SimSteps.py

import os
import galsim
import traceback
import gc
import psutil
import numpy as np
from astropy.io import fits
from datetime import datetime
from numpy.random import Generator, PCG64

from ObservationSim._util import get_shear_field
from ObservationSim.Straylight import calculateSkyMap_split_g
from ObservationSim.Config.Header import generatePrimaryHeader, generateExtensionHeader
from ObservationSim.PSF import PSFGauss, FieldDistortion, PSFInterp, PSFInterpSLS
from ObservationSim.Instrument.Chip import ChipUtils as chip_utils
from ObservationSim.Instrument.Chip import Effects
from ObservationSim.Instrument.Chip.libCTI.CTI_modeling import CTI_sim

class SimSteps:
    def __init__(self, overall_config, chip_output, all_filters):
        self.overall_config = overall_config
        self.chip_output = chip_output
        self.all_filters = all_filters
    
    def prepare_headers(self, chip, pointing):
        
        self.h_prim = generatePrimaryHeader(
            xlen=chip.npix_x, 
            ylen=chip.npix_y, 
            pointNum = str(pointing.id),
            ra=pointing.ra, 
            dec=pointing.dec, 
            pixel_scale=chip.pix_scale,
            time_pt = pointing.timestamp,
            exptime=pointing.exp_time,
            im_type=pointing.pointing_type,
            sat_pos=[pointing.sat_x, pointing.sat_y, pointing.sat_z],
            sat_vel=[pointing.sat_vx, pointing.sat_vy, pointing.sat_vz],
            project_cycle=self.overall_config["project_cycle"],
            run_counter=self.overall_config["run_counter"],
            chip_name=str(chip.chipID).rjust(2, '0'))
        self.h_ext = generateExtensionHeader(
            chip=chip,
            xlen=chip.npix_x, 
            ylen=chip.npix_y, 
            ra=pointing.ra, 
            dec=pointing.dec, 
            pa=pointing.img_pa.deg, 
            gain=chip.gain, 
            readout=chip.read_noise, 
            dark=chip.dark_noise, 
            saturation=90000, 
            pixel_scale=chip.pix_scale, 
            pixel_size=chip.pix_size,
            xcen=chip.x_cen,
            ycen=chip.y_cen,
            extName=pointing.pointing_type,
            timestamp = pointing.timestamp,
            exptime = pointing.exp_time,
            readoutTime = chip.readout_time)
        return self.h_prim, self.h_ext

    def add_sky_background(self, chip, filt, tel, pointing, catalog, obs_param):
        flat_normal = np.ones_like(chip.img.array)
        if obs_param["flat_fielding"] == True:
            flat_normal = flat_normal * chip.flat_img.array / np.mean(chip.flat_img.array)
        if obs_param["shutter_effect"] == True:
            flat_normal = flat_normal * chip.shutter_img
            flat_normal = np.array(flat_normal, dtype='float32')
        
        if obs_param["enable_straylight_model"]:
            # Filter.sky_background, Filter.zodical_spec will be updated
            filt.setFilterStrayLightPixel(
                jtime = pointing.jdt,
                sat_pos = np.array([pointing.sat_x, pointing.sat_y, pointing.sat_z]),
                pointing_radec = np.array([pointing.ra,pointing.dec]),
                sun_pos = np.array([pointing.sun_x, pointing.sun_y, pointing.sun_z]))
            self.chip_output.Log_info("================================================")
            self.chip_output.Log_info("sky background + stray light pixel flux value: %.5f"%(filt.sky_background))
        
        if chip.survey_type == "photometric":
            sky_map = filt.getSkyNoise(exptime = obs_param["exptime"])
            sky_map = sky_map * np.ones_like(chip.img.array) * flat_normal
            sky_map = galsim.Image(array=sky_map)
        else:
            # chip.loadSLSFLATCUBE(flat_fn='flat_cube.fits')
            sky_map = calculateSkyMap_split_g(
                    skyMap=flat_normal,
                    blueLimit=filt.blue_limit,
                    redLimit=filt.red_limit,
                    conf=chip.sls_conf,
                    pixelSize=chip.pix_scale,
                    isAlongY=0,
                    flat_cube=chip.flat_cube, 
                    zoldial_spec = filt.zodical_spec)
            sky_map = sky_map + filt.sky_background
        
        # sky_map = sky_map * tel.pupil_area * obs_param["exptime"]
        chip.img += sky_map
        return chip, filt, tel, pointing

    def add_objects(self, chip, filt, tel, pointing, catalog, obs_param):

        # Prepare output file(s) for this chip
        self.chip_output.create_output_file()

        # Prepare the PSF model
        if self.overall_config["psf_setting"]["psf_model"] == "Gauss":
            psf_model = PSFGauss(chip=chip, psfRa=self.overall_config["psf_setting"]["psf_rcont"])
        elif self.overall_config["psf_setting"]["psf_model"] == "Interp":
            if chip.survey_type == "spectroscopic":
                psf_model = PSFInterpSLS(chip, filt, PSF_data_prefix=self.overall_config["psf_setting"]["psf_sls_dir"])
            else:
                psf_model = PSFInterp(chip=chip, npsf=chip.n_psf_samples, PSF_data_file=self.overall_config["psf_setting"]["psf_pho_dir"])
        else:
            self.chip_output.Log_error("unrecognized PSF model type!!", flush=True)
        
        # Apply field distortion model
        if obs_param["field_dist"] == True:
            fd_model = FieldDistortion(chip=chip, img_rot=pointing.img_pa.deg)
        else:
            fd_model = None

        # Update limiting magnitudes for all filters based on the exposure time
        # Get the filter which will be used for magnitude cut
        for ifilt in range(len(self.all_filters)):
            temp_filter = self.all_filters[ifilt]
            temp_filter.update_limit_saturation_mags(exptime=pointing.get_full_depth_exptime(temp_filter.filter_type), chip=chip)
            if temp_filter.filter_type.lower() == self.overall_config["obs_setting"]["cut_in_band"].lower():
                cut_filter = temp_filter

        # Read in shear values from configuration file if the constant shear type is used
        if self.overall_config["shear_setting"]["shear_type"] == "constant":
            g1_field, g2_field, _ = get_shear_field(config=self.overall_config)
            self.chip_output.Log_info("Use constant shear: g1=%.5f, g2=%.5f"%(g1_field, g2_field))

        # Get chip WCS
        if not hasattr(self, 'h_ext'):
            _, _ = self.prepare_headers(chip=chip, pointing=pointing)
        chip_wcs = galsim.FitsWCS(header = self.h_ext)
        
        # Loop over objects
        nobj = len(catalog.objs)
        missed_obj = 0
        bright_obj = 0
        dim_obj = 0
        for j in range(nobj):
            # # [DEBUG] [TODO]
            # if j >= 10:
            #     break
            obj = catalog.objs[j]

            # load and convert SED; also caculate object's magnitude in all CSST bands
            try:
                sed_data = catalog.load_sed(obj)
                norm_filt = catalog.load_norm_filt(obj)
                obj.sed, obj.param["mag_%s"%filt.filter_type.lower()], obj.param["flux_%s"%filt.filter_type.lower()] = catalog.convert_sed(
                    mag=obj.param["mag_use_normal"],
                    sed=sed_data,
                    target_filt=filt, 
                    norm_filt=norm_filt,
                )
                _, obj.param["mag_%s"%cut_filter.filter_type.lower()], obj.param["flux_%s"%cut_filter.filter_type.lower()] = catalog.convert_sed(
                    mag=obj.param["mag_use_normal"],
                    sed=sed_data,
                    target_filt=cut_filter, 
                    norm_filt=norm_filt,
                )
            except Exception as e:
                traceback.print_exc()
                self.chip_output.Log_error(e)
                continue

            # [TODO] Testing
            # self.chip_output.Log_info("mag_%s = %.3f"%(filt.filter_type.lower(), obj.param["mag_%s"%filt.filter_type.lower()]))

            # Exclude very bright/dim objects (for now)
            if cut_filter.is_too_bright(
                mag=obj.param["mag_%s"%self.overall_config["obs_setting"]["cut_in_band"].lower()],
                margin=self.overall_config["obs_setting"]["mag_sat_margin"]):
                self.chip_output.Log_info("obj %s too birght!! mag_%s = %.3f"%(obj.id, cut_filter.filter_type, obj.param["mag_%s"%self.overall_config["obs_setting"]["cut_in_band"].lower()]))
                bright_obj += 1
                obj.unload_SED()
                continue
            if filt.is_too_dim(
                mag=obj.getMagFilter(filt),
                margin=self.overall_config["obs_setting"]["mag_lim_margin"]):
                self.chip_output.Log_info("obj %s too dim!! mag_%s = %.3f"%(obj.id, filt.filter_type, obj.getMagFilter(filt)))
                dim_obj += 1
                obj.unload_SED()
                continue

            # Get corresponding shear values
            if self.overall_config["shear_setting"]["shear_type"] == "constant":
                if obj.type == 'star':
                    obj.g1, obj.g2 = 0., 0.
                else:
                    # Figure out shear fields from overall configuration shear setting
                    obj.g1, obj.g2 = g1_field, g2_field
            elif self.overall_config["shear_setting"]["shear_type"] == "catalog":
                pass
            else:
                self.chip_output.Log_error("Unknown shear input")
                raise ValueError("Unknown shear input")

            # Get position of object on the focal plane
            pos_img, _, _, _, fd_shear = obj.getPosImg_Offset_WCS(img=chip.img, fdmodel=fd_model, chip=chip, verbose=False, chip_wcs=chip_wcs, img_header=self.h_ext)

            # [TODO] For now, only consider objects which their centers (after field distortion) are projected within the focal plane
            # Otherwise they will be considered missed objects
            # if pos_img.x == -1 or pos_img.y == -1 or (not chip.isContainObj(x_image=pos_img.x, y_image=pos_img.y, margin=0.)):
            if pos_img.x == -1 or pos_img.y == -1:
                self.chip_output.Log_info('obj_ra = %.6f, obj_dec = %.6f, obj_ra_orig = %.6f, obj_dec_orig = %.6f'%(obj.ra, obj.dec, obj.ra_orig, obj.dec_orig))
                self.chip_output.Log_error("Objected missed: %s"%(obj.id))
                missed_obj += 1
                obj.unload_SED()
                continue

            # Draw object & update output catalog
            try:
                if self.overall_config["run_option"]["out_cat_only"]:
                    isUpdated = True
                    obj.real_pos = obj.getRealPos(chip.img, global_x=obj.posImg.x, global_y=obj.posImg.y, img_real_wcs=obj.chip_wcs)
                    pos_shear = 0.
                elif chip.survey_type == "photometric" and not self.overall_config["run_option"]["out_cat_only"]:
                    isUpdated, pos_shear = obj.drawObj_multiband(
                        tel=tel,
                        pos_img=pos_img, 
                        psf_model=psf_model, 
                        bandpass_list=filt.bandpass_sub_list, 
                        filt=filt, 
                        chip=chip, 
                        g1=obj.g1, 
                        g2=obj.g2, 
                        exptime=pointing.exp_time,
                        fd_shear=fd_shear)

                elif chip.survey_type == "spectroscopic" and not self.overall_config["run_option"]["out_cat_only"]:
                    isUpdated, pos_shear = obj.drawObj_slitless(
                        tel=tel, 
                        pos_img=pos_img, 
                        psf_model=psf_model, 
                        bandpass_list=filt.bandpass_sub_list, 
                        filt=filt, 
                        chip=chip, 
                        g1=obj.g1, 
                        g2=obj.g2, 
                        exptime=pointing.exp_time,
                        normFilter=norm_filt,
                        fd_shear=fd_shear)

                if isUpdated == 1:
                    # TODO: add up stats
                    self.chip_output.cat_add_obj(obj, pos_img, pos_shear)
                    pass
                elif isUpdated == 0:
                    missed_obj += 1
                    self.chip_output.Log_error("Objected missed: %s"%(obj.id))
                else:
                    self.chip_output.Log_error("Draw error, object omitted: %s"%(obj.id))
                    continue
            except Exception as e:
                traceback.print_exc()
                self.chip_output.Log_error(e)

            # Unload SED:
            obj.unload_SED()
            del obj
            gc.collect()
        del psf_model
        gc.collect()

        self.chip_output.Log_info("Running checkpoint #1 (Object rendering finished): pointing-%d chip-%d pid-%d memory-%6.2fGB"%(pointing.id, chip.chipID, os.getpid(), (psutil.Process(os.getpid()).memory_info().rss / 1024 / 1024 / 1024) ))

        self.chip_output.Log_info("# objects that are too bright %d out of %d"%(bright_obj, nobj))
        self.chip_output.Log_info("# objects that are too dim %d out of %d"%(dim_obj, nobj))
        self.chip_output.Log_info("# objects that are missed %d out of %d"%(missed_obj, nobj))

        # Apply flat fielding (with shutter effects)
        flat_normal = np.ones_like(chip.img.array)
        if obs_param["flat_fielding"] == True:
            flat_normal = flat_normal * chip.flat_img.array / np.mean(chip.flat_img.array)
        if obs_param["shutter_effect"] == True:
            flat_normal = flat_normal * chip.shutter_img
            flat_normal = np.array(flat_normal, dtype='float32')
        chip.img *= flat_normal
        del flat_normal

        return chip, filt, tel, pointing
    
    def add_cosmic_rays(self, chip, filt, tel, pointing, catalog, obs_param):
        self.chip_output.Log_info("  Adding Cosmic-Ray")
        chip.img, crmap_gsimg, cr_event_num = chip_utils.add_cosmic_rays(
            img=chip.img, 
            chip=chip, 
            exptime=pointing.exp_time, 
            seed=self.overall_config["random_seeds"]["seed_CR"]+pointing.id*30+chip.chipID)
        # [TODO] output cosmic ray image
        return chip, filt, tel, pointing
    
    def apply_PRNU(self, chip, filt, tel, pointing, catalog, obs_param):
        chip.img *= chip.prnu_img
        return chip, filt, tel, pointing
    
    def add_poisson_and_dark(self, chip, filt, tel, pointing, catalog, obs_param):
        # Add dark current & Poisson noise
        InputDark = False
        if obs_param["add_dark"] == True:
            if InputDark:
                chip.img = chip_utils.add_inputdark(img=chip.img, chip=chip, exptime=pointing.exp_time)
            else:
                chip.img, _ = chip_utils.add_poisson(img=chip.img, chip=chip, exptime=pointing.exp_time, poisson_noise=chip.poisson_noise)
        else:
            chip.img, _ = chip_utils.add_poisson(img=chip.img, chip=self, exptime=pointing.exp_time, poisson_noise=chip.poisson_noise, dark_noise=0.)
        return chip, filt, tel, pointing
    
    def add_brighter_fatter(self, chip, filt, tel, pointing, catalog, obs_param):
        chip.img = chip_utils.add_brighter_fatter(img=chip.img)
        return chip, filt, tel, pointing
    
    def add_detector_defects(self, chip, filt, tel, pointing, catalog, obs_param):
        # Add Hot Pixels or/and Dead Pixels
        rgbadpix = Generator(PCG64(int(self.overall_config["random_seeds"]["seed_defective"]+chip.chipID)))
        badfraction = 5E-5*(rgbadpix.random()*0.5+0.7)
        chip.img = Effects.DefectivePixels(
            chip.img, 
            IfHotPix=obs_param["hot_pixels"], 
            IfDeadPix=obs_param["dead_pixels"],
            fraction=badfraction, 
            seed=self.overall_config["random_seeds"]["seed_defective"]+chip.chipID, biaslevel=0)
        # Apply Bad columns 
        if obs_param["bad_columns"] == True:
            chip.img = Effects.BadColumns(chip.img, seed=self.overall_config["random_seeds"]["seed_badcolumns"], chipid=chip.chipID)
        return chip, filt, tel, pointing
    
    def add_nonlinearity(self, chip, filt, tel, pointing, catalog, obs_param):
        self.chip_output.Log_info("  Applying Non-Linearity on the chip image")
        chip.img = Effects.NonLinearity(GSImage=chip.img, beta1=5.e-7, beta2=0)
        return chip, filt, tel, pointing
    
    def add_blooming(self, chip, filt, tel, pointing, catalog, obs_param):
        self.chip_output.Log_info("  Applying CCD Saturation & Blooming")
        chip.img = Effects.SaturBloom(GSImage=chip.img, nsect_x=1, nsect_y=1, fullwell=int(chip.full_well))
        return chip, filt, tel, pointing

    def apply_CTE(self, chip, filt, tel, pointing, catalog, obs_param):
        self.chip_output.Log_info("  Apply CTE Effect")
        ### 2*8 -> 1*16 img-layout
        img = chip_utils.formatOutput(GSImage=chip.img)
        chip.nsecy = 1
        chip.nsecx = 16

        img_arr = img.array
        ny, nx = img_arr.shape
        dx = int(nx/chip.nsecx)
        dy = int(ny/chip.nsecy)
        newimg = galsim.Image(nx, int(ny+chip.overscan_y), init_value=0)
        for ichannel in range(16):
            print('\n***add CTI effects: pointing-{:} chip-{:} channel-{:}***'.format(pointing.id, chip.chipID, ichannel+1))
            noverscan, nsp, nmax = chip.overscan_y, 3, 10
            beta, w, c = 0.478, 84700, 0
            t = np.array([0.74, 7.7, 37],dtype=np.float32)
            rho_trap = np.array([0.6, 1.6, 1.4],dtype=np.float32)
            trap_seeds = np.array([0, 1000, 10000],dtype=np.int32) + ichannel + chip.chipID*16
            release_seed = 50 + ichannel + pointing.id*30  + chip.chipID*16
            newimg.array[:, 0+ichannel*dx:dx+ichannel*dx] = CTI_sim(img_arr[:, 0+ichannel*dx:dx+ichannel*dx],dx,dy,noverscan,nsp,nmax,beta,w,c,t,rho_trap,trap_seeds,release_seed)
        newimg.wcs = img.wcs
        del img
        img = newimg

        ### 1*16 -> 2*8 img-layout
        chip.img = chip_utils.formatRevert(GSImage=img)
        chip.nsecy = 2
        chip.nsecx = 8
        
        # [TODO] make overscan_y == 0
        chip.overscan_y = 0
        return chip, filt, tel, pointing
    
    def add_prescan_overscan(self, chip, filt, tel, pointing, catalog, obs_param):
        self.chip_output.Log_info("Apply pre/over-scan")
        chip.img = chip_utils.AddPreScan(GSImage=chip.img,
                                         pre1=chip.prescan_x,
                                         pre2=chip.prescan_y,
                                         over1=chip.overscan_x,
                                         over2=chip.overscan_y)
        return chip, filt, tel, pointing
    
    def add_bias(self, chip, filt, tel, pointing, catalog, obs_param):
        self.chip_output.Log_info("  Adding Bias level and 16-channel non-uniformity")
        if obs_param["bias_16channel"] == True:
            chip.img = Effects.AddBiasNonUniform16(chip.img, 
                    bias_level=float(chip.bias_level), 
                    nsecy = chip.nsecy, nsecx=chip.nsecx, 
                    seed=self.overall_config["random_seeds"]["seed_biasNonUniform"]+chip.chipID)
        elif obs_param["bias_16channel"] == False:
            chip.img += self.bias_level
        return chip, filt, tel, pointing

    def add_readout_noise(self, chip, filt, tel, pointing, catalog, obs_param):
        seed = int(self.overall_config["random_seeds"]["seed_readout"]) + pointing.id*30 + chip.chipID
        rng_readout = galsim.BaseDeviate(seed)
        readout_noise = galsim.GaussianNoise(rng=rng_readout, sigma=chip.read_noise)
        chip.img.addNoise(readout_noise)
        return chip, filt, tel, pointing
    
    def apply_gain(self, chip, filt, tel, pointing, catalog, obs_param):
        self.chip_output.Log_info("  Applying Gain")
        if obs_param["gain_16channel"] == True:
            chip.img, chip.gain_channel = Effects.ApplyGainNonUniform16(
                chip.img, gain=chip.gain, 
                nsecy = chip.nsecy, nsecx=chip.nsecx, 
                seed=self.overall_config["random_seeds"]["seed_gainNonUniform"]+chip.chipID)
        elif obs_param["gain_16channel"] == False:
            chip.img /= chip.gain
        return chip, filt, tel, pointing
    
    def quantization_and_output(self, chip, filt, tel, pointing, catalog, obs_param):
        chip.img.array[chip.img.array > 65535] = 65535
        chip.img.replaceNegative(replace_value=0)
        chip.img.quantize()
        chip.img = galsim.Image(chip.img.array, dtype=np.uint16)
        hdu1 = fits.PrimaryHDU(header=self.h_prim)
        hdu1.add_checksum()
        hdu1.header.comments['CHECKSUM'] = 'HDU checksum'
        hdu1.header.comments['DATASUM'] = 'data unit checksum'
        hdu2 = fits.ImageHDU(chip.img.array, header=self.h_ext)
        hdu2.add_checksum()
        hdu2.header.comments['XTENSION'] = 'extension type'
        hdu2.header.comments['CHECKSUM'] = 'HDU checksum'
        hdu2.header.comments['DATASUM'] = 'data unit checksum'
        hdu2.header.comments["XTENSION"] = "image extension"
        hdu1 = fits.HDUList([hdu1, hdu2])
        fname = os.path.join(self.chip_output.subdir, self.h_prim['FILENAME'] + '.fits')
        hdu1.writeto(fname, output_verify='ignore', overwrite=True)
        return chip, filt, tel, pointing

SIM_STEP_TYPES = {
    "scie_obs": "add_objects",
    "sky_background": "add_sky_background",
    "cosmic_rays": "add_cosmic_rays",
    "PRNU_effect": "apply_PRNU",
    "poisson_and_dark": "add_poisson_and_dark",
    "bright_fatter": "add_brighter_fatter",
    "detector_defects": "add_detector_defects",
    "nonlinearity": "add_nonlinearity",
    "blooming": "add_blooming",
    "CTE_effect": "apply_CTE",
    "prescan_overscan": "add_prescan_overscan",
    "bias": "add_bias",
    "readout_noise": "add_readout_noise",
    "gain": "apply_gain",
    "quantization_and_output": "quantization_and_output"
}