pyfeelpp documentation

docs/math/antora.yml

@@ -7,3 +7,4 @@ nav:
 - modules/analyse-fonctionnelle/nav.adoc
 - modules/fem/nav.adoc
 - modules/hdg/nav.adoc
+- modules/rbm/nav.adoc

docs/mor/modules/ROOT/nav.adoc

@@ -7,3 +7,4 @@
 ** xref:pbdw.adoc#_using_pbdw[Using PBDW]
 * xref:index.adoc#_cases[Cases]
 ** xref:opusheat:index.adoc[OpusHeat]
+** xref:thermalfin:index.adoc[ThermalFin]

docs/mor/modules/ROOT/pages/index.adoc

@@ -12,3 +12,4 @@ The Parametrized Background Data Weak (xref:pbdw.adoc[PBDW]) method has also bee
 == Cases
 
 - xref:opusheat:index.adoc[OpusHeat]
+- xref:thermalfin:index.adoc[ThermalFin]

docs/mor/modules/thermalfin/nav.adoc

@@ -0,0 +1 @@
+* xref:index.adoc[Thermal fin]

docs/mor/modules/thermalfin/pages/index.adoc

@@ -0,0 +1,3 @@
+= Thermal fin
+
+The redaction of this page is in progress.

docs/user/modules/python/nav.adoc

@@ -13,7 +13,9 @@
 *** xref:pyfeelpptoolboxes/fluid.adoc[Fluid Mechanics]
 *** xref:pyfeelpptoolboxes/solid.adoc[Solid Mechanics]
 ** xref:pyfeelppmor/index.adoc[{pyfeelppmor}]
-*** xref:pyfeelppmor/parameters.adoc[Parameters]
 *** xref:pyfeelppmor/parameterSpace.adoc[Parameter space]
+*** xref:pyfeelppmor/parameters.adoc[Parameters]
 *** xref:pyfeelppmor/petscDouble.adoc[PETSc vector and matrix]
+*** xref:pyfeelppmor/toolboxmor.adoc[Toolbox MOR]
 *** xref:pyfeelppmor/reducedbasis.adoc[Python module `reducedbasis`]
+**** xref:pyfeelppmor/reducedbasis_offline.adoc[Offline stage]

docs/user/modules/python/pages/pyfeelpp/filters.adoc

@@ -24,30 +24,30 @@ app = feelpp.Environment(["myapp"],config=feelpp.localRepository("")) # <1>
 
 [source,python]
 ----
-app.setLogVerbosityLevel(0) # <0>
+app.setLogVerbosityLevel(0) # <1>
 geofilename=feelpp.download( "github:{repo:feelpp,path:feelpp/quickstart/laplacian/cases/feelpp2d/feelpp2d.geo}", worldComm=app.worldCommPtr() )[0]
-mesh = feelpp.load(feelpp.mesh(dim=2,realdim=2), name=geofilename, h=0.1, verbose=1) # <1>
-Xh=feelpp.functionSpace(mesh=mesh,space="Pch",order=1) # <2>
-P0h=feelpp.functionSpace(mesh=mesh,space="Pdh",order=0) # <3>
-u=Xh.element() # <4>
-u.on(range=feelpp.elements(mesh), expr=feelpp.expr("sin(2*pi*x)*cos(pi*y):x:y")) # <5>
-e = feelpp.exporter(mesh=mesh)  # <6>
-e.addScalar("a_scalar", 1.) # <7>
-e.addP1c("u",u) # <8>
-e.addP0d("pid",feelpp.pid( P0h )) # <9>
-e.save()   # <10>
+mesh = feelpp.load(feelpp.mesh(dim=2,realdim=2), name=geofilename, h=0.1, verbose=1) # <2>
+Xh=feelpp.functionSpace(mesh=mesh,space="Pch",order=1) # <3>
+P0h=feelpp.functionSpace(mesh=mesh,space="Pdh",order=0) # <4>
+u=Xh.element() # <5>
+u.on(range=feelpp.elements(mesh), expr=feelpp.expr("sin(2*pi*x)*cos(pi*y):x:y")) # <6>
+e = feelpp.exporter(mesh=mesh)  # <7>
+e.addScalar("a_scalar", 1.) # <8>
+e.addP1c("u",u) # <9>
+e.addP0d("pid",feelpp.pid( P0h )) # <10>
+e.save()   # <11>
 ----
-<0> The verbosity level is set to 0 (only `VLOG(level)` with `level <= 0` are displayed)
-<1> The mesh is loaded from the {uri-github-feelpp} repository
-<2> The function space stem:[X_h] is built from the mesh stem:[mesh] and the space stem:[Pch] and order stem:[1]
-<3> The function space stem:[P0_h] is built from the mesh stem:[mesh] and the space stem:[Pdh] and order stem:[0]
-<4> The element stem:[u] is built from the function space stem:[X_h]
-<5> The element stem:[u] is interpolated with the expression stem:[{sin(2*pi*x)*cos(pi*y)}:x:y]
-<6> The exporter is built from the mesh stem:[mesh]
-<7> The scalar stem:[a_scalar] is added to the exporter stem:[e]
-<8> The element stem:[u] is added to the exporter stem:[e]
-<9> The element stem:[pid] is added to the exporter stem:[e]
-<10> The exporter stem:[e] is saved to the disk
+<1> The verbosity level is set to 0 (only `VLOG(level)` with `level <= 0` are displayed)
+<2> The mesh is loaded from the {uri-github-feelpp} repository
+<3> The function space stem:[X_h] is built from the mesh `mesh` and the space `Pch` and order `1`
+<4> The function space stem:[P0_h] is built from the mesh `mesh` and the space `Pdh` and order `0`
+<5> The element `u` is built from the function space `Xh`
+<6> The element `u` is interpolated with the expression `{sin(2*pi*x)*cos(pi*y)}:x:y`
+<7> The exporter is built from the mesh `mesh`
+<8> The scalar `a_scalar` is added to the exporter `e`
+<9> The element `u` is added to the exporter `e`
+<10> The element `pid` is added to the exporter `e`
+<11> The exporter `e` is saved to the disk
 
 
 [%collapsible.result]
@@ -59,7 +59,7 @@ e.save()   # <10>
 ----
 ====
 
-The results are stored in the directory `feelppdb`` and can be visualized with {paraview-website} or {ensight}. 
+The results are stored in the directory `feelppdb` and can be visualized with {paraview-website} or {ensight}. 
 
 INFO: The file `exports/ensightgold/myapp/myapp.case` can be opened directly in {paraview} or {ensight} to visualize the mesh and the fields.
 

docs/user/modules/python/pages/pyfeelpp/mesh.adoc

@@ -208,4 +208,4 @@ i1 = [3.24282386], i2 = [149.465], i3 = [0.]
 == Next steps
 
 - [x] xref:pyfeelpp/discr.adoc[Manipulate functions and function spaces]
-- [x] Export to paraview
+- [x] xref:pyfeelpp/filters.adoc[Export to paraview]

docs/user/modules/python/pages/pyfeelppmor/index.adoc

@@ -1,4 +1,3 @@
 = {feelpp} model order reduction in Python
 include::user:ROOT:partial$header-macros.adoc[]
 
-This sections present the modules developped and the function implemented to handle {feelpp} MOR objects in Python. The script link:https://github.com/feelpp/feelpp/blob/develop/mor/pyfeelpp-mor/feelpp/mor/toolboxmor.py[toolboxmor.py] presents the main features, on the case opus-heat.

docs/user/modules/python/pages/pyfeelppmor/petscDouble.adoc

@@ -7,12 +7,12 @@ Class `VectorPetscDouble` (`feelpp._alg.VectorPetscDouble`)
 
 === Methods
 
-* `__init__(*args, **kwargs)` : Overloaded function.
+* `\\__init__(*args, **kwargs)` : Overloaded function.
 
-    1. `__init__()`. The vector created is empty.
-    2. `__init__(arg0: int, arg1: feelpp._core.WorldComm)`. The vector created has a size `arg0`.
-    3. `__init__(arg0: int, arg1: int, arg2: feelpp._core.WorldComm)`. The vector created has a global size `arg0`, and a local size `arg1`.
-    4. `__init__(arg0: feelpp._alg.DataMap, arg1: bool)`
+    1. `\\__init__()`. The vector created is empty.
+    2. `\\__init__(arg0: int, arg1: feelpp._core.WorldComm)`. The vector created has a size `arg0`.
+    3. `\\__init__(arg0: int, arg1: int, arg2: feelpp._core.WorldComm)`. The vector created has a global size `arg0`, and a local size `arg1`.
+    4. `\\__init__(arg0: feelpp._alg.DataMap, arg1: bool)`
 * `clear()` : clear PETSc vector
 * `size()` -> `int` : return  PETSc Vector size
 * `vec()` -> `vec` : return a PETSc Vector that can be manipulated with funciton of the module https://pypi.org/project/petsc4py/[petsc4py]. The functions of this class can be found on the https://www.mcs.anl.gov/petsc/petsc4py-current/docs/apiref/petsc4py.PETSc.Vec-class.html[documentation of PETSc4py].
@@ -26,10 +26,10 @@ Class `MatrixPetscDouble` (`feelpp._alg.MatrixPetscDouble`). The methods of this
 
 === Methods
 
-* `__init__(*args, **kwargs)` : Overloaded function.
-    1. `__init__(arg0: feelpp._core.WorldComm)`
-    2. `__init__(arg0: feelpp._alg.DataMap, arg1: feelpp._alg.DataMap)`
-    3. `__init__(arg1: feelpp._alg.DataMap, arg2: feelpp._core.WorldComm)`
+* `\\__init__(*args, **kwargs)` : Overloaded function.
+    1. `\\__init__(arg0: feelpp._core.WorldComm)`
+    2. `\\__init__(arg0: feelpp._alg.DataMap, arg1: feelpp._alg.DataMap)`
+    3. `\\__init__(arg1: feelpp._alg.DataMap, arg2: feelpp._core.WorldComm)`
 * `clear()` : clear PETSc matrix
 * `mat(...)` -> `mat` : return a PETSc sparse matrix. See https://www.mcs.anl.gov/petsc/petsc4py-current/docs/apiref/petsc4py.PETSc.Mat-class.html[the documentation].
 * `rowStart()` -> `int` : return PETSc Matrix row start. This method is usefull is the matrix is shared with many processors, as the following.

docs/user/modules/python/pages/pyfeelppmor/reducedbasis.adoc

@@ -20,19 +20,24 @@ To set the environment, those module also need to be imported :
 [source, python]
 ----
 import sys, os
-from feelpp.toolboxes.heat import *
-import feelpp.toolboxes.core as core
-from feelpp.operators import mass
-import feelpp.mor as mor
+from feelpp.toolboxes.heat import * <1>
+import feelpp.toolboxes.core as core <2>
+from feelpp.operators import mass <3>
+import feelpp.mor as mor <4>
 import feelpp
 ----
 
+<1> The toolbox heat is used to simulate the heat equation on the case studied.
+<2> The module `core` is used to set the environment.
+<3> The `mass` function is needed to create the mass matrix if we work on a time-dependent case.
+<4> The module `mor` is used to get the functions neededd by the toolbox mor.
+
 
 == Set the environment
 
 [source, python]
 ----
-config = feelpp.globalRepository("reducedbasis")
+config = feelpp.localRepository("reducedbasis")
 sys.argv = ["reducedbasis"]
 o = core.toolboxes_options("heat")
 o.add(mor.makeToolboxMorOptions())
@@ -43,22 +48,39 @@ e = feelpp.Environment(sys.argv, opts=o, config=config)
 
 == Set the model
 
+=== Description of the model
+
+In this document, the cases used is `thermal-fin` in 2D, with a stem:[\mathbb{P}_1] discretization and stationnary.
+This model is described in xref:mor:thermalfin:index.adoc[this page].
+
+The usage would be the same for another case.
+
+
+=== Code
+
 Set the parameters nedded to the model :
+
 - `config_file` : path to the cfg file
 - `dim` : the dimension of the case
 - `order` : the order of discretization (`1` for stem:[\mathbb{P}_1], `2` for stem:[\mathbb{P}_2])
-- `time_dependant` : is the case stationnary or transient ? (for now always to `False`)
+- `time_depedent` : is the case stationnary or transient ? (for now always to `False`)
 - `name` : name of the CRB case
 
+
+
 [source,python]
 ----
 config_file = "thermal-fin.cfg"
 dim = 2
 order = 1
-time_dependant = False
+time_depedent = False
 name = "thermalfin-2d"
 ----
 
+The configuration files for this case are available on github : https://raw.githubusercontent.com/feelpp/feelpp/develop/mor/cases/thermal-fin/2d/fin.geo[geometry file], https://raw.githubusercontent.com/feelpp/feelpp/develop/mor/cases/thermal-fin/2d/thermal-fin.cfg[cfg file], and https://raw.githubusercontent.com/feelpp/feelpp/develop/mor/cases/thermal-fin/2d/thermal-fin.json[json file].
+
+Then, we can create the model :
+
 [source,python]
 ----
 cfg = feelpp.readCfg(config_file)
@@ -118,7 +140,7 @@ default_parameter = modelParameters.toParameterValues()
 
 [source, python]
 ----
-model = mor.toolboxmor(name=name, dim=dim, time_dependent=time_dependant)
+model = mor.toolboxmor(name=name, dim=dim, time_dependent=time_depedent)
 model.setFunctionSpaces( Vh=heatBox.spaceTemperature() )
 
 def assembleDEIM(mu):
@@ -305,6 +327,8 @@ rb.computeOfflineErrorRhs()
 rb.computeOfflineError()
 ----
 
+The script `generate_basis.py` run this offline part. See the  xref:pyfeelppmor/reducedbasis_offline.adoc[dedicated page] for more details.
+
 == Online phase
 
 [source, python]
@@ -355,4 +379,6 @@ rb.getSolutionsFE(mu, k=0)
 ----
 ====
 
-The two functions `getSolutions` (resp. `getSolutionsFE`) return the solution stem:[u_N(\mu)] (resp. stem:[u(\mu)]) and the value of the output stem:[s_N^k(\mu)] (res. stem:[s^k(\mu)]).
+The two functions `getSolutions` (resp. `getSolutionsFE`) return the solution stem:[u_N(\mu)] (resp. stem:[u(\mu)]) and the value of the output stem:[s_N^k(\mu)] (res. stem:[s^k(\mu)]).
+See xref:pyfeelppmor/parameters.adoc[the dedicated page] for the API of `ParameterSpaceElement`.
+

docs/user/modules/python/pages/pyfeelppmor/reducedbasis_offline.adoc

@@ -0,0 +1,146 @@
+= Usage of Feelpp / Mor : Offline part
+
+:stem: latexmath
+
+== Usage
+
+This part can be run in parallel.
+
+.Usage of `generate_basis` script
+[source, bash]
+----
+[mpirun -np <nproc>] python3 generate_basis.py
+    --config-file <path to cfg file>
+    --odir <path to export directory>
+    --dim <DIM>
+    --case <name of the case>
+    [--algo {0|1|2}]
+    [--train-size <size>]
+    [--tol <tol>]
+    [--time-dependant={True | False}]
+----
+
+[[parameters]]
+.Parameters
+[cols="1,1,1"]
+|===
+|Name|Description|Default value
+
+|`config-file`
+|path to config `cfg` file
+|_mandatory_
+
+|`odir`
+|path to output directory, where data will be saved
+|_mandatory_
+
+|`case`
+|name of the case
+|`generate_basis`
+
+|`dim`
+|dimension of the case (must be `2` or `3`)
+|_mandatory_
+
+|`algo`
+|algorithm used to generate a basis, see <<Algorithms, table below>>
+|`1`
+
+|`train-size`
+|Size of the train set (depending on algorithm used)
+|`40`
+
+|`tol`
+|tolerance used for generating (depends on algorithm)
+|`1e-6`
+
+|`time-dependant`
+|time dependant case
+|`False`
+|===
+
+
+
+[[Algorithms]]
+== Algorithms
+
+
+.Algorithms
+[cols='1,2,5']
+|===
+|Value|Algorithm|Description
+
+|`0`
+|From sample
+|Generates a basis of size stem:[N=]`train-size` elements, [log-]randomly taken in the space.
+
+|`1`
+|Greedy algorithm
+|Run the greedy algorithm on a train set of element of size `train-size`. This algorithm also stores the evolution of the maximal error bound at each step.
+
+|`2`
+|POD generation
+|Takes the largest POD modes from a basis of size `train-size`. The resulting basis will have a size stem:[N\leqslant]`train-size`. This algorithm also stores the evolution of the maximal error bound at each step.
+
+|===
+
+WARNING: For now, the computation of error bound is only valid when the decomposition is coercive (_i.e._ stem:[A(\mu)=\displaystyle\sum_{q}\beta_A(\mu)A^q], with stem:[\beta_A(\mu)\geqslant 0])
+
+[[offline]]
+== Exported files
+
+Here is a description of the generated files :
+
+1. A `JSON` file, exported in `odir` directory, containing the following informations :
+    - `Qa` : Size of the decomposition of stem:[A(\mu)]
+    - `Qf` : Size of the decomposition of stem:[F(\mu)]
+    - `QLk` : Sizes of the decompositions of stem:[L_k(\mu)] for stem:[k\in\{1,n_\text{output}\}], gathers in a list.
+    - `N_output` : Number of output described in the JSON (`CRBOutput`)
+    - `output_names` : names of those outputs
+    - `N` : Size of the reduced basis
+    - `path` : Path where `h5` file is stored 
+    - `mubar` : Values of stem:[\bar{\mu}]
+
+2. A `h5` file, containing all the matrices used in the online part (of « small » size).
+
+
+== Documentation for developpers
+
+The function to call to generate the basis is `generatebasis` :
+
+[source, python]
+----
+import feelpp.mor.generate_basis as g               # import the module
+g.generatebasis(worldcomm=worldcomm, config=config) # run the script
+----
+
+Where :
+
+1. `worldcomm` is a pointer to the MPI communicator. This argument is optionnal. If none (or `None`) is given, the function will get the communicaotr from `feelpp.Environment`.
+
+2. `config` is an object of type `g.generateBasisConfig`. It contains the paramters used by the script to generate the application.
+
+[source, python]
+----
+config = generateBasisConfig(dim, config_file, time_dependant, odir, case, algo, size, tol)
+----
+
+The description and the default values of those parameters are descirbed <<parameters,above>>.
+
+If `odir` contains `$name`, this expression will be replaced by the _name_ of the case, defined by concatenation `<case>-np_<nproc>`, where `case` is the name of the case given in the configuration, and `nproc` is the number of processors where the simulation is run.
+
+
+[[files]]
+== Files created by {feelpp}
+
+Some files are created by {feelpp}. The emplacement of those file is important because they are used by the online part. In the following list, `<feel-dir>` represents the path to the `feel` directory. If nothing has been change in `~/.feelppconfig`, the path to the `<feel-dir>` is `~/feel`.
+
+* Model repository : `<feel-dir>/crbdb/<name>/<uuid>`, where :
+** `<name>` is the name of the case, defined by the option `--case`.
+** `<uuid>` is a unique identifier of the model, computed by the toolbox.
+This repository contains the meshes for the EIM decomposition.
+
+* DEIM and MDEIM sampling files : `<feel-dir>/generate_basis-<name>/np_<nproc>/DmuDEim-P<dim>-Ne<size>-generated-by-master-proc` :
+** `<dim>` is the dimension of the parameter space.
+** `<size>` is the size of the train set used on the EIM algorithm. This corresponds to the option `toolboxmor.trainset-deim-size` and `trainset-mdeim-size` in the cfg file.
+Those two files (one for the EIM decomposition of the matrix, and one for the EIM decomposition of the right-hand side) are generated by the toolbox, and contain the values of the parameters used to generate the basis.