Hi Ashley, Unfortunately Stephanie's suggestion won't work, since units like Mpccm, which depend on details of the dataset, aren't available in yt.units. In particular the conversion from comoving megaparsecs to physical units depends on redshift and assumed cosmology. I think what's happening here is that you are dealing with a non-cosmological simulation which does not have Mpccm defined. It looks like yt's StarFormationRate class that you're trying to use assumes that it's being passed an output from a cosmological simulation and is failing in a less than graceful fashion when you try pass a non-cosmological output. We should probably emphasize more in the documentation that StarFormationRate only accepts a cosmological output (or improve it so that it *can* work with non-cosmological outputs) and improve the error message you saw. If you'd like to file an issue about this documentation problem (at https://github.com/yt-project/yt/issues/new) that will help us to avoid losing track of this deficiency. Rather than using the StarFormationRate class, I'd suggest looking at this example in the cookbook section of the docs: http://yt-project.org/doc/cookbook/calculating_information.html#using- particle-filters-to-calculate-star-formation-rates To adapt that example for your data, you're going to need to figure out how to select the star particles in your simulation and then either adapt the particle filter definition in the example for your data, or if you are using a frontend like Gadget where the output format defines a star particle type, just use that instead of the particle filter. The rest of the cookbook recipe just uses numpy to bin the star particle ages according to the formation time of the particle, and then backs out the star formation rate history. Hope that helps, Nathan On Tue, Sep 12, 2017 at 3:23 PM Stephanie Tonnesen <stonnes@gmail.com> wrote:

I am not sure, but the first thing I would do would be to add

from yt.units import Mpccm

At the top and see what happens next.

Best, Stephanie

On Tue, Sep 12, 2017 at 11:36 AM Ashley Jarvis <ajarvis@star.sr.bham.ac.uk> wrote:

...

Hi all,

I’m attempting to adapt the example script for plotting star formation rates (found at http://yt-project.org/doc/analyzing/analysis_modules/star _analysis.html#star-formation-rate) to work with output data from a simulation run using Enzo. However, when I attempt to run the script, I get the following error:

———————————————————

File "PlotSFR.py", line 26, in <module> sfr = StarFormationRate(data, star_mass=mass_old, star_creation_time=ct_old, volume=sp.volume()) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/analysis _modules/star_analysis/sfr_spectrum.py", line 115, in __init__ self._ds.quan(1.0, 'Mpccm**3').units File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/yt_array.py", line 1355, in __new__ dtype=dtype, bypass_validation=bypass_validation) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/yt_array.py", line 430, in __new__ units = Unit(input_units, registry=registry) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 257, in __new__ unit_data = _get_unit_data_from_expr(unit_expr, registry.lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 572, in _get_unit_data_from_expr unit_data = _get_unit_data_from_expr(unit_expr.args[0], unit_symbol_lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 566, in _get_unit_data_from_expr return _lookup_unit_symbol(str(unit_expr), unit_symbol_lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 657, in _lookup_unit_symbol "symbols." % symbol_str) yt.units.unit_registry.UnitParseError: Could not find unit symbol 'Mpccm' in the provided symbols.

———————————————————

I am using version 3.3.5 of yt.

Does anyone know how I can resolve this? Any suggestions would be greatly appreciated.

Thanks, Ashley

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

-- -- Dr. Stephanie Tonnesen Alvin E. Nashman Postdoctoral Fellow Carnegie Observatories, Pasadena, CA stonnes@gmail.com _______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

On Wed, Sep 13, 2017 at 7:58 AM, Ashley Jarvis <ajarvis@star.sr.bham.ac.uk> wrote:

Thanks for your help. I have submitted a new issue about the documentation problem as you suggested.

I have used the other example to produce a plot, but I’m not sure if the particles are being filtered correctly. I am using Enzo simulation data. I think I want to select all particles that are stars, and then only keep those that have a creation time greater than 0; I am unsure how to do this. Is it possible to combine two conditions within the filter definition?

Yup, you can combine boolean arrays using boolean operators in Numpy. For your problem it would look like this: def formed_star(pfilter, data): # identify stars (particle_type 2) that have creation times after the beginning # of the simulation filter = (data["all", "creation_time"] > 0) & (data['all', 'particle_type'] == 2) return filter -Nathan

Many thanks, Ashley

On 12 Sep 2017, at 21:57, nathan12343@gmail.com wrote:

Hi Ashley,

Unfortunately Stephanie's suggestion won't work, since units like Mpccm, which depend on details of the dataset, aren't available in yt.units. In particular the conversion from comoving megaparsecs to physical units depends on redshift and assumed cosmology.

I think what's happening here is that you are dealing with a non-cosmological simulation which does not have Mpccm defined.

It looks like yt's StarFormationRate class that you're trying to use assumes that it's being passed an output from a cosmological simulation and is failing in a less than graceful fashion when you try pass a non-cosmological output. We should probably emphasize more in the documentation that StarFormationRate only accepts a cosmological output (or improve it so that it *can* work with non-cosmological outputs) and improve the error message you saw. If you'd like to file an issue about this documentation problem (at https://github.com/yt-project/yt/issues/new) that will help us to avoid losing track of this deficiency.

Rather than using the StarFormationRate class, I'd suggest looking at this example in the cookbook section of the docs:

http://yt-project.org/doc/cookbook/calculating_information. html#using-particle-filters-to-calculate-star-formation-rates

To adapt that example for your data, you're going to need to figure out how to select the star particles in your simulation and then either adapt the particle filter definition in the example for your data, or if you are using a frontend like Gadget where the output format defines a star particle type, just use that instead of the particle filter. The rest of the cookbook recipe just uses numpy to bin the star particle ages according to the formation time of the particle, and then backs out the star formation rate history.

Hope that helps,

Nathan

On Tue, Sep 12, 2017 at 3:23 PM Stephanie Tonnesen <stonnes@gmail.com> wrote:

...

I am not sure, but the first thing I would do would be to add

from yt.units import Mpccm

At the top and see what happens next.

Best, Stephanie

On Tue, Sep 12, 2017 at 11:36 AM Ashley Jarvis < ajarvis@star.sr.bham.ac.uk> wrote:

...

Hi all,

I’m attempting to adapt the example script for plotting star formation rates (found at http://yt-project.org/doc/a nalyzing/analysis_modules/star_analysis.html#star-formation-rate) to work with output data from a simulation run using Enzo. However, when I attempt to run the script, I get the following error:

———————————————————

File "PlotSFR.py", line 26, in <module> sfr = StarFormationRate(data, star_mass=mass_old, star_creation_time=ct_old, volume=sp.volume()) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/analysis _modules/star_analysis/sfr_spectrum.py", line 115, in __init__ self._ds.quan(1.0, 'Mpccm**3').units File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/yt_array.py", line 1355, in __new__ dtype=dtype, bypass_validation=bypass_validation) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/yt_array.py", line 430, in __new__ units = Unit(input_units, registry=registry) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 257, in __new__ unit_data = _get_unit_data_from_expr(unit_expr, registry.lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 572, in _get_unit_data_from_expr unit_data = _get_unit_data_from_expr(unit_expr.args[0], unit_symbol_lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 566, in _get_unit_data_from_expr return _lookup_unit_symbol(str(unit_expr), unit_symbol_lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/units/unit_object.py", line 657, in _lookup_unit_symbol "symbols." % symbol_str) yt.units.unit_registry.UnitParseError: Could not find unit symbol 'Mpccm' in the provided symbols.

———————————————————

I am using version 3.3.5 of yt.

Does anyone know how I can resolve this? Any suggestions would be greatly appreciated.

Thanks, Ashley

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

-- -- Dr. Stephanie Tonnesen Alvin E. Nashman Postdoctoral Fellow Carnegie Observatories, Pasadena, CA stonnes@gmail.com _______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

Hi Ashley, That depends on the star formation rate prescription you are using. The formula in the cookbook recipe calculates the star formation rate based on the masses of particles that formed over a small time period in the past. The star formation rate calculated by the recipe in a given time bin is the initial masses of star particles that formed in that bin divided by the width of the time bin. In the limit where the width of the time bin goes to zero, the instantaneous star formation rate and the star formation rate you measure through this method should agree. If your star particles participate in a feedback algorithm and lose mass or change mass for any other reason, that would impact the star formation rates one would measure with this algorithm. In this case, one should use the original mass of the star particles, which would need to be recorded somehow if the masses of newly formed star particles aren't known a priori. A good check to make sure you're getting the right answer is to compute the instantaneous star formation rate in a snapshot early in the simulation and compare it with the star formation rate you measured at the time of that snapshot based on the star particles present in the *final* snapshot. Assuming that you are forming lots of star particles in a given time bin and that your star formation algorithm is stochastic, these numbers should match closely although you will likely not get perfect agreement. For example, see Figure 1 of my 2015 paper ( http://adsabs.harvard.edu/abs/2015ApJ...814..131G). The code used to generate the figure is available in these two notebooks: http://nbviewer.jupyter.org/urls/bitbucket.org/ngoldbaum/galaxy_analysis/raw... http://nbviewer.jupyter.org/urls/bitbucket.org/ngoldbaum/galaxy_analysis/raw... Note that this paper did not include the effects of feedback, so there isn't any attempt to correct for the fact that the masses of star particles. However, for my 2016 paper I did account for that effect and just used the fixed 300 solar masses each new star particle in my simulation is created with. See Figure 3 of my 2016 paper ( http://adsabs.harvard.edu/abs/2016ApJ...827...28G) and this notebook: http://nbviewer.jupyter.org/urls/bitbucket.org/ngoldbaum/galaxy_analysis/raw... Hope that's helpful, Nathan On Thu, Oct 5, 2017 at 9:13 AM, <ajarvis@star.sr.bham.ac.uk> wrote:

Hi Nathan,

I managed to implement the filter definition that you provided, but I have come across another problem. When I try to generate the SFR plots for the simulation at different time steps, there is always a sharp decline in the rate around 25Myr before the output time of the data dump (see attached plots).

Is this happening because of something in the simulation itself, e.g. there is a delay between the region reaching a given density limit and the star particle actually being created (to represent the formation time of the stars), or is it being caused by something in the yt plotting process?

Many thanks, Ashley

https://i.imgur.com/mjCQvZy.png https://i.imgur.com/Cv5D1HK.png https://i.imgur.com/6ZA2xgq.png

...

On Wed, Sep 13, 2017 at 7:58 AM, Ashley Jarvis <ajarvis@star.sr.bham.ac.uk> wrote:

...

Thanks for your help. I have submitted a new issue about the documentation problem as you suggested.

I have used the other example to produce a plot, but I’m not sure if the particles are being filtered correctly. I am using Enzo simulation data. I think I want to select all particles that are stars, and then only keep those that have a creation time greater than 0; I am unsure how to do this. Is it possible to combine two conditions within the filter definition?

Yup, you can combine boolean arrays using boolean operators in Numpy. For your problem it would look like this:

def formed_star(pfilter, data): # identify stars (particle_type 2) that have creation times after the beginning # of the simulation filter = (data["all", "creation_time"] > 0) & (data['all', 'particle_type'] == 2) return filter

-Nathan

...

Many thanks, Ashley

On 12 Sep 2017, at 21:57, nathan12343@gmail.com wrote:

Hi Ashley,

Unfortunately Stephanie's suggestion won't work, since units like Mpccm, which depend on details of the dataset, aren't available in yt.units. In particular the conversion from comoving megaparsecs to physical units depends on redshift and assumed cosmology.

I think what's happening here is that you are dealing with a non-cosmological simulation which does not have Mpccm defined.

It looks like yt's StarFormationRate class that you're trying to use assumes that it's being passed an output from a cosmological simulation and is failing in a less than graceful fashion when you try pass a non-cosmological output. We should probably emphasize more in the documentation that StarFormationRate only accepts a cosmological output (or improve it so that it *can* work with non-cosmological outputs) and improve the error message you saw. If you'd like to file an issue about this documentation problem (at https://github.com/yt-project/yt/issues/new) that will help us to avoid losing track of this deficiency.

Rather than using the StarFormationRate class, I'd suggest looking at this example in the cookbook section of the docs:

http://yt-project.org/doc/cookbook/calculating_information. html#using-particle-filters-to-calculate-star-formation-rates

To adapt that example for your data, you're going to need to figure out how to select the star particles in your simulation and then either adapt the particle filter definition in the example for your data, or if you are using a frontend like Gadget where the output format defines a star particle type, just use that instead of the particle filter. The rest of the cookbook recipe just uses numpy to bin the star particle ages according to the formation time of the particle, and then backs out the star formation rate history.

Hope that helps,

Nathan

On Tue, Sep 12, 2017 at 3:23 PM Stephanie Tonnesen <stonnes@gmail.com> wrote:

...

I am not sure, but the first thing I would do would be to add

from yt.units import Mpccm

At the top and see what happens next.

Best, Stephanie

On Tue, Sep 12, 2017 at 11:36 AM Ashley Jarvis < ajarvis@star.sr.bham.ac.uk> wrote:

...

Hi all,

I’m attempting to adapt the example script for plotting star formation rates (found at http://yt-project.org/doc/a nalyzing/analysis_modules/star_analysis.html#star-formation-rate) to work with output data from a simulation run using Enzo. However, when I attempt to run the script, I get the following error:

———————————————————

File "PlotSFR.py", line 26, in <module> sfr = StarFormationRate(data, star_mass=mass_old, star_creation_time=ct_old, volume=sp.volume()) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/analysis _modules/star_analysis/sfr_spectrum.py", line 115, in __init__ self._ds.quan(1.0, 'Mpccm**3').units File "/usr/local/anaconda/lib/python2.7/site-packages/yt/ units/yt_array.py", line 1355, in __new__ dtype=dtype, bypass_validation=bypass_validation) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/ units/yt_array.py", line 430, in __new__ units = Unit(input_units, registry=registry) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/ units/unit_object.py", line 257, in __new__ unit_data = _get_unit_data_from_expr(unit_expr, registry.lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/ units/unit_object.py", line 572, in _get_unit_data_from_expr unit_data = _get_unit_data_from_expr(unit_expr.args[0], unit_symbol_lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/ units/unit_object.py", line 566, in _get_unit_data_from_expr return _lookup_unit_symbol(str(unit_expr), unit_symbol_lut) File "/usr/local/anaconda/lib/python2.7/site-packages/yt/ units/unit_object.py", line 657, in _lookup_unit_symbol "symbols." % symbol_str) yt.units.unit_registry.UnitParseError: Could not find unit symbol 'Mpccm' in the provided symbols.

———————————————————

I am using version 3.3.5 of yt.

Does anyone know how I can resolve this? Any suggestions would be greatly appreciated.

Thanks, Ashley

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

-- -- Dr. Stephanie Tonnesen Alvin E. Nashman Postdoctoral Fellow Carnegie Observatories, Pasadena, CA stonnes@gmail.com _______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

_______________________________________________ yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org