A variety of methods exist for temporally compressing (shortening) optical pulses. These methods typically start with pulses in the picosecond or femtosecond range, and end up with pulses that can be as short as a few optical cycles. The optical bandwidth of the initial pulse is usually increased using a nonlinear interaction such as self-phase modulation; this leads to a chirped pulse, which sometimes ends up being longer than the original pulse. A well-known technique for generating sub-100 fs pulses is nonlinear compression in a fiber, where the fiber's nonlinearity is used to broaden the optical spectrum. Thereafter, the pulse duration is reduced using linear dispersive compression, which removes the chirp by flattening the spectral phase. This is accomplished by sending the pulse through an optical element with a suitable amount of dispersion, such as a prism pair, an optical fiber, a grating compressor, or a chirped mirror.

In the 1960s, Gires and Tournois and Giordmaine et al. independently proposed the shortening of optical pulses using compression techniques analogous to those used at microwave frequencies. Fisher et al. suggested that femtosecond optical pulses could be obtained by first passing a short pulse through an optical Kerr liquid in order to impress a frequency sweep or “chirp” on the pulse's carrier. Pulse compression was then to be achieved by compensating the frequency sweep in the pulse frequency spectrum using a dispersive delay line. In 1982, Shank et al.