## Technical Details of the GalFish Calculator

**Inputs:**The inputs required by GalFish are the survey plan (a list containing the wavelength and depth of the proposed observations) and the properties of the target galaxies. GalFish combines them as described in Acquaviva et al, ApJ 749 (2012) to compute the Fisher matrix, the covariance matrix, and the forecasted uncertainties on the parameters over which the user chooses to marginalize.

**Target Galaxies:**The uncertainties computed by GalFish depend on the physical properties of the galaxy being observed. A choice of target galaxies is available, differing in redshift (z = 1, 2, or 3), age (100 Myr, 600 Myr, or 2 Gyr), dust content parametrized by the excess color E(B-V) (E(B-V) = 0, 0.3 or 0.6 mag), star formation history (constant or starburst), and mass (which is a free parameter chosen by the user).

**Simulated Data and Photometric Errors:**A list of 33 filters from the most common instruments from ground and space, ranging from rest-frame UV to rest-frame IR, is available. The user can choose to use a certain band by inserting the depth of the planned observation (in magnitudes) in each box. The limiting magnitude is used to compute the photometric error associated to the measurement in each filter. If a box is left empty, the default assumption is that no information is available in that band. The user can also specify an additional error, added in quadrature to the photometric one, and expressed in terms of a fraction of the flux for all bands. This allows one to account for e.g., calibration errors, or to investigate the effect of an unknown source of systematic uncertainty. This error can be specified as a decimal number (0.1 = 10%) in the box labeled `systematic error'.

**Stellar Population Synthesis Models:**GalFish uses the latest models from Bruzual and Charlot (2011), assumes a Salpeter IMF, uses the prescription of Calzetti (2000) to compute the dust attenuation from the excess color E(B-V), and assumes that there are no photons emitted more energetic than 912 A (rest-frame). While the measurement of SED fitting parameters may depend significantly on the assumptions made in SPS modeling, the uncertainties on them, which are the output of GalFish, are expected to be much less model-dependent.

**Priors:**If the data are not able to constrain the values of SED fitting parameters with high precision, prior available information on the probability distribution function becomes important. In particular, the requirement that galaxies cannot be older than the age of the Universe at their redshift needs to be implemented. This is usually achieved through a top-hat prior on the age of the galaxy; however, as demonstrated in Acquaviva et al, ApJ 749 (2012), this prior can be effectively approximated by a Gaussian prior in ln(age) of width &Delta ln(age) ~ 3.5 or 4.0. Such prior can be implemented by populating the corresponding box with the chosen width; if the box is left empty, no prior is included.

**Units:**GalFish computes and reports (if requested) the Fisher and covariance matrices in units of ln(age) and ln(mass). The square root of the diagonal elements of the covariance matrix in these units corresponds to the relative error on the corresponding parameters, (σ Age)/Age and (σ Mass)/Mass. The uncertainty on the parameter E(B-V) is an absolute error and is expressed in magnitudes.

**Results:**The results reported by default are the uncertainties on each parameter in the units specified above. By ticking the corresponding boxes, the user can also request to output the Fisher matrix and its inverse, the covariance matrix. The latter can be used to evaluate the correlation between parameters and to plot error ellipses. GalFish also prints to screen the input values associated to each run, so that each configuration can be easily saved.

**Saving a Calculation**: Many input parameters are used in a GalFish calculation, and it can be tedious to type them in, especially if you'd just like to try a slight variation on a previous calculation. The GalFish input forms use the HTML 'GET' method to access the input variables you provide, which means that all your input values are appended to URL (the web address) upon submission. To save a submission, you can copy the URL and save it in a regular text file. To re-run the same parameters later, simply paste the URL back into the address bar of your browser.

**Previous Submissions:**At the bottom of the Compute page, GalFish keeps track of your previous 5 submissions, with the most recent (ie. current) submission numbered '1', and the 5th most recent numbered '5'. Each item in the list is a complete listing of the settings used in the submission, and is a link to re-submit the inputs. To save the inputs for a previous submission, right-click on the link and select 'Copy link address' (or similar menu option for your browser). The web address contains all the input values and can be saved in a regular text file for later use.