In 2004, Taylor et al.[1] observed the cyclohexadienyl radical reaction with oxygen in non-polar solvents using a liquid-phase spectrometer. The spectrometer measured the amount of UV light exiting a solution of 1,4-cyclohexadiene and a peroxide that was exposed to the pulse of an excimer laser. This pulse initiated a series of reactions which produced cyclohexadienyl radicals. Cyclohexadienyl radicals react with oxygen to ultimately form benzene and hydroperoxyl radical. This technique, known as laser-flash photolysis, both produced and measured cyclohexadienyl radical concentration over a period of 5 millionths of a second or 5 microseconds.

The reactions proceeded under a variety of different conditions. Both the concentration of oxygen and the temperature were varied. Photodetectors collected the thousands of data points. Numerical models based on the hypothetical chemistry (see Table 1) simulated the data within error. However, optimizing the kinetic parameters in the numerical model proved disastrous. No reasonable fits were found that modeled the data with confidence. Taylor et al. were not sure if the model was incorrect or if the procedure they were using to fit the data was incorrect. They needed a method to determine the best set of parameters, also known as the globally optimum set, that would fit the data to their model.

#	Reaction	k298 (M-1μs-1 or μs-1)
1	(CH3)3CO + 1,4-C6H8 → c-C6H7 + (CH3)3COH	53
2	c-C6H8 + O-2 ⇔ p-C6H7OO	[1, 1200]
3	c-C6H8 + O-2 ⇔ o-C6H7OO	[1, 1200]
4	o-C6H8 + O-2 ⇔ C6H6 + HO2	[0.001, 100]
5	2 c-C6H8 → Products	1200