

Data associated with the article "Tuning local microstructure of colloidal gels by ultrasound-activated deformable inclusions" by B. Saint-Michel, G. Petekidis, V. Garbin. 

Preprint available on arxiv.org: https://arxiv.org/abs/2112.07592 
 
Contact for dataset: bsaintmichel (as seen on github, and gmail)

# Contents

## 1. This Readme file 

## 2. Experiment files

Essentially we have two datasets :

* Dissolution : Medium-sized bubble, *without* ultrasound
* Acoustics : Medium-large sized bubble, *with* ultrasound [for details about the pulse, see Acoustics.log, Section 3]

These files are labeled as :
* `XXXX_bubble.json` contains a table containing bubble information to be read with Pandas, with keys/columns
  * `bub_x` (in pixels), 
  * `bub_y` (in pixels), 
  * `bub_rad` (in pixels),
  * `bub_ecc` (eccentricity), 
  * `exclu_rad` (exclusion radius, i.e. bubble + estimated solvent pocket radius)
  * `frame` (= index), the frame at which these observations are made
* `XXXX_info.json` contains a dict containing metadata for the experiment ; notable fields of the dict are 
  * `width` (in pixels), 
  * `height` (in pixels), 
  * `date`, 
  * `pixsize` (in micrometers), 
  * `time` (straightforward)
  * `figure_frames` (the frames used to produce the figure)
* `XXXX_particles.json` contains a (large) table containing information on the particles surrounding the gel, to be read with Pandas, with keys/columns 
  * `x` (in pixels), 
  * `y` (in pixels), 
  * `n_neighbors` (straightforward)
  * `frame` (where they are found)
* `XXXX_structure.json` contains a dict of processed data regarding the particle distribution around the bubble [see main article for more details]. 
  * `phieff` (phi, bins_ctrs) : profile of $\varphi(r)$ (in the paper), the estimated volume fraction of the particles based on particle detection :
  * `vv_pdf` (avg, conditional, bins_ctrs) : void sizes (**not volumes**) distribution, (either in all the sample, or conditioned by $r/R$ through `segment_edges`)
  * `bond_pdf` (avg, conditional, bins_ctrs) : bond orientation $\phi^{ij}$ (in the paper), conditional averaged $\phi^{ij}_s$, also number of bonds detected in each segment $N_s$ defined by `segment_edges. 
  NB : $\sum_s \phi^{ij}_{\rm cond, s} N_s / \sum_s N_s = \phi^{ij}_{\rm avg}$ for all $r$
  * `g` (avg, conditional, bins_ctrs) : pair correlation function of the particles. Corrections due to image boundary and bubble are already implemented.
Note: all conditional arrays are in the form (frame, segment n°, extra axis [r between particles, void size, etc.]), while all average arrays ditch the second dimension (no segment n°).

## 3. Acoustics log file 

The file `Acoustics.txt` logs the different acoustic pulses that have been sent to the transducer. It is essentially a TAB delimited file with fields 

* `frames` : Index of the frames between which the pulse has been applied (we never apply when acquiring a frame)
* `V` : the voltage amplitude sent by the waveform generator (in mV)
* `Gain` : the % gain of the radiofrequency amplifier (AG 2021 by T&C Power Conversion Inc.) 
* `f` :  the frequency of the pulse (in kHz)
* `Ncyc` : the number of oscillation cycles
* `Rem` : (*remarque*) Comments and observations after applying the pulse

# How were these files acquired ?

All data were either directly input by myself (Acoustics.txt) or acquired using a Zeiss LSM 710 confocal microscope. The corresponding .lsm, saved using ZEN (black edition) files have then been processed using `Trackpy` to obtain the position of the particles, and through a Jupyter Notebook script to extract the bubble radius, position and exclusion area, and the corresponding processed variables of `XXXX_data.JSON`. Nearest neighbours were computed using `scikit-learn` library NearestNeighbours.

# License 

CC0