.04034155170192192, 0.14753118116840172]
 [0.05933195621751569, 0.15089699947601312]
 [0.0002772715611076463, 0.1530273856659445]
 [0.03497875213568734, 0.15371414368230196]
 [0.06086038742068811, 0.15709298444433004]
 [0.0028132618482891642, 0.15940880929987014]
 [0.00015708255353278367, 0.15942252727283682]

We can plot this result, using either Plots.jl or Makie.jl. We'll use Plots here:

using Plots
plot(result)

See the plotting guide for more details on how to plot the results of an analysis.

We can also get artifacts from this result, like:

using DyadInterface: artifacts
artifacts(result, :SimulationSolutionTable)

91×3 DataFrame

Row	timestamp	sys₊integrator1₊x(t)	sys₊integrator2₊x(t)
	Float64	Float64	Float64
1	0.0	0.0	0.0
2	0.0001	1.5708e-7	5.23599e-12
3	0.0011	1.90047e-5	6.96868e-9
4	0.00832501	0.00108246	3.0107e-6
5	0.0209942	0.006676	4.74072e-5
6	0.0364343	0.0186723	0.000237158
7	0.0567452	0.0385259	0.00081571
8	0.0828699	0.0591628	0.00211851
9	0.112216	0.0613464	0.00395135
10	0.146423	0.0354001	0.00566764
11	0.184829	0.00354746	0.0063481
12	0.224383	0.0088908	0.00643985
13	0.264834	0.0461333	0.00752472
14	0.310568	0.0619237	0.010216
15	0.362219	0.0199111	0.0124693
16	0.411188	0.00194739	0.0127396
17	0.466172	0.0473151	0.0139535
18	0.516467	0.0594961	0.016941
19	0.580588	0.00574339	0.0190607
20	0.632796	0.0154559	0.0192735
21	0.6872	0.0611255	0.021478
22	0.745374	0.0364467	0.0247294
23	0.803878	0.000242572	0.0254656
24	0.861963	0.0435171	0.0264955
25	0.917198	0.0591318	0.029718
26	0.989732	0.00165813	0.0318257
27	1.04438	0.0262546	0.0322472
28	1.10222	0.063598	0.0351578
29	1.16647	0.0161093	0.0380138
30	1.22394	0.00860495	0.038271
31	1.29123	0.0624728	0.0408316
32	1.34945	0.0323937	0.0439749
33	1.4124	0.00240111	0.0445805
34	1.47551	0.0547086	0.0462708
35	1.53363	0.0475053	0.0497086
36	1.60113	3.67886e-5	0.0509386
37	1.65983	0.0415178	0.0518786
38	1.71818	0.0586226	0.0552505
39	1.79737	0.000142258	0.057305
40	1.85342	0.0352737	0.0579998
41	1.91401	0.0606615	0.0613702
42	1.98676	0.00275927	0.0636649
43	2.04403	0.0259363	0.0640852
44	2.10312	0.0635508	0.0670641
45	2.16924	0.0137913	0.069906
46	2.22743	0.0111571	0.0701579
47	2.28931	0.0619277	0.0725662
48	2.34958	0.0323035	0.0758348
49	2.41129	0.00203085	0.0764362
50	2.47436	0.0539275	0.0780685
51	2.53211	0.0488407	0.0814976
52	2.60061	5.63697e-5	0.0828028
53	2.65891	0.0406761	0.0837072
54	2.71739	0.059074	0.0870724
55	2.79728	0.000185943	0.0891758
56	2.85322	0.0351114	0.0898656
57	2.91396	0.0607154	0.0932425
58	2.98705	0.00267976	0.0955432
59	3.04434	0.0262769	0.0959729
60	3.10355	0.0635414	0.0989733
61	3.16994	0.0132561	0.1018
62	3.22812	0.0117199	0.102052
63	3.289	0.06186	0.104435
64	3.34957	0.0323441	0.107725
65	3.41102	0.00197455	0.108329
66	3.47402	0.0537142	0.109945
67	3.53166	0.0492557	0.113373
68	3.60048	9.09425e-5	0.114703
69	3.65866	0.0404666	0.115599
70	3.71718	0.0592206	0.118964
71	3.79729	0.000221938	0.121083
72	3.85321	0.0351384	0.121774
73	3.91401	0.0607318	0.125157
74	3.9873	0.00262081	0.127458
75	4.04459	0.0265607	0.127896
76	4.10388	0.0635415	0.130912
77	4.17044	0.0128922	0.133727
78	4.22861	0.0121332	0.133981
79	4.28881	0.0618343	0.136349
80	4.3496	0.0323585	0.139654
81	4.41085	0.0019565	0.140258
82	4.47381	0.0535974	0.141866
83	4.53138	0.0495299	0.145293
84	4.6004	0.000127936	0.14664
85	4.65849	0.0403416	0.147531
86	4.71704	0.059332	0.150897
87	4.7971	0.000277272	0.153027
88	4.85301	0.0349788	0.153714
89	4.9138	0.0608604	0.157093
90	4.9869	0.00281326	0.159409
91	5.0	0.000157083	0.159423

For the artifacts allowed for a given analysis, see the pages which define the given analysis.

Defining and Running an Analysis

Any analysis is available and meant to run in Julia. For an analysis called MyAnalysis in Dyad, the generated Julia code will be created with the following format:

result = MyAnalysis(; kwargs...)

where the kwargs are keyword arguments corresponding to the arguments of the analysis run definition (i.e. the fields that can be set at the Dyad level). This invocation will be set to automatically have the keyword arguments default to the values defined in the Dyad version, for example for a TransientAnalysis if one had set stop = 10.0, then stop = 10.0 will be the default value in the generated function. These defaults can thus be overruled by declaring the keyword argument, for example MyAnalysis(; stop=20.0) will run a version of the analysis which overrides the stop time to 20.0.

Analysis Result Interface

The analysis results can be interacted with in two ways: there are the customizable plot and the artifacts. The most basic usage is to get the artifacts via:

artifacts(result, name::Symbol)

The allowed artifacts can be queried via:

artifacts(results)

which returns a list of all the available artifact names. For more details about all the artifacts,

AnalysisSolutionMetadata(result)

can be used, which also includes descriptions for all the artifacts.

Additionally, some analyses include a customizable visualization:

customizable_visualization(::AbstractAnalysisSolution, ::PlotlyVisualizationSpec)

This will be defined fully in future versions.

Defining Your Own Analysis

At the lowest level, analyses are implemented in Julia, see the Custom analysis tutorial for more details on how to write your own.