eturb.operators

Operators

Information regarding mesh, mathematical operators, and domain decomposition.

Todo

Move length to abl.par file and extrude it to allow double precision values.

Note

We deliberately try not to use the variable names used in Nek5000, as those are ambiguously named.

Functions

next_power(value[, base])

Compute the perfect power of base greater than or equal to value.

Classes

Operators([sim, params])

Container for parameters and writing box and SIZE files.

class eturb.operators.Operators(sim=None, params=None)

Container for parameters and writing box and SIZE files.

Note

Some values are not available as parameters and instead automatically computed for generating the SIZE file.

SIZE	properties	Comment
lelg	max_n_seq	Max. number of elements globally
lelt	max_n_loc	Max. number of elements per processor (should be not smaller than lelg/lpmin, i.e.
lelx	max_nx	Automatically computed. Max. number of elements along x direction for the global tensor product solver / dimensions.
lely	max_ny	Same as above for y direction.
lelz	max_nz	Same as above for z direction.
lbelt	order_mhd	Automatically computed as lelt if "MHD" in params.nek.problem_type.equation.
lpelt	order_linear	Automatically computed as lelt if "linear" in params.nek.problem_type.equation.
lcvelt	order_cvode	Automatically computed as lelt if params.nek.cvode._enabled is True
lx1m	order_mesh_solver	p-order for mesh solver. Automatically computed based params.nek.general.stress_formulation and whether Arbitrary Lagrangian-Eulerian (ALE) methods are used or not.

Documentation for params

Documentation for params.oper

The following table matches counterpart of mandatory SIZE variables.

SIZE	params.oper	Comment
ldim	dim	Domain dimensions (2 or 3)
lpmin	nproc_min	Min MPI ranks
lpmax	nproc_max	Max MPI ranks
ldimt	scalars	Number of auxilary fields (minimum 1).

Documentation for params.oper.max

The following table matches counterpart of optional SIZE variables. These refer to upper bound number of something. The parameters are considered “optional” and would be ignored with the default values.

SIZE	params.oper.max	Comment
mxprev	dim_proj	Max. dimension of projection space
lgmres	dim_krylov	Max. dimension of Krylov space for GMRES
lhis	hist	Max. number of history (i.e. monitoring) points.
maxobj	obj	Max. number of objects?
lorder	order_time	Max. temporal order (minimum: 2)
lpert	perturb	Max. number of perturbations
toteq	scalars_cons	Max. conserved scalars
ldimt_proj	scalars_proj	Max. scalars for residual projection
nsessmax	sessions	Max. sessions to NEKNEK

Documentation for params.oper.elem

The following table relate to element configuration for certain operations. The parameters are considered “optional” (except for lx1 / order and lxd / coef_dealiasing which are mandatory) and would be ignored with the default values.

SIZE	params.oper.elem	Comment
lxd	coef_dealiasing	p-order 1 for over-integration. Automatically computed from float coef_dealiasing. See order_dealiasing
lx1	order	p-order (avoid uneven and values <6).
lxo	order_out	Max. p-order on output (should be >= order)
lx2	staggered	p-order for pressure. Automatically computed from boolean staggered (True implies \(\mathbb{P}_N - \mathbb{P}_{N-2}\) and False implies \(\mathbb{P}_N - \mathbb{P}_{N}\) or a collocated grid). See order_pressure

1

Polynomial order which means the number of GLL / GL points per element.

Documentation for params.oper.misc

Other miscellaneous parameters:

SIZE	params.oper.misc	Comment
lfdm	fast_diag	Equals to True for global tensor product solver (that uses fast diagonalization method). False otherwise.

property max_n_loc

Equivalent to lelt. The next integer greater than or equals max_n_seq / params.oper.nproc_min.

property max_n_seq

Equivalent to lelg.

property max_nx

Equivalent to lelx.

property max_ny

Equivalent to lely.

property max_nz

Equivalent to lelz.

memory_required()

According to FAQ the following estimate:

lx1*ly1*lz1*lelt * 3000byte + lelg * 12byte + MPI + optional libraries
(e.g. CVODE)

property nb_fields

Used in .box file.

property order

Equivalent to lx1.

property order_cvode

Equivalent to lcvelt.

property order_dealiasing

Equivalent to lxd.

property order_linear

Equivalent to lpelt.

property order_mesh_solver

Equivalent to lx1m.

Todo

Must include a condition to check if ALE methods are used or not.

property order_mhd

Equivalent to lbelt.

property order_pressure

Equivalent to lx2.

produce_long_str_describing_oper(oper_method='Base')

Produce a string describing the operator.

produce_str_describing_oper()

Produce a string describing the operator.

write_box(template, fp=<_io.TextIOWrapper name='<stdout>' mode='w' encoding='UTF-8'>, comments='')

Write the .box file which is input for the genbox meshing tool.

Parameters

• template (jinja2.environment.Template) – Template instance like abl.templates.box

• fp (io.TextIOWrapper) – File handler to write to

write_size(template, fp=<_io.TextIOWrapper name='<stdout>' mode='w' encoding='UTF-8'>, comments='')

Write the SIZE file which is required for compilation.

Parameters

template (jinja2.environment.Template) – Template instance like abl.templates.size

eturb.operators.next_power(value, base=2)

Compute the perfect power of base greater than or equal to value.

Parameters

• value (int or float) –

• base (int) –

Return int