dtnguyen2

@Doug0Burke, You are a genius! Because your suggestion ```Python def lmdif(loads-of-arguments, transform=None): ... if transform is None: transform = NoTransformation() ``` works.

Using the following code: ```Python def check_transformation(arg): if arg is False: return NoTransformation() elif arg is True or isinstance(arg, Transformation): return Transformation() elif inspect.isclass(arg) and issubclass(arg, Transformation): return arg else:...

cp the following files: ``` (testing-PR) [dtn@devel12 sherpa]$ ls -l /pool7/dtn/transform total 100K -rw-rw-r--. 1 dtn dtn 45K Apr 4 15:08 optfcts.py -rw-rw-r--. 1 dtn dtn 29K Apr 4 15:08...

The most recent push accepts False, True or a user's inherit transformation from the class ```Transformation```. I can get the tests to pass on my machine now, let's see if...

BTW, the file _sherpa/optmethods/optfcts.py_ is getting too big, so I moved the relevant transformation classes to a separate file so users who want to write their own transformation can see...

I slightly modified @DougBurke example: ```Python from sherpa.astro.ui import * load_pha("/home/dburke/sherpa/sherpa-test-data/sherpatest/3c273.pi") set_source(xsphabs.gal * xspowerlaw.pl) print(get_source()) set_method_opt("transformation", True) fit() ``` That is I changed the command: ```Python set_method_opt("transformation", True) ``` to...

Hang on, I think I have a simple fix that will return a reasonable fit without having the user tighten the ranges. A more sophisticated fix would entail more code...

The following excerpt: > The user must therefore be aware of the fact that, for example, if he puts limits of (0, 10**10) on a parameter, then the values 0.0...

1. correct about soft limits are more restrictive than the hard limits 2. correct 3. I double checked the ```covar``` and ```conf`` code to verify that they use the hard...

So we are going to keep the chi2xspecvar but make the following two changes right? 1) "drop", ie exclude the accumulation of the stat if yraw[ii] == 0 2) remove...