Procedures

Get or add a integer value of a given name

Get or add a integer array of a given name

Get or add a logical value of a given name

Get or add a logical array of a given name

Get or add a real value of a given name

Get or add a real array of a given name

Get a character value of a given name

Get or add a character array of a given name

Add a configuration variable with an integer value

Add a configuration variable with an array of type integer

Add a configuration variable with an logical value

Add a configuration variable with an array of type logical

Add a configuration variable with a real value

Add a configuration variable with an array of type

Add a configuration variable with an character value

Add a configuration variable with an array of type character

gradient of temperature:

gradient of temperature:

Advance all the grids over one time step, including all sources

Routine handles the particle evolution

Advance all grids over one time step, but without taking dimensional splitting or split source terms into account

Integrate all grids by one partial step

Payload update

Allocate the cell metrics of all max_blocks blocks in one go. The ixGext range is the one over which dvolume and ds are known, the face areas start one cell lower than ixG in every direction.

allocate arrays on igrid node

allocate arrays on igrid node

allocate arrays on igrid node

allocate arrays on igrid node

allocate memory to physical state of igrid node

allocate memory to physical state of igrid node

allocate memory to physical state of igrid node

allocate memory to physical state of igrid node

allocate memory to one-level coarser physical state of igrid node

allocate memory to one-level coarser physical state of igrid node

allocate memory to one-level coarser physical state of igrid node

allocate memory to one-level coarser physical state of igrid node

This routine fills the fine faces before prolonging. It is the face equivalent of fix_conserve

coarsen and refine blocks to update AMR grid

Construct Morton-order as a global recursive lexicographic ordering.

asign tree node neighor

calculate magnetic field from vector potential A at cell edges

Set boundary conditions according to user data

fill ghost cells at a physical boundary x is this block's slice of bgeo/bgeoc rather than s%x, so that device code never has to dereference a per-block pointer to reach the positions.

Read this module"s parameters from a file

Performa a binary search for the variable 'var_name'

Determine the shape of a block for output (whether to include ghost cells, and on which sides)

Perform Helmholtz operator on a box

Perform Helmholtz operator on a box

Perform Helmholtz operator on a box

Perform Laplacian operator on a box in cylindrical geometry, using (r,z) and (r,phi,z) coordinates in 2D/3D.

Get ghost cells for a fine neighbor

Get ghost cells for a fine neighbor

Get ghost cells for a fine neighbor

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Laplacian operator in cylindrical geometry. TODO: in 3D this does not converge well, maybe it will for a stretched grid.

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Laplacian operator

Perform Gauss-Seidel relaxation on box for a Laplacian operator

Perform Gauss-Seidel relaxation on box for a Laplacian operator

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Helmholtz operator

Perform Gauss-Seidel relaxation on box for a Laplacian operator

Perform Gauss-Seidel relaxation on box for a Laplacian operator

Perform Gauss-Seidel relaxation on box for a Laplacian operator

Perform Helmholtz operator on a box

Perform Helmholtz operator on a box

Perform Helmholtz operator on a box

Perform Laplacian operator on a box

Perform Laplacian operator on a box

Perform Laplacian operator on a box

Perform Helmholtz operator on a box

Perform Helmholtz operator on a box

Perform Helmholtz operator on a box

Perform Laplacian operator on a box

Perform Laplacian operator on a box

Perform Laplacian operator on a box

Enumerate every chunk this rank sends and receives, and lay each peer's chunks out as one contiguous run of the exchange buffer. Fills snd_igrid / snd_ibuf / snd_group for the pack kernels, ibuf_offset for fix_conserve, and the send_pe_ / recv_pe_ runs for the MPI calls.

Compute both corner as well as cell-centered values for output

computes cell corner (xC) and cell center (xCC) coordinates

Interface to add variables to the configuration

Interface to get variables from the configuration

Interface to get variables from the configuration

Get the size of a variable

Get the type of a given variable of a configuration type

Update the variables in the configartion with the values found in 'filename'

Sort the variables for faster lookup

This routine writes the current configuration to a file with descriptions

This routine writes the current configuration to a markdown file

Validate a configured polar axis, once, on the host.

Sample point on a circle with given radius

coarsen one grid to its coarser representative

coarsen sibling blocks into one block

When refining a coarse block with fine neighbours, it is necessary prolong consistently with the already fine faces. This routine takes care of the communication of such faces.

Finalize (or shutdown) the MPI environment

Initialize the MPI environment

SRHD iteration solves for p via NR, and then gives xi as output

SRHD iteration solves for p via NR, and then gives xi as output

Compute number of elements in index range

Standard method for creating a new output file

Cross product of two vectors

Calculate curl of a vector qvec within ixL Options to employ standard second order CD evaluations use Gauss theorem for non-Cartesian grids use Stokes theorem for non-Cartesian grids

clean up destroyed particles on all cores

Solve a diffusion equation implicitly, assuming a constant diffusion coefficient. The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming a constant diffusion coefficient. The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming a constant diffusion coefficient. The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming anisotropic diffusion coefficient which has been stored in mg_iveps1, mg_iveps2, mg_iveps3 (also on coarse grids). The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming anisotropic diffusion coefficient which has been stored in mg_iveps1, mg_iveps2, mg_iveps3 (also on coarse grids). The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming anisotropic diffusion coefficient which has been stored in mg_iveps1, mg_iveps2, mg_iveps3 (also on coarse grids). The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming a variable diffusion coefficient which has been stored in mg_iveps (also on coarse grids). The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming a variable diffusion coefficient which has been stored in mg_iveps (also on coarse grids). The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Solve a diffusion equation implicitly, assuming a variable diffusion coefficient which has been stored in mg_iveps (also on coarse grids). The solution at time t should be stored in mg_iphi, which is on output replaced by the solution at time t+dt.

Calculate divergence of a vector qvec within ixL

Calculate divergence of a vector qvec within ixL using limited extrapolation to cell edges

Limit the centered dwC differences within ixC for iw in direction idim. The limiter is chosen according to typelim.

Routine to ensure that enough storage is allocated for the configuration type. If not the new size will be twice as much as the current size. If no storage is allocated yet a minumum amount of starage is allocated.

Do all local error estimation which determines (de)refinement

Exit MPI-AMRVAC with an error message

Davis (1988) min/max wave speed estimates for HLL, in their relativistic form: wL = min(lambda^-), wR = max(lambda^+) over the left and right states.

Wave speed estimates: min/max acoustic bounds (Davis 1988) Reference: Toro 2010, Chapter 10.

Davis (1988) min/max wave speed estimates wL = min(v_n - c_f), wR = max(v_n + c_f) (fast magnetosonic) over left/right states; used in HLL flux estimation.

Wave speed estimates for HLL/HLLC using Toro (2010) PVRS pressure estimate Implements Eq. (10.67)-(10.69)

Evaluate Implicit part in place, i.e. psa==>F_im(psa)

Exchange and apply the non-local half of a coarsening step, one message per neighbouring rank.

Exchange and apply the migrating blocks, one message per neighbouring rank, with the buffers sized by what actually moves.

fake advance a step to calculate magnetic field

fake update magnetic field from vector potential

Determine fields from MHD variables

Physical boundary conditions

Physical boundary conditions

Build block igrid's cell metrics directly on the device.

Build block igrid's cell metrics directly on the device.

Build block igrid's cell metrics directly on the device.

Build block igrid's cell metrics directly on the device.

This routine fills the particle grid variables with the default for mhd, i.e. only E and B

Fill this block's extra w-variables on the device from usr_set_nwextra.

Adaptive search (combination of linear and binary search) of sorted list for the smallest ix such that list(ix) >= val. On failure, returns size(list)+1

Binary search of sorted list for the smallest ix such that list(ix) >= val. On failure, returns size(list)+1

Linear search of sorted list for the smallest ix such that list(ix) >= val. On failure, returns size(list)+1

Get the id of neighbor nb of boxes(id), through its parent

Get the id of neighbor nb of boxes(id), through its parent

Get the id of neighbor nb of boxes(id), through its parent

find neighors of all blocks

find neighors of level-one root blocks

Deallocate grid variables for particles

Correct the coarse cells abutting a refinement boundary: take out the coarse face flux the update used and put back the sum of the fine ones.

Fix particle position when crossing the 0,2pi boundary in noncartesian coordinates

Make a reconstructed primitive face state admissible before it is handed to the Riemann solver.

Make a reconstructed primitive face state admissible before it is handed to the Riemann solver. Genuinely empty by default: a physics module without auxiliary variables and without a floor of its own needs nothing here, and the call is inlined away.

Make a reconstructed primitive face state admissible before it is handed to the Riemann solver. Genuinely empty by default: a physics module without auxiliary variables and without a floor of its own needs nothing here, and the call is inlined away.

Make a reconstructed primitive face state admissible before it is handed to the Riemann solver. Genuinely empty by default: a physics module without auxiliary variables and without a floor of its own needs nothing here, and the call is inlined away.

Return true if the AMR grid should not be adapted any more. This is controlled by tfixgrid or itfixgrid. Other conditions may be included.

pointwise evaluations used in con2prim compute pointwise value for pressure p and dpdxi

pointwise evaluations used in con2prim compute pointwise value for pressure p and dpdxi

Make every rank's isafety flags known to every rank.

Update payload subroutine

Returns maximum local signal speed from primitive state u in direction flux_dim; used in LLF/TVDLF flux estimation.

Returns maximum local signal speed from primitive state u in direction flux_dim; used in LLF/TVDLF flux estimation.

Returns maximum local signal speed |v_n| + c_f (fast magnetosonic) from primitive state u in direction flux_dim; used in LLF/TVDLF flux estimation.

Returns maximum local signal speed |v_n| + c_f (fast magnetosonic) from primitive state u in direction flux_dim; used in LLF/TVDLF flux estimation.

Compute the current density j = curl(B)

Calculate div B within ixO

Routine to find entries in a string

Routine to find the indices of entries in a string

Compute global maxima of iw variables over the leaves of the grid.

Compute global minima of iw variables over the leaves of the grid.

Block indices, per level, of the blocks a slice passes through.

Get a integer value of a given name

Get a integer array of a given name

Get Lorentz factor from relativistic momentum

Get Lorentz factor from velocity

Get a logical value of a given name

Get a logical array of a given name

Set the Morton range for each processor

Get a real value of a given name

Get a real array of a given name

get physical quantities from simulation result in specific physics module

get physical quantities from simulation result in specific physics module

get physical quantities from simulation result in specific physics module

Get a character value of a given name

Get a character array of a given name

Return the index of the variable with name 'var_name', or -1 if not found.

Compute mean(w**power) over the leaves of the grid. The first mode (power=1) corresponds to the mean, the second to the mean squared values and so on.

Compute the volume average of func(w) over the leaves of the grid.

Compute how much of the domain is covered by each grid level. This routine does not take a non-Cartesian geometry into account.

do update ghost cells of all blocks including physical boundaries

get all variables names

fill inner boundary values

Get first available igrid on processor ipe

Get first available igrid on processor ipe

Get first available igrid on processor ipe

Get first available igrid on processor ipe

Implicit global update step within IMEX schemes, advance psa=psb+dtfactorqdtF_im(psa) First copy all variables from a to b, this is necessary to account for

Calculate gradient of a scalar q within ixL in direction idir

This routine will be called if an error occurs in one of the subroutines of this module.

Let particles evolve in time. The routine also handles grid variables and output.

improve initial condition after initialization

Create and store the MPI types that will be used for parallel communication

build root forest

Initialize grid variables for particles

Initialize a collection of rng's for parallel use

Initialize communicators for particles

fill in initial condition

Initialize amrvac: read par files and initialize variables

Initialize (and allocate) simulation and grid variables

Generate and initialize all grids at the coarsest level (level one)

Return 4-byte integer

Return 8-byte integer

Modified implementation of Koren limiter, to avoid division and the min/max functions, which can be problematic / expensive. In most literature, you have r = a / b (ratio of gradients). Then the limiter phi(r) is multiplied with b. With this implementation, you get phi(r) * b

Modified implementation of Koren limiter, to avoid division and the min/max functions, which can be problematic / expensive. In most literature, you have r = a / b (ratio of gradients). Then the limiter phi(r) is multiplied with b. With this implementation, you get phi(r) * b

Modified implementation of Koren limiter, to avoid division and the min/max functions, which can be problematic / expensive. In most literature, you have r = a / b (ratio of gradients). Then the limiter phi(r) is multiplied with b. With this implementation, you get phi(r) * b

Modified implementation of Koren limiter, to avoid division and the min/max functions, which can be problematic / expensive. In most literature, you have r = a / b (ratio of gradients). Then the limiter phi(r) is multiplied with b. With this implementation, you get phi(r) * b

Give every peer one contiguous run of the exchange buffer, with the chunks inside a run ordered by the key. Both ranks compute the same key for the same chunk, so sorting by it makes the sender's run and the receiver's run agree element for element without any handshake.

build Morton space filling curve for level 1 grid

reallocate blocks into processors for load balance

Relativistic particle integrator

Momentum update subroutine for full Lorentz dynamics

Update payload subroutine

This function returns a new lookup table

This function returns a new lookup table from regularly spaced data

Get the value of a single column at x

Get the value of a single column at a location

Get a location in the lookup table

Fill the column with index col_ix using linearly interpolated data

This function returns a new lookup table

This function returns a new lookup table from regularly spaced data

Get the value of a single column at x

Get the value of a single column at a location

Get a location in the lookup table

Add a new column by linearly interpolating the (x, y) data

This function returns a new lookup table

This function returns a new lookup table from regularly spaced data

Read the lookup table from file (in binary, potentially unportable)

Get the value of a single column at x

Get the value of a single column at a location

Get the x-coordinates and the columns of the lookup table

Get a location in the lookup table

Get the values of all columns at x

Get the values of all columns at a location

Linearly interpolate the (x, y) input data to the new_x coordinates

Returns the x-coordinates of the lookup table

Compute by use of linear interpolation the value in the middle

Fill the column with index col_ix using the linearly interpolated (x, y) data

Write the lookup table to file (in binary, potentially unportable)

The main driver, called after MPI and GPUs are initialized

Monotonised central-difference limiter with tunable beta (AMRVAC's mcbeta; beta=2 recovers the classic MC limiter, van Leer 1979)

Monotonised central-difference limiter with tunable beta (AMRVAC's mcbeta; beta=2 recovers the classic MC limiter, van Leer 1979)

Monotonised central-difference limiter with tunable beta (AMRVAC's mcbeta; beta=2 recovers the classic MC limiter, van Leer 1979)

Monotonised central-difference limiter with tunable beta (AMRVAC's mcbeta; beta=2 recovers the classic MC limiter, van Leer 1979)

Allocate communication buffers and local boxes for a tree that has already been created

Allocate communication buffers and local boxes for a tree that has already been created

Allocate communication buffers and local boxes for a tree that has already been created

Apply operator to the tree and store in variable i_out

Apply operator to the tree and store in variable i_out

Apply operator to the tree and store in variable i_out

Prolong from a parent to a child with index offset dix. This method could sometimes work better than the default prolongation, which does not take the variation in epsilon into account Initialize MPI if needed, and store MPI information

Prolong from a parent to a child with index offset dix. This method could sometimes work better than the default prolongation, which does not take the variation in epsilon into account Initialize MPI if needed, and store MPI information

Prolong from a parent to a child with index offset dix. This method could sometimes work better than the default prolongation, which does not take the variation in epsilon into account Initialize MPI if needed, and store MPI information

Set multigrid boundary conditions for the solution according to variable iw

Copy a variable from the multigrid tree

Copy from multigrid tree with one layer of ghost cells. Corner ghost cells are not used/set.

Copy a variable to the multigrid tree, including a layer of ghost cells

Deallocate all allocatable arrays

Deallocate all allocatable arrays

Deallocate all allocatable arrays

Perform FAS-FMG cycle (full approximation scheme, full multigrid).

Perform FAS-FMG cycle (full approximation scheme, full multigrid).

Perform FAS-FMG cycle (full approximation scheme, full multigrid).

Perform FAS V-cycle (full approximation scheme).

Perform FAS V-cycle (full approximation scheme).

Perform FAS V-cycle (full approximation scheme).

Fill ghost cells at all grid levels

Fill ghost cells at all grid levels

Fill ghost cells at all grid levels

Fill ghost cells at a grid level

Fill ghost cells at a grid level

Fill ghost cells at a grid level

Get the offset of a box with respect to its parent (e.g. in 2d, there can be a child at offset 0,0, one at n_cell/2,0, one at 0,n_cell/2 and one at n_cell/2, n_cell/2) n_cell / 2

Get the offset of a box with respect to its parent (e.g. in 2d, there can be a child at offset 0,0, one at n_cell/2,0, one at 0,n_cell/2 and one at n_cell/2, n_cell/2) n_cell / 2

Get the offset of a box with respect to its parent (e.g. in 2d, there can be a child at offset 0,0, one at n_cell/2,0, one at 0,n_cell/2 and one at n_cell/2, n_cell/2) n_cell / 2

Get coordinates at the face of a box

Get coordinates at the face of a box

Get coordinates at the face of a box

Specify minimum buffer size (per process) for communication

Specify minimum buffer size (per process) for communication

Specify minimum buffer size (per process) for communication

Return .true. if a box has children

Return .true. if a box has children

Return .true. if a box has children

Compute the child index for a box with spatial index ix. With child index we mean the index in the children(:) array of its parent.

Compute the child index for a box with spatial index ix. With child index we mean the index in the children(:) array of its parent.

Compute the child index for a box with spatial index ix. With child index we mean the index in the children(:) array of its parent.

Load balance all boxes in the multigrid tree. Compared to mg_load_balance_simple, this method does a better job of setting the ranks of parent boxes

Load balance all boxes in the multigrid tree. Compared to mg_load_balance_simple, this method does a better job of setting the ranks of parent boxes

Load balance all boxes in the multigrid tree. Compared to mg_load_balance_simple, this method does a better job of setting the ranks of parent boxes

Load balance the parents (non-leafs). Assign them to the rank that has most children.

Load balance the parents (non-leafs). Assign them to the rank that has most children.

Load balance the parents (non-leafs). Assign them to the rank that has most children.

Load balance all boxes in the multigrid tree, by simply distributing the load per grid level. This method will only work well for uniform grids.

Load balance all boxes in the multigrid tree, by simply distributing the load per grid level. This method will only work well for uniform grids.

Load balance all boxes in the multigrid tree, by simply distributing the load per grid level. This method will only work well for uniform grids.

Determine total number of unknowns (on leaves)

Determine total number of unknowns (on leaves)

Determine total number of unknowns (on leaves)

Store boundary conditions for the solution variable, this can speed up calculations if the same boundary conditions are re-used.

Store boundary conditions for the solution variable, this can speed up calculations if the same boundary conditions are re-used.

Store boundary conditions for the solution variable, this can speed up calculations if the same boundary conditions are re-used.

Prolong variable iv from lvl to variable iv_to at lvl+1

Prolong variable iv from lvl to variable iv_to at lvl+1

Prolong variable iv from lvl to variable iv_to at lvl+1

Specify minimum buffer size (per process) for communication

Specify minimum buffer size (per process) for communication

Specify minimum buffer size (per process) for communication

Prolong from a parent to a child with index offset dix

Prolong from a parent to a child with index offset dix

Prolong from a parent to a child with index offset dix

Restrict all levels

Restrict all levels

Restrict all levels

Specify minimum buffer size (per process) for communication

Specify minimum buffer size (per process) for communication

Specify minimum buffer size (per process) for communication

Restrict from lvl to lvl-1

Restrict from lvl to lvl-1

Restrict from lvl to lvl-1

Create a list of leaves and a list of parents for a level

Create a list of leaves and a list of parents for a level

Create a list of leaves and a list of parents for a level

Create a list of refinement boundaries (from the coarse side)

Create a list of refinement boundaries (from the coarse side)

Create a list of refinement boundaries (from the coarse side)

Setup multigrid for usage

Generate a multigrid tree that includes the amrvac tree, but also contains coarser grid levels. A number of checks has already been performed in mg_setup_multigrid, so we don't repeat these checks here.

If the grid has changed, rebuild the full multigrid tree

modify initial condition

Determine most popular rank in the list. In case of ties, assign the rank with the least work.

Determine most popular rank in the list. In case of ties, assign the rank with the least work.

Determine most popular rank in the list. In case of ties, assign the rank with the least work.

Exit MPI-AMRVAC with an error message

MUSCL reconstruction in primitive variables for two faces using a 5-point stencil. Returns uL(:,iface), uR(:,iface) for iface=1 (between cells 2-3) and iface=2 (between 3-4).

MUSCL reconstruction in primitive variables for two faces using a 5-point stencil. Returns uL(:,iface), uR(:,iface) for iface=1 (between cells 2-3) and iface=2 (between 3-4).

MUSCL reconstruction in primitive variables for two faces using a 5-point stencil. Returns uL(:,iface), uR(:,iface) for iface=1 (between cells 2-3) and iface=2 (between 3-4).

MUSCL reconstruction in primitive variables for two faces using a 5-point stencil. Returns uL(:,iface), uR(:,iface) for iface=1 (between cells 2-3) and iface=2 (between 3-4).

Interal routine: get the next value (returned as 64 bit signed integer)

Single normal random number

Helper routine for quicksort, to perform partitioning

Convert to noncartesian coordinates

Convert position to Cartesian coordinates

Give initial values to parameters

Calculate idirmin and the idirmin:3 components of the common current array make sure that dxlevel(^D) is set correctly.

Check if particle is inside computational domain

Quick check if particle is still in igrid

Create initial particles

Finalize the particles module

Initialize particle data and parameters

Read this module's parameters from a file

Convert vector from Cartesian coordinates

Convert vector to Cartesian coordinates

Initialize the module

Initialize the module GLM MHD uses split source addition in psi:

Initialize the module

Initialize the module

Transform primitive variables into conservative ones

Transform primitive variables into conservative ones

Transform primitive variables into conservative ones

Transform primitive variables into conservative ones

Transform conservative variables into primitive ones

Transform conservative variables into primitive ones

Transform conservative variables into primitive ones

Transform conservative variables into primitive ones

here no SI_UNIT used by default, to be implemented

here no SI_UNIT used by default, to be implemented eq_state_units by default, to be implemented

here no SI_UNIT used by default, to be implemented eq_state_units by default, to be implemented

here no SI_UNIT used by default, to be implemented eq_state_units by default, to be implemented

Point the metric components of state s at block igrid's slice of geo. Bounds remapping is what keeps the index ranges (which start below 1 for the face areas, and for the volumes when nghostcells is odd) intact: a plain pointer assignment would silently rebase them to 1.

Point the metric components of state s at block igrid's slice of geo. Bounds remapping is what keeps the index ranges (which start below 1 for the face areas, and for the volumes when nghostcells is odd) intact: a plain pointer assignment would silently rebase them to 1.

Point the metric components of state s at block igrid's slice of geo. Bounds remapping is what keeps the index ranges (which start below 1 for the face areas, and for the volumes when nghostcells is odd) intact: a plain pointer assignment would silently rebase them to 1.

Point the metric components of state s at block igrid's slice of geo. Bounds remapping is what keeps the index ranges (which start below 1 for the face areas, and for the volumes when nghostcells is odd) intact: a plain pointer assignment would silently rebase them to 1.

Quick check if particle coordinate is inside igrid (ghost cells included, except the last ngh)

Return Poisson random variate with rate lambda. Works well for lambda < 30 or so. For lambda >> 1 it can produce wrong results due to roundoff error.

Helper routine to get variables. This is useful because a lot of the same code is executed for the different types of variables.

Helper routine to store variables. This is useful because a lot of the same code is executed for the different types of variables.

Write volume-averaged values and other information to the log file

Print a log that can be used to check whether the code still produces the same output (regression test)

process is a user entry in time loop, before output and advance allows to modify solution, add extra variables, etc. Warning: CFL dt already determined (and is not recomputed)!

process_advanced is user entry in time loop, just after advance allows to modify solution, add extra variables, etc. added for handling two-way coupled PIC-MHD Warning: w is now at global_time^(n+1), global time and iteration at global_time^n, it^n

do 1st order prolongation

do 2nd order prolongation

This subroutine performs a 2nd order prolongation for a staggered field F, preserving the divergence of the coarse cell. This is useful for preserving DivF=0. If DivF=f(x), a different algorithm must be used.

prolong one block

Prolong onto a child box

Prolong onto a child box

Prolong onto a child box

In case the fine grid is on a different CPU, perform the prolongation and store the fine-grid values in the send buffer.

In case the fine grid is on a different CPU, perform the prolongation and store the fine-grid values in the send buffer.

In case the fine grid is on a different CPU, perform the prolongation and store the fine-grid values in the send buffer.

For all grids on all processors, do a check on refinement flags. Make sure that neighbors will not differ more than one level of refinement.

Simple implementation of quicksort algorithm to sort the variable list alphabetically.

The radial coordinate map, from the logical coordinate s to the physical radius r, and its inverse.

the params require to be read in phys template file phys_init

the params require to be read in phys template file phys_init

the params require to be read in phys template file phys_init

Radiative cooling initialization

Radiative cooling initialization

Radiative cooling initialization

Rank key(1:n) ascending: perm(k) is the index of the k-th smallest key. Heapsort, so O(n log n) worst case and no scratch beyond perm itself. Local rather than the equivalent mrgrnk vendored with octree-mg, to keep the callers independent of the multigrid solver.

Read the command line arguments passed to agile

Set values according to user data

Read in the user-supplied parameter-file

Read this module's parameters from a file

Read this module's parameters from a file

Read this module's parameters from a file

Read this module's parameters from a file

Routine to read in snapshots (.dat files). When it cannot recognize the file version, it will automatically try the 'old' reader.

re-calculate the magnetic field from the vector potential in a completely divergency free way

MUSCL (primitive-variable) reconstruction with slope limiter; HLL two-wave approximate Riemann flux at faces. Uses estimated left/right signal speeds (Davis (1988)) for less diffusion than LLF, no contact resolution.

MUSCL (primitive-variable) reconstruction with slope limiter; HLL two-wave approximate Riemann flux at faces. Uses estimated left/right signal speeds (Davis (1988)) for less diffusion than LLF, no contact resolution.

MUSCL (primitive-variable) reconstruction with slope limiter; HLL two-wave approximate Riemann flux at faces. Uses estimated left/right signal speeds (Davis (1988)) for less diffusion than LLF, no contact resolution.

MUSCL (primitive-variable) reconstruction with slope limiter; HLL two-wave approximate Riemann flux at faces. Uses estimated left/right signal speeds (Davis (1988)) for less diffusion than LLF, no contact resolution.

MUSCL (primitive-variable) reconstruction with slope limiter; HLLC approximate Riemann flux at faces. Restores the contact wave (and shear in Euler/HD), typically sharper than HLL for similar cost.

MUSCL (primitive-variable) reconstruction with slope limiter; HLLC approximate Riemann flux at faces. Restores the contact wave (and shear in Euler/HD), typically sharper than HLL for similar cost.

MUSCL (primitive-variable) reconstruction with slope limiter; HLLC approximate Riemann flux at faces. Restores the contact wave (and shear in Euler/HD), typically sharper than HLL for similar cost.

MUSCL (primitive-variable) reconstruction with slope limiter; HLLC approximate Riemann flux at faces. Restores the contact wave (and shear in Euler/HD), typically sharper than HLL for similar cost.

MUSCL (primitive-variable) reconstruction with slope limiter; LLF/Rusanov numerical flux at faces. Robust and diffusive; uses local max wavespeed for upwinding. Adaptive diffusion (Rempel et al. 2009) is compiled in only when FLUX_AD is defined.

MUSCL (primitive-variable) reconstruction with slope limiter; LLF/Rusanov numerical flux at faces. Robust and diffusive; uses local max wavespeed for upwinding. Adaptive diffusion (Rempel et al. 2009) is compiled in only when FLUX_AD is defined.

MUSCL (primitive-variable) reconstruction with slope limiter; LLF/Rusanov numerical flux at faces. Robust and diffusive; uses local max wavespeed for upwinding. Adaptive diffusion (Rempel et al. 2009) is compiled in only when FLUX_AD is defined.

MUSCL (primitive-variable) reconstruction with slope limiter; LLF/Rusanov numerical flux at faces. Robust and diffusive; uses local max wavespeed for upwinding. Adaptive diffusion (Rempel et al. 2009) is compiled in only when FLUX_AD is defined.

Reconstruct scalar q within ixO^L to 1/2 dx in direction idir Return both left and right reconstructed values

refine one block to its children blocks

reset AMR and (de)allocate boundary flux storage at level changes

Force the tree to desired level(s) from level_io(_min/_max) used for conversion to vtk output

Resize the storage size of variable, which can be of type integer, logical, real or character

Read this module's parameters from a file

One-face HLL numerical flux from primitive L/R states. takes flux_type(flux_dim,nw_flux) into account (fallback to LLF for selected variables)

One-face HLL numerical flux from primitive L/R states. takes flux_type(flux_dim,nw_flux) into account (fallback to LLF for selected variables)

One-face HLL numerical flux from primitive L/R states. takes flux_type(flux_dim,nw_flux) into account (fallback to LLF for selected variables)

One-face HLL numerical flux from primitive L/R states. takes flux_type(flux_dim,nw_flux) into account (fallback to LLF for selected variables)

One-face HLLC numerical flux from primitive L/R states (HD only). Uses scalar pressure assumption; not valid for MHD (use HLLD instead). Reference: Toro (2010), chapter 10 (Variant 2) does not yet take flux_type(flux_dim,nw_flux) into account

One-face HLLC numerical flux from primitive L/R states (HD only). Uses scalar pressure assumption; not valid for MHD (use HLLD instead). Reference: Toro (2010), chapter 10 (Variant 2) does not yet take flux_type(flux_dim,nw_flux) into account

One-face HLLC numerical flux from primitive L/R states (HD only). Uses scalar pressure assumption; not valid for MHD (use HLLD instead). Reference: Toro (2010), chapter 10 (Variant 2) does not yet take flux_type(flux_dim,nw_flux) into account

One-face HLLC numerical flux from primitive L/R states (HD only). Uses scalar pressure assumption; not valid for MHD (use HLLD instead). Reference: Toro (2010), chapter 10 (Variant 2) does not yet take flux_type(flux_dim,nw_flux) into account

One-face LLF/Rusanov numerical flux from primitive L/R states. phi(nw_flux) is the adaptive-diffusion reduction factor; only present when FLUX_AD is defined.

One-face LLF/Rusanov numerical flux from primitive L/R states. phi(nw_flux) is the adaptive-diffusion reduction factor; only present when FLUX_AD is defined.

One-face LLF/Rusanov numerical flux from primitive L/R states. phi(nw_flux) is the adaptive-diffusion reduction factor; only present when FLUX_AD is defined.

One-face LLF/Rusanov numerical flux from primitive L/R states. phi(nw_flux) is the adaptive-diffusion reduction factor; only present when FLUX_AD is defined.

Helper function for next()

Inverse of r_of_s. See there.

Payload update

Pack the outgoing flux chunks and post one message per peer. Where each chunk goes was decided by build_message_layout; nothing here walks the tree.

Set the coordinate system to be used

Set the coordinate system from the GEOM compile-time define.

Set the index range over which the cell metrics are known. With an even number of ghost cells this is just the ghosted block; with an odd number (or a staggered grid) ghost-cell prolongation needs one extra layer.

The routine that actually fills the ghost cells

The routine that actually fills the ghost cells

The routine that actually fills the ghost cells

This routine sets the indexes for the correction of the circulation according to several different cases, as grids located in different cpus, presence of corners, and different relative locations of the fine grid respect to the coarse one

Set a seed for the rng

setdt - set dt for all levels between levmin and levmax. dtpar>0 --> use fixed dtpar for all level dtpar<=0 --> determine CFL limited timestep

setdt - set dt for all levels between levmin and levmax. dtpar>0 --> use fixed dtpar for all level dtpar<=0 --> determine CFL limited timestep

setdt - set dt for all levels between levmin and levmax. dtpar>0 --> use fixed dtpar for all level dtpar<=0 --> determine CFL limited timestep

setdt - set dt for all levels between levmin and levmax. dtpar>0 --> use fixed dtpar for all level dtpar<=0 --> determine CFL limited timestep

Build up AMR

added for nvidia compilers

Fill ghost cells near refinement boundaries which preserves diffusive fluxes.