Data set for the article to be published in PNAS "Adiabatic computing for optimal thermodynamic efficiency of information processing" This data set contains: optimal_1ms.zip : raw data files of the response to an optimal erasure protocol of characteristic duration 1ms. This Matlab file includes all necessary quantitites to compute the average work and heat during the procedure ( ie position trajectory and driving parameters). optimal_3ms_part1 & _part2.zip: same for optimal protocols of 3ms. Analysis_code_optimal.m: Analysis Matlab code to be run on the mat files above. The code run the full analysis and output the work, heat, kinetic and potential energy and some calibration parameters. The processed data is saved in optimal_1 & 3ms_result.mat. Fast_erasure_v1_0.3 & _0.12.zip: same for basic protocol at v1=0 & v1=0.12. The processed data is saved in Fast_erasure_v1_0.3 & _0.12.mat. This data corresponds tp the two speeds used to draw the heating kinetic energy curves of Fig.4 BasicAndOptimalTranslations.mat: Mat file containing the data used to create Fig. 3: tprotocol (ms), x1basprotocol and x1optprotocol (sigma) define the ramps of the well center for the basic and optimal protocols (time and value of x1 for each ramp). t (ms), xbas and xopt (sigma) are the recording from the experiment (time, and arrays corresponding to 2000 measurements with 10ms of data sampled at 2Mhz each). The data is post processed with a low pass filter at 8kHz (filtfilt function in Matlab, using a 4th order butterworth filter). Fig. 3 correspond to the average of those 2000 trajectories. f0 (Hz) is the resonance frequency of the cantilever. FigX.fig: Matlab format figure source used to draw all of the plots of the article, embedding the all data.