pearl-public/pmsco-public
2
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

public distro 2.1.0

Browse Source
This commit is contained in:
muntwiler_m
2019-07-19 12:54:54 +02:00
parent acea809e4e
commit fbd2d4fa8c
40 changed files with 2813 additions and 345 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
docs/config.dox
View File

@@ -763,6 +763,7 @@ src/introduction.dox \
src/concepts.dox \
src/concepts-tasks.dox \
src/concepts-emitter.dox \
src/concepts-atomscat.dox \
src/installation.dox \
src/execution.dox \
src/commandline.dox \

27
docs/src/commandline.dox
View File

@@ -21,9 +21,6 @@ Do not include the extension <code>.py</code> or a trailing slash.
@c path/to/project.py should be the path and name to your project module.
Common args and project args are described below.
Note: In contrast to earlier versions, the project module is not executed directly any more.
Rather, it is loaded by the main pmsco module as a 'plug-in'.
\subsection sec_common_args Common Arguments
@@ -43,15 +40,14 @@ The following table is ordered by importance.
| --log-level | DEBUG, INFO, WARNING (default), ERROR, CRITICAL | Minimum level of messages that should be added to the log. |
| --log-file | file system path | Name of the main log file. Under MPI, the rank of the process is inserted before the extension. Default: output-file + log, or pmsco.log. |
| --log-disable | | Disable logging. By default, logging is on. |
| --pop-size | integer | Population size (number of particles) in swarm optimization mode. The default value is the greater of 4 or two times the number of calculation processes. |
| --pop-size | integer | Population size (number of particles) in swarm and genetic optimization mode. The default value is the greater of 4 or the number of parallel calculation processes. |
| --seed-file | file system path | Name of the population seed file. Population data of previous optimizations can be used to seed a new optimization. The file must have the same structure as the .pop or .dat files. See @ref pmsco.project.Project.seed_file. |
| --table-file | file system path | Name of the model table file in table scan mode. |
| -c, --code | edac (default) | Scattering code. At the moment, only edac is supported. |
\subsubsection sec_file_categories File Categories
The following category names can be used with the @c --keep-files option.
The following category names can be used with the `--keep-files` option.
Multiple names can be specified and must be separated by spaces.
| Category | Description | Default Action |
@@ -59,7 +55,7 @@ Multiple names can be specified and must be separated by spaces.
| all | shortcut to include all categories | |
| input | raw input files for calculator, including cluster and phase files in custom format | delete |
| output | raw output files from calculator | delete |
| phase | phase files in portable format for report | delete |
| atomic | atomic scattering and emission files in portable format | delete |
| cluster | cluster files in portable XYZ format for report | keep |
| debug | debug files | delete |
| model | output files in ETPAI format: complete simulation (a_-1_-1_-1_-1) | keep |
@@ -67,9 +63,20 @@ Multiple names can be specified and must be separated by spaces.
| symmetry | output files in ETPAI format: symmetry (a_b_c_-1_-1) | delete |
| emitter | output files in ETPAI format: emitter (a_b_c_d_-1) | delete |
| region | output files in ETPAI format: region (a_b_c_d_e) | delete |
| report| final report of results | keep |
| report| final report of results | keep always |
| population | final state of particle population | keep |
| rfac | files related to models which give bad r-factors | delete |
| rfac | files related to models which give bad r-factors, see warning below | delete |
\note
The `report` category is always kept and cannot be turned off.
The `model` category is always kept in single calculation mode.
\warning
If you want to specify `rfac` with the `--keep-files` option,
you have to add the file categories that you want to keep, e.g.,
`--keep-files rfac cluster model scan population`
(to return the default categories for all calculated models).
Do not specify `rfac` alone as this will effectively not return any file.
\subsection sec_project_args Project Arguments
@@ -125,4 +132,4 @@ The job script is written to @c $DESTDIR/$JOBNAME which is also the destination
| MODE | single, swarm, grid, genetic | PMSCO operation mode. This value is passed on to PMSCO as the @c --mode argument. |
| ARGS (optional) | | Any further arguments are passed on verbatim to PMSCO. You don't need to specify the mode and time limit here. |
*/
*/

114
docs/src/concepts-atomscat.dox Normal file
View File

@@ -0,0 +1,114 @@
/*! @page pag_concepts_atomscat Atomic scattering
\section sec_atomscat Atomic scattering
\subsection sec_atomscat_intro Introduction
The process of calculating atomic scattering factors (phase shifts) can be customized in several ways.
1. Internal processing.
Some multiple scattering programs, like EDAC, contain a built-in facility to calculate phase shifts.
This is the most simple and default behaviour.
2. Automatic calculation in a separate program.
PMSCO has an interface to run the PHAGEN program from
the [MsSpec-1.0 package](https://ipr.univ-rennes1.fr/msspec) to calculate scattering factors.
Note that the PHAGEN code is not included in the public distribution of PMSCO.
3. Manual calculation.
Scattering files created manually using an external program can be used by providing the file names.
The files must have the format required by the multiple scattering code,
and they must be linked to the corresponding atoms of the cluster.
In the case of automatic calculation, the project code can optionally hook into the process
and modify clusters before and after scattering factors are calculated.
For instance, it may provide an extended cluster in order to reduce boundary effects,
or it may modify the assignment of scattering files to cluster atoms
so that the scattering factors of selected atom classes are used
(cf. section \ref sec_atomscat_atomclass).
\subsection sec_atomscat_usage Usage
\subsubsection sec_atomscat_internal Internal processing
This is the default behaviour selected in the inherited pmsco.project.Project class.
Make sure not to override the `atomic_scattering_factory` attribute.
Its default value is pmsco.calculators.calculator.InternalAtomicCalculator.
\subsubsection sec_atomscat_external Automatic calculation in a separate program
To select the atomic scattering calculator,
assign its interface class to the project's `atomic_scattering_factory` attribute.
For example, to use PHAGEN, add the following code to your project's `__init__` constructor:
@code{.py}
from pmsco.calculators.phagen import PhagenCalculator
self.atomic_scattering_factory = PhagenCalculator
@endcode
\subsubsection sec_atomscat_manual Manual calculation
If you want to keep the scattering factors constant during an optimization,
you should run PMSCO in _single_ mode and provide the model parameters and cluster
that will return the desired scattering files.
In the `create_params` method of your project,
you should then set the `phase_files` attribute,
which is a dictionary that maps atom classes to the names of the scattering files.
Unless you set specific values in the cluster object, the atom class defaults to the element number.
The file names should include a path relative to the working directory.
\subsection sec_atomscat_implement Implementation
\subsubsection sec_atomscat_atomclass Atom classes
Atomic scattering programs classify atoms based on chemical element, charge state and symmetry of the local environment.
This means that two atoms of the same chemical element may have different scattering factors.
For example, if you have EDAC output the cluster after calculation of the muffin tin potential,
you will find that the chemical element number has been replaced by an arbitrary integer.
By default, PMSCO will do the linking of atom classes and scattering files transparently.
However, if you want to reduce the number of atom classes,
or if you have the scattering factors calculated on a reference cluster,
you will have to provide project code to do the assignment.
This is described further below.
\subsubsection sec_atomscat_calculator Atomic scattering calculator
The project selects the atomic scattering calculation mode by specifying its `atomic_scattering_factory` attributed.
This is the name of a class that inherits from @ref pmsco.calculators.calculator.AtomicCalculator.
The following calculators are currently implemented:
| Class | Description |
| --- | --- |
| pmsco.calculators.calculator.InternalAtomicCalculator | Calculate the atomic scattering factors in the multiple-scattering program. |
| pmsco.calculators.phagen.PhagenCalculator | Calculate the atomic scattering factors in the PHAGEN program. |
An atomic calculator class essentially defines a `run` method that operates on a cluster and scattering parameters object.
It generates the necessary scattering files, updates the cluster with the new atom classes
and updates the parameters with the file names of the scattering files.
Note that the scattering files have to be in the correct format for the multiple scattering calculator.
\subsubsection sec_atomscat_hooks Project hooks
Before and after calculation of the scattering factors,
the project's `before_atomic_scattering` and `after_atomic_scattering` methods are called
with the cluster and input parameters.
The _before_ method provides the cluster to be used for atomic scattering calculations.
It may,
1. just return the original cluster,
2. modify the provided cluster to include additional atoms or modify the charge state of the emitter,
3. create a completely different cluster,
4. return None to suppress the atomic scattering calculation.
The method is called once at the beginning of the PMSCO job with model -1,
where it may return the global reference cluster.
Later on it is called once for each calculation task with the specific task index.
Similarly, the _after_ method collects the results and updates the `phase_files` dictionary of the input parameters.
It is free to consolidate atom classes and remove unwanted atoms.
However, it must make sure that for each atom class in the cluster,
there is a corresponding link to a scattering file.
*/

23
docs/src/installation.dox
View File

@@ -39,8 +39,8 @@ The code depends on the following libraries:
- BLAS
- LAPACK
- Python 2.7 or 3.6
- Numpy >= 1.11
- Python packages from PyPI listed in the requirements.txt file
- Numpy >= 1.13
- Python packages listed in the requirements.txt file
Most of these requirements are available from the Linux distribution.
For an easily maintainable Python environment, [Miniconda](https://conda.io/miniconda.html) is recommended.
@@ -50,11 +50,11 @@ and it's difficult to switch between different Python versions.
On the PSI cluster machines, the environment must be set using the module system and conda (on Ra).
Details are explained in the PEARL Wiki.
PMSCO runs under Python 2.7 or Python 3.6 or higher.
Since Python 2 is being deprecated, the code has been ported to Python 3.6.
Compatibility with Python 2.7 is maintained by using
the [future package](http://python-future.org/compatible_idioms.html).
New code should be written according to their guidelines.
PMSCO runs under Python 2.7 or Python 3.6.
Since Python 2 is being deprecated, Python 3.6 is recommended.
Compatibility with Python 2.7 is currently maintained by using
the [future package](http://python-future.org/compatible_idioms.html)
but may be dropped at any time.
\subsection sec_install_instructions Instructions
@@ -86,7 +86,6 @@ nano \
openmpi-bin \
openmpi-common \
sqlite3 \
swig \
wget
@endcode
@@ -102,11 +101,11 @@ Install Miniconda according to their [instructions](https://conda.io/docs/user-g
then configure the Python environment:
@code{.sh}
conda create -q --yes -n pmsco python=2.7
conda create -q --yes -n pmsco python=3.6
source activate pmsco
conda install -q --yes -n pmsco \
pip \
numpy \
"numpy>=1.13" \
scipy \
ipython \
mpi4py \
@@ -114,7 +113,9 @@ conda install -q --yes -n pmsco \
nose \
mock \
future \
statsmodels
statsmodels \
swig \
gitpython
pip install periodictable attrdict fasteners
@endcode

20
docs/src/introduction.dox
View File

@@ -9,13 +9,15 @@ The actual scattering calculation is done by code developed by other parties.
While the scattering program typically calculates a diffraction pattern based on a set of static parameters and a specific coordinate file in a single process,
PMSCO wraps around that program to facilitate parameter handling, cluster building, structural optimization and parallel processing.
In the current version, the [EDAC](http://garciadeabajos-group.icfo.es/widgets/edac/) code
developed by F. J. García de Abajo, M. A. Van Hove, and C. S. Fadley (1999) is used for scattering calculations.
Other code can be integrated as well.
Initially, support for the MSC program by Kaduwela, Friedman, and Fadley was planned but is currently not maintained.
PMSCO is written in Python 2.7.
EDAC is written in C++, MSC in Fortran.
PMSCO interacts with the calculation programs through Python wrappers for C++ or Fortran.
In the current version, PMSCO can make use of the following programs.
Other programs may be integrated as well.
- [EDAC](http://garciadeabajos-group.icfo.es/widgets/edac/)
by F. J. García de Abajo, M. A. Van Hove, and C. S. Fadley,
[Phys. Rev. B 63 (2001) 075404](http://dx.doi.org/10.1103/PhysRevB.63.075404)
- PHAGEN from the [MsSpec package](https://ipr.univ-rennes1.fr/msspec)
by C. R. Natoli and D. Sébilleau,
[Comp. Phys. Comm. 182 (2011) 2567](http://dx.doi.org/10.1016/j.cpc.2011.07.012)
\section sec_intro_highlights Highlights
@@ -63,11 +65,11 @@ An open distribution of PMSCO is available under the [Apache License, Version 2.
- Please acknowledge the use of the code.
- Please share your development of the code with the original author.
Due to different copyright, the MSC and EDAC programs are not contained in the public software repository.
Due to different copyright terms, the third-party calculation programs are not contained in the public software repository.
These programs may not be used without an explicit agreement by the respective original authors.
\author Matthias Muntwiler, <mailto:matthias.muntwiler@psi.ch>
\version This documentation is compiled from version $(REVISION).
\copyright 2015-2018 by [Paul Scherrer Institut](http://www.psi.ch)
\copyright 2015-2019 by [Paul Scherrer Institut](http://www.psi.ch)
\copyright Licensed under the [Apache License, Version 2.0](http://www.apache.org/licenses/LICENSE-2.0)
*/

54
docs/src/optimizers.dox
View File

@@ -30,6 +30,55 @@ The domain parameters have the following meanings:
| step | Not used. |
\subsubsection sec_opt_seed Seeding a population
By default, one particle is initialized with the start value declared in the parameter domain,
and the other are set to random values within the domain.
You may initialize more particles of the population with specific values by providing a seed file.
The seed file must have a similar format as the result `.dat` files
with a header line specifying the column names and data rows containing the values for each particle.
A good practice is to use a previous `.dat` file and remove unwanted rows.
To continue an interrupted optimization,
the `.dat` file from the previous optimization can be used as is.
The seeding procedure can be tweaked by several optimizer parameters (see above).
PMSCO normally loads the first rows up to population size - 1 or up to the `seed_limit` parameter,
whichever is lower.
If an `_rfac` column is present, the file is first sorted by R-factor and only the best models are loaded.
Models that resulted in an R-factor above the `rfac_limit` parameter are always ignored.
During the optimization process, all models loaded from the seed file are normally re-calculated.
This may waste CPU time if the calculation is run under the same conditions
and would result in exactly the same R-factor,
as is the case if the seed is used to continue a previous optimization, for example.
In these situations, the `recalc_seed` parameter can be set to False,
and PMSCO will use the R-factor value from the seed file rather than calculating the model again.
\subsubsection sec_opt_patch Patching a running optimization
While an optimization process is running, the user can manually patch the population with arbitrary values,
for instance, to kick the population out of a local optimum or to drive it to a less sampled parameter region.
To patch a running population, prepare a population file named `pmsco_patch.pop` and copy it to the work directory.
The file must have a similar format as the result `.dat` files
with a header line specifying the column names and data rows containing the values.
It should contain as many rows as particles to be patched but not more than the size of the population.
The columns must include a `_particle` column which specifies the particle to patch
as well as the model parameters to be changed.
Parameters that should remain unaffected can be left out,
extra columns including `_gen`, `_rfac` etc. are ignored.
PMSCO checks the file for syntax errors and ignores it if errors are present.
Parameter values that lie outside the domain boundary are ignored.
Successful or failed patching is logged at warning level.
The patch file is re-applied whenever its time stamp has changed.
\attention Do not edit the patch file in the working directory
to prevent it from being read in an unfinished state or multiple times.
\subsection sec_opt_genetic Genetic optimization
The genetic algorithm evolves a population of individuals
@@ -73,11 +122,14 @@ The domain parameters have the following meanings:
| max | Upper limit of the parameter range. |
| step | Standard deviation of the Gaussian distribution of weak mutations. The step should not be much lower than the the parameter range divided by the population size and not greater than one third of the parameter range. |
The population of the genetic optimizer can be seeded and patched in the same way as the particle swarm,
cf. sections @ref sec_opt_seed and @ref sec_opt_swarm.
\subsection sec_opt_grid Grid search
The grid search algorithm samples the parameter space at equidistant steps.
The order of calculations is randomized so that early results represent various parts of the parameter space.
The order of calculations is randomized so that distant parts of the parameter space are sampled at an early stage.
| Parameter | Description |
| --- | --- |

2
docs/src/uml/project-classes.puml
View File

@@ -25,7 +25,7 @@ class Scan {
thetas
phis
alphas
set_scan()
import_scan_file()
}
class Domain {