Update:
I will provide 200 spectra. 100 of them are uniformly distributed across stellar parameters, Teff = 6300-7300K, log(g/(cm/s^2)) = 3.5-4.5, [Fe/H] = -1 - 0.4. The remaining 100 spectra are also uniformly distributed, but the range is wider: Teff = 6000-7500K, log(g/(cm/s^2)) = 1-5, [Fe/H] = -5 - 1. In both samples, [alpha/Fe] depends on [Fe/H].

Then, the spectra are doppler shifted with a radial velocity randomly selected from -200 to 200 km/s. Galactic extinction is also applied to the spectra with `F99ExtinctionCurve` at a fixed ra/dec=0deg/0deg.

200 model spectra of F stars were made in the pfsSimObject format. The relevant catalog contains catId, objId, tract/patch, magnitude, stellar parameters, etc. I use catId = 9 but I can change it if you have any recommendations. type is set to be FLUXSTD. I did not calculate J band magnitudes.
 
model spectra and catalog:
https://hscdata.mtk.nao.ac.jp/hsc_bin_dist/temp/deleted_on_20221101/model_fstars_2022_08_17.tar.gz
 
The objId = 0 – 99 are stars of the sample with wider parameter ranges, which includes stars scarcely observed (the latter in the above description). On the other hand, the objId = 100-199 are of the sample composed of stars with likely stellar parameters (the former). For the integration test and weekly, it would be better to select stars of objId = 100-199.
 
Paul, could you have a look at these spectra and catalog? price
 

The catalog file includes columns m_g, m_r, m_i, m_z and m_Y. What photometric system is this? I would have guessed HSC, except you use Y instead of y. We need to be able to map these to the bandpasses in obs_pfs. sogo.mineo, is there a lookup-table somewhere that matches the filter name in the pfsConfig to the bandpass file?

I think that we need to add noise to these photometric measurements. Do you have a prescription that you'd like to use, or should I invent something?

Everything has r = 18 mag. Perhaps it would be better to have a variety of brightnesses, so that you can investigate performance as a function of brightness? (I could artificially dim spectra in the simulator if you'd like; but I need from you a prescription for selecting the magnitudes.)

I've verified that the simulator can read the spectra, and that the flux scaling is reasonable.

Is it OK to randomly select stars from your catalog for the simulated exposures? We currently have three modes for the simulator:

• The "integration test" has fifteen fibers, three of which are flux standards. The purpose of this dataset is to allow a relatively quick check that changes have not broken the pipeline. Due to the small number of flux standards, I would probably lean towards using brighter sources so that operations downstream from flux calibration are not affected by lower signal-to-noise. There are two designs (the second is just a shuffled version of the first), so we could potentially use up to 6 different spectra.
• The "weekly" has a single spectrograph with full fiber sampling, with 60 flux standards. The purpose of this dataset is to check that the pipeline is performing well on data that looks something like we'll get from the real instrument. There are two designs (the second is just a shuffled version of the first), so we could potentially use up to 120 different spectra.
• The "science" set has a single spectrograph with full fiber sampling, with 59 flux standards. The purpose of this dataset is to produce science-like outputs for the 1D team. There are two designs (one stars, one extragalactic) so we could use a total of 118 different spectra.

is there a lookup-table somewhere that matches the filter name in the pfsConfig to the bandpass file?

Here is it: https://github.com/Subaru-PFS/drp_stella/blob/13d4aef8fd33b22f49bd0d959b6334e68f762733/python/pfs/drp/stella/fitReference.py#L111

I am not sure whether Takuji Yamashita uses this mapping or not. In particular, this mapping maps "i" to "HSC-I" filter as opposed to "HSC-I2". If he were to use it, m_i would not be HSC-I2 but HSC-I.

I am sorry for this late reply...
The photometric system is HSC. I used g, r2, i2, z, and y of HSC in FilterCurve that Mineo-san mentioned above.

For magnitude and error, I am sorry, I misunderstood. I meant 18 mag is just a default value. I was thinking of changing magnitudes by using --fluxStdMag option or by editing the catalog on my side for each run. I was also thinking of adding magnitude errors by editing pfsConfig. I will send an updated catalog which includes magnitude errors and a variety of magnitudes.

For the selection of spectra, I will select 3 spectra for integration test, 60 for weekly, and 59 for science.

This is the updated catalog.
https://hscdata.mtk.nao.ac.jp/pfs_tempfiles/deleted_on_20221231/model_fstars_2022_10_06.tar.gz

I added columns of magnitude errors (e_m_x), fluxes, and flux errors. I use uniform flux errors for all stars, which corresponds to S/N = 100 at 20 mag (it is slightly worse than one of HSC-SSP). The magnitudes are distributed from g = 14 to 20 mag. This magnitude range is close to one of our flux standard F stars.

I recommend objId = 103, 170, and 182 for the integration test. Because the parameters in the catalog do not depend on objId, we can use 60 spectra with objId = 100 - 159 for weekly. They are the main sample. Similarly, for the science mode, we can use 59 spectra with objId = 141 - 199.

As part of this, we will need to check that we're using the correct filter names, following PIPE2D-1110.

The filters I used in the above model spectra are g/r2/i2/z/y of HSC. But I have updated some minor points. Could you use the following new spectra and catalog?
 
new: 
https://hscdata.mtk.nao.ac.jp/pfs_tempfiles/deleted_on_20230331/model_fstars_2022_12_22.tar.gz
 

• Filters were changed from HSC to PS1. The filter curves, from g_ps1 to y_ps1, in fitReference.py were used.
• A flux error was changed to 350nJy, which corresponds to a sensitivity of PS1 g-band in 3pi survey.

These spectra have been incorporated into a new integration test and weekly data set (SIM2D-151).