Rework of Particle class, and Particle data creation by VeckoTheGecko · Pull Request #2134 · Parcels-code/Parcels

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko · 2025-08-05T13:13:09Z

Changes:

Cleares up Variable and Particle classes, making the dictinction between a ParticleClass (a type of Particle), and a KernelParticle (an accessor class used to read and set particle data in the arrays). This clarification allows for clearer abstractions and consolidation of responsibility (helping with testing).
Removed lonlatdepth_dtype from particleset init
- Created a public helper function get_default_particle(dtype: np.float32 | np.float64) which allows users to create particles with the right precision from the beginning (rather than redefining after the fact).
Consolidate particle data creation to a function create_particle_data that works from the particle class, ngrids, time_interval, initial
Remove MPI related code from ParticleFile, and some cleanup of the ParticleFile class (particlefile remains untested at the moment - more changes will come in a future PR)

Would be good to also discuss variable naming - let me know if you have any suggestions.

Chose the correct base branch (v4-dev for v4 changes)
Fixes #
Added tests

…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Make it explicitly an initial dict rather than as kwargs

Not sure why this is necessary since I only refactored the way the particle data was created?

erikvansebille · 2025-08-06T05:17:53Z

So the code (see below) to temporarily set time to np.nat until we know the sign of dt in execute() stays as it is?
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L100-L103
and
https://github.com/OceanParcels/Parcels/blob/30aec22d2939e285bfbd44f6617e36019002f6d3/parcels/particleset.py#L536-L540
That would be fine with me; but it could perhaps be a bit more transparently coded?

I see, something to keep in mind when we refactor the particleset execute loop

@erikvansebille

For internal consistency with the CF trajectory output spec. cc @erikvansebille

VeckoTheGecko · 2025-08-06T15:55:28Z

@erikvansebille after syncing this branch with v4-dev, I am getting some strange infinite loop bugs in the following tests to do with the RK45 kernel ..... I'm not really sure how to go about solving them at the moment, perhaps if you have a second tomorrow we can look through together since you're familiar with the rk45 kernel? (or can pytest.skip them for now until we have time to work through this. I think the pfile stuff needs to remain my priority for now

pytest 'tests/v4/test_advection.py::test_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_decaying_moving_eddy[RK45-1e-05]' 'tests/v4/test_advection.py::test_gyre_flowfields[stommel_gyre-A-RK45-1]' 'tests/v4/test_advection.py::test_gyre_flowfields[peninsula-A-RK45-1]'

For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

VeckoTheGecko added 21 commits August 5, 2025 11:30

Update tests to use np.testing.assert_allclose
543d729

Remove lonlatdepth_dtype
8529d82

Remove MPI particle file writing code, clean up execution paths, and …
0a5970b
…assert .zarr file extension MPI is not yet supported in v4-dev anyway, and MPI will likely be implemented differently (#2074).

Remove particle.py
f0f8cbe

Add new Particle class implementation
7d04de4
This makes it distinct what a particle is on the "particle type" level (defining what attributes exist etc) and on the kernel level (i.e., just a link to the particle data. Naming is a WIP

Update references to fix tests
33a10bb

Remove old particle implementation
c0766a2

Create create_particle_data method on Particle class
e6fdd34

Add tests for particle.py
51bc4d5

Update create_particle_data to be a function
3c3ffa1

Update create_particle_data to pass arbitrary initial values
0fb9fdb

Update particle data naming to 'id' instead of 'trajectory'
e181004

Update initial value for dt Variable
986ce99

Update time dtype for particle
848d981

Create get_default_particle
59275f5
This allows users to get a default particle of differing spatial precision. This replaces the argument `lonlatdepth_dtype` that was previously used on the ParticleSet init.

Add dlon,dlat,ddepth variables to particle
b8d3315

Create particle data based on fieldset time interval
c67f6c4

Update create_particle_data arg
38709fb
Make it explicitly an initial dict rather than as kwargs

Add test_create_particle_data
31e2670

Update test_advection.py to use lon/lat instead of lon_nextloop...
b182a75
Not sure why this is necessary since I only refactored the way the particle data was created?

Remove commented code
0701a0b

github-project-automationBot added this to Parcels development Aug 5, 2025

github-project-automationBot moved this to Backlog in Parcels development Aug 5, 2025

Add example usage
69e3759

erikvansebille reviewed Aug 5, 2025
View reviewed changes

Comment threadparcels/particle.py
Comment threadparcels/particle.py Outdated
Comment threadparcels/particlefile.py Outdated
Comment threadparcels/particleset.py Outdated
Comment threadtests/v4/test_advection.py Outdated

erikvansebille reviewed Aug 6, 2025
View reviewed changes

VeckoTheGecko force-pushed the particle-particledata branch from 0afb92a to 6d8bbceCompare August 6, 2025 11:34

VeckoTheGecko mentioned this pull request Aug 6, 2025
More xgrid generated datasets #2135
Merged
3 tasks

VeckoTheGecko added 2 commits August 6, 2025 13:40

Review feedback
79bdb88

Add test_particleclass_add_variable_in_loop
ab67bd4

VeckoTheGecko force-pushed the particle-particledata branch from 6d8bbce to ab67bd4Compare August 6, 2025 11:40

erikvansebille approved these changes Aug 6, 2025
View reviewed changes

erikvansebille added a commit that referenced this pull request Aug 6, 2025
Removing unit test that is now part of#2134
568f133

VeckoTheGecko added 2 commits August 6, 2025 17:03

Update particle ID varname to be 'trajectory'
d0a2c6a
For internal consistency with the CF trajectory output spec. cc @erikvansebille

Merge branch 'v4-dev' into particle-particledata
07f7c94

VeckoTheGeckoand others added 3 commits August 6, 2025 18:06

Skip RK45 advection tests
4c56dde

Increasing RK45_tol to speed up tests
79e4420
For some reason RK45 was set so strict that it reduced dt to 1 second, severely slowing down the tests

Update LocalStore to DirectoryStore
c43c694
Local environment was desynced with CI - accidentally included zarr v3 code. Let's stick to v2 for now then upgrade later

VeckoTheGecko merged commit 9c6ce08 into v4-devAug 7, 2025
9 checks passed

VeckoTheGecko deleted the particle-particledata branch August 7, 2025 08:59

github-project-automationBot moved this from Backlog to Done in Parcels development Aug 7, 2025

Uh oh!

Conversation

VeckoTheGecko commented Aug 5, 2025• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

VeckoTheGecko commented Aug 6, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

VeckoTheGecko commented Aug 5, 2025• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

VeckoTheGecko commented Aug 6, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

VeckoTheGecko commented Aug 5, 2025• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

VeckoTheGecko commented Aug 6, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

VeckoTheGecko commented Aug 5, 2025• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

VeckoTheGecko commented Aug 5, 2025•
edited
Loading

VeckoTheGecko commented Aug 5, 2025•
edited
Loading

VeckoTheGecko commented Aug 5, 2025•
edited
Loading

VeckoTheGecko commented Aug 5, 2025•
edited
Loading

VeckoTheGecko commented Aug 5, 2025•
edited
Loading