Skip to content

Commit 95a3a43

Browse files
Matias LorenzMatias Lorenz
Matias Lorenz
authored and
Matias Lorenz
committed
updated README and Dockerfile
1 parent 72db35d commit 95a3a43

File tree

4 files changed

+196
-83
lines changed

4 files changed

+196
-83
lines changed

‎Dockerfile‎

Lines changed: 10 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,7 @@
55
ARG MATLAB_RELEASE=R2024b
66

77
# Specify the extra products to install into the image. These products can either be toolboxes or support packages.
8-
ARG ADDITIONAL_PRODUCTS="Symbolic_Math_Toolbox Deep_Learning_Toolbox_Model_for_ResNet-50_Network"
8+
ARG ADDITIONAL_PRODUCTS="Statistics_and_Machine_Learning_Toolbox Optimization_Toolbox Parallel_Computing_Toolbox Signal_Processing_Toolbox"
99

1010
# This Dockerfile builds on the Ubuntu-based mathworks/matlab image.
1111
# To check the available matlab images, see: https://hub.docker.com/r/mathworks/matlab
@@ -29,8 +29,11 @@ RUN export DEBIAN_FRONTEND=noninteractive \
2929
&& rm -rf /var/lib/apt/lists/*
3030

3131
# Install C compiler (gcc)
32-
RUN apt-get update && apt-get install -y gcc g++ && rm -rf /var/lib/apt/lists/*
33-
32+
RUN sed -i 's|http://archive.ubuntu.com|http://mirror.math.princeton.edu/pub/mirrors|' /etc/apt/sources.list && \
33+
apt-get update && \
34+
apt-get install -y gcc g++ && \
35+
rm -rf /var/lib/apt/lists/*
36+
3437
# Run mpm to install MathWorks products into the existing MATLAB installation directory,
3538
# and delete the mpm installation afterwards.
3639
# Modify it by setting the ADDITIONAL_PRODUCTS defined above,
@@ -77,11 +80,11 @@ ENV MW_DDUX_FORCE_ENABLE=true MW_CONTEXT_TAGS=$MW_CONTEXT_TAGS,MATLAB:TOOLBOXES:
7780
WORKDIR /home/matlab
7881
# Inherit ENTRYPOINT and CMD from base image.
7982
# Install git for cloning the repository
80-
RUN sudo apt-get update && sudo apt-get install -y git && sudo apt-get clean
83+
# RUN apt-get update && apt-get install -y git && apt-get clean
8184

8285
# Clone the MINT repository and set working directory
83-
RUN git clone https://github.com/panzerilab/MINT.git /home/matlab/MINT
84-
WORKDIR /home/matlab/MINT
86+
# RUN git clone https://github.com/panzerilab/MINT.git /home/matlab/MINT
87+
# WORKDIR /home/matlab/MINT
8588

8689
# Run BuildMINT.m script
87-
RUN matlab -batch "BuildMINT.m"
90+
# RUN matlab -batch "BuildMINT.m"

‎How_to_use_MINT/How_to_compute_MI_live.mlx‎

2.5 KB
Binary file not shown.

‎README.md‎

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -34,16 +34,16 @@
3434
### Build with Docker
3535
1. If you have Docker installed, you can build the Dockerfile in this package by running
3636
```
37-
docker build --build-arg ADDITIONAL_PRODUCTS="Statistics_and_Machine_Learning_Toolbox Optimization_Toolbox Parallel_Computing_Toolbox Signal_processing_Toolbox" -t matlab4mint:R2024b .
37+
docker build --platform=linux/amd64 -t matlab4mint:R2024b .
3838
```
3939
2. After the Docker image has been built with all the necessary toolboxes, run
4040
```
41-
docker run --init -v /path/on/host:/home/matlab/shared -e MLM_LICENSE_FILE=27000@MyServerName matlab4mint:R2024b
41+
docker run --init --platform=linux/amd64 -v /path/on/host:/home/matlab/Documents -it matlab4mint:R2024b
4242
```
43-
Where /path/on/host can be replaced with the directory in your computer where you can work persistently after the Docker image is used. If you do not know your license hostname, you can skip -e MLM_LICENSE_FILE=27000@MyServerName and you can login with your username and password when you are inside the container.
43+
Where /path/on/host can be replaced with the directory in your computer where you can work persistently after the Docker image is used, be sure that MINT is inside that directory. If you do not know your license hostname, you can skip -e MLM_LICENSE_FILE=27000@MyServerName and you can login with your username and password when you are inside the container.
4444
3. Once inside the container, write
4545
```
46-
cd /home/matlab/MINT
46+
cd /path/to/MINT
4747
run('BuildMINT.m')
4848
```
4949

‎src/core_functions/BiasCorrection/extrapolation.m‎

Lines changed: 182 additions & 72 deletions
Original file line numberDiff line numberDiff line change
@@ -55,95 +55,201 @@
5555
end
5656
nTrials = size(inputs{1},length(size(inputs{1})));
5757

58-
for i=1:opts.xtrp
59-
randidx = randperm(nTrials, nTrials);
60-
npartition = [1 2 4];
61-
PID2 = repmat({zeros(1,nTimepoints)}, 1, length(outputs));
62-
PID4 = repmat({zeros(1,nTimepoints)}, 1, length(outputs));
63-
I1_2 = 0;
64-
I2_2 = 0;
65-
I12_2 = 0;
66-
I1_4 = 0;
67-
I2_4 = 0;
68-
I12_4 = 0;
69-
inputs_s = inputs;
70-
idx = repmat({':'}, 1, ndims(inputs{1}));
71-
idx{end} = randidx;
72-
for var = 1:length(inputs)
73-
inputs_s{var} = inputs{var}(idx{:});
74-
end
75-
for np=2:length(npartition)
76-
for pidx = 1:npartition(np)
77-
inputs_p = partition(inputs_s, npartition(np), pidx,0);
78-
PID_p = feval(corefunc, inputs_p, outputs, plugin_opts);
79-
for outIdx = 1:length(outputs)
80-
if npartition(np)==2
81-
PID2{outIdx} = PID2{outIdx} + PID_p{outIdx}/2;
82-
elseif npartition(np)==4
83-
PID4{outIdx} = PID4{outIdx} + PID_p{outIdx}/4;
58+
if opts.parallel
59+
noutputs = length(outputs);
60+
corrected_v_par = zeros(opts.xtrp,nTimepoints);
61+
I1_corrected_par = zeros(opts.xtrp,nTimepoints);
62+
I2_corrected_par = zeros(opts.xtrp,nTimepoints);
63+
I12_corrected_par = zeros(opts.xtrp,nTimepoints);
64+
parfor i=1:opts.xtrp
65+
randidx = randperm(nTrials, nTrials);
66+
npartition = [1 2 4];
67+
PID2 = repmat({zeros(1,nTimepoints)}, 1, length(outputs));
68+
PID4 = repmat({zeros(1,nTimepoints)}, 1, length(outputs));
69+
I1_2 = 0;
70+
I2_2 = 0;
71+
I12_2 = 0;
72+
I1_4 = 0;
73+
I2_4 = 0;
74+
I12_4 = 0;
75+
inputs_s = inputs;
76+
idx = repmat({':'}, 1, ndims(inputs{1}));
77+
idx{end} = randidx;
78+
for var = 1:length(inputs)
79+
inputs_s{var} = inputs{var}(idx{:});
80+
end
81+
for np=2:length(npartition)
82+
for pidx = 1:npartition(np)
83+
inputs_p = partition(inputs_s, npartition(np), pidx,0);
84+
PID_p = feval(corefunc, inputs_p, outputs, plugin_opts);
85+
for outIdx = 1:noutputs
86+
if npartition(np)==2
87+
PID2{outIdx} = PID2{outIdx} + PID_p{outIdx}/2;
88+
elseif npartition(np)==4
89+
PID4{outIdx} = PID4{outIdx} + PID_p{outIdx}/4;
90+
end
8491
end
85-
end
86-
if opts.pid_constrained
87-
I1_p = MI({inputs_p{1}, inputs_p{end}}, {'I(A;B)'}, plugin_opts);
88-
I2_p = MI({inputs_p{2}, inputs_p{end}}, {'I(A;B)'}, plugin_opts);
89-
I12_p = MI({cat(1, inputs_p{1}, inputs_p{2}), inputs_p{end}}, {'I(A;B)'}, plugin_opts);
90-
if npartition(np)==2
91-
I1_2 = I1_2 + I1_p{1}/2;
92-
I2_2 = I2_2 + I2_p{1}/2;
93-
I12_2 = I12_2 + I12_p{1}/2;
94-
elseif npartition(np)==4
95-
I1_4 = I1_4 + I1_p{1}/4;
96-
I2_4 = I2_4 + I2_p{1}/4;
97-
I12_4 = I12_4 + I12_p{1}/4;
92+
if opts.pid_constrained
93+
I1_p = MI({inputs_p{1}, inputs_p{end}}, {'I(A;B)'}, plugin_opts);
94+
I2_p = MI({inputs_p{2}, inputs_p{end}}, {'I(A;B)'}, plugin_opts);
95+
I12_p = MI({cat(1, inputs_p{1}, inputs_p{2}), inputs_p{end}}, {'I(A;B)'}, plugin_opts);
96+
if npartition(np)==2
97+
I1_2 = I1_2 + I1_p{1}/2;
98+
I2_2 = I2_2 + I2_p{1}/2;
99+
I12_2 = I12_2 + I12_p{1}/2;
100+
elseif npartition(np)==4
101+
I1_4 = I1_4 + I1_p{1}/4;
102+
I2_4 = I2_4 + I2_p{1}/4;
103+
I12_4 = I12_4 + I12_p{1}/4;
104+
end
98105
end
99106
end
100107
end
101-
end
102-
x_extrap = npartition./nTrials;
103-
if strcmp(opts.bias,'qe')
104-
for t = 1:nTimepoints
105-
for outIdx = 1:length(outputs)
106-
y = [plugin_v{outIdx}(t), PID2{outIdx}(t), PID4{outIdx}(t)];
107-
p = polyfit(x_extrap, y, 2);
108-
corrected_v{outIdx}(1,t) = corrected_v{outIdx}(t) + p(3)/xtrp;
108+
x_extrap = npartition./nTrials;
109+
if strcmp(opts.bias,'qe')
110+
for t = 1:nTimepoints
111+
for outIdx = 1:noutputs
112+
y = [plugin_v{outIdx}(t), PID2{outIdx}(t), PID4{outIdx}(t)];
113+
p = polyfit(x_extrap, y, 2);
114+
corrected_v_par(outIdx, i, t) = p(3);
115+
end
116+
end
117+
elseif strcmp(opts.bias,'le')
118+
for t = 1:nTimepoints
119+
for outIdx = 1:noutputs
120+
y = [plugin_v{outIdx}(t), PID2{outIdx}(t)];
121+
p = polyfit(x_extrap, y, 1);
122+
corrected_v_par(outIdx, i, t) = p(2);
123+
end
109124
end
110125
end
111-
elseif strcmp(opts.bias,'le')
112-
for t = 1:nTimepoints
113-
for outIdx = 1:length(outputs)
114-
y = [plugin_v{outIdx}(t), PID2{outIdx}(t)];
115-
p = polyfit(x_extrap, y, 1);
116-
corrected_v{outIdx}(1,t) = corrected_v{outIdx}(t) + p(2)/xtrp;
126+
if opts.pid_constrained
127+
if strcmp(opts.bias,'qe')
128+
for t = 1:nTimepoints
129+
y = [I1_plugin(t), I1_2(t), I1_4(t)];
130+
p = polyfit(x_extrap, y, 2);
131+
I1_corrected_par(i,t) = p(3);
132+
y = [I2_plugin(t), I2_2(t), I2_4(t)];
133+
p = polyfit(x_extrap, y, 2);
134+
I2_corrected_par(i,t) = p(3);
135+
y = [I12_plugin(t), I12_2(t), I12_4(t)];
136+
p = polyfit(x_extrap, y, 2);
137+
I12_corrected_par(i,t) = p(3);
138+
end
139+
elseif strcmp(opts.bias,'le')
140+
for t = 1:nTimepoints
141+
y = [I1_plugin(t), I1_2(t)];
142+
p = polyfit(x_extrap, y, 1);
143+
I1_corrected_par(i,t) = p(2);
144+
y = [I2_plugin(t), I2_2(t)];
145+
p = polyfit(x_extrap, y, 1);
146+
I2_corrected_par(i,t) = p(2);
147+
y = [I12_plugin(t), I12_2(t)];
148+
p = polyfit(x_extrap, y, 1);
149+
I12_corrected_par(i,t) = p(2);
150+
end
117151
end
118152
end
119153
end
120-
if opts.pid_constrained
154+
for spar=1:noutputs
155+
corrected_v{spar}(1,:) = mean(corrected_v_par(spar,:,:),2);
156+
end
157+
I1_corrected = mean(I1_corrected_par,1);
158+
I2_corrected = mean(I2_corrected_par,1);
159+
I12_corrected = mean(I12_corrected_par,1);
160+
161+
else
162+
for i=1:opts.xtrp
163+
randidx = randperm(nTrials, nTrials);
164+
npartition = [1 2 4];
165+
PID2 = repmat({zeros(1,nTimepoints)}, 1, length(outputs));
166+
PID4 = repmat({zeros(1,nTimepoints)}, 1, length(outputs));
167+
I1_2 = 0;
168+
I2_2 = 0;
169+
I12_2 = 0;
170+
I1_4 = 0;
171+
I2_4 = 0;
172+
I12_4 = 0;
173+
inputs_s = inputs;
174+
idx = repmat({':'}, 1, ndims(inputs{1}));
175+
idx{end} = randidx;
176+
for var = 1:length(inputs)
177+
inputs_s{var} = inputs{var}(idx{:});
178+
end
179+
for np=2:length(npartition)
180+
for pidx = 1:npartition(np)
181+
inputs_p = partition(inputs_s, npartition(np), pidx,0);
182+
PID_p = feval(corefunc, inputs_p, outputs, plugin_opts);
183+
for outIdx = 1:length(outputs)
184+
if npartition(np)==2
185+
PID2{outIdx} = PID2{outIdx} + PID_p{outIdx}/2;
186+
elseif npartition(np)==4
187+
PID4{outIdx} = PID4{outIdx} + PID_p{outIdx}/4;
188+
end
189+
end
190+
if opts.pid_constrained
191+
I1_p = MI({inputs_p{1}, inputs_p{end}}, {'I(A;B)'}, plugin_opts);
192+
I2_p = MI({inputs_p{2}, inputs_p{end}}, {'I(A;B)'}, plugin_opts);
193+
I12_p = MI({cat(1, inputs_p{1}, inputs_p{2}), inputs_p{end}}, {'I(A;B)'}, plugin_opts);
194+
if npartition(np)==2
195+
I1_2 = I1_2 + I1_p{1}/2;
196+
I2_2 = I2_2 + I2_p{1}/2;
197+
I12_2 = I12_2 + I12_p{1}/2;
198+
elseif npartition(np)==4
199+
I1_4 = I1_4 + I1_p{1}/4;
200+
I2_4 = I2_4 + I2_p{1}/4;
201+
I12_4 = I12_4 + I12_p{1}/4;
202+
end
203+
end
204+
end
205+
end
206+
x_extrap = npartition./nTrials;
121207
if strcmp(opts.bias,'qe')
122208
for t = 1:nTimepoints
123-
y = [I1_plugin(t), I1_2(t), I1_4(t)];
124-
p = polyfit(x_extrap, y, 2);
125-
I1_corrected(t) = I1_corrected(t) + p(3)/xtrp;
126-
y = [I2_plugin(t), I2_2(t), I2_4(t)];
127-
p = polyfit(x_extrap, y, 2);
128-
I2_corrected(t) = I2_corrected(t) + p(3)/xtrp;
129-
y = [I12_plugin(t), I12_2(t), I12_4(t)];
130-
p = polyfit(x_extrap, y, 2);
131-
I12_corrected(t) = I12_corrected(t) + p(3)/xtrp;
209+
for outIdx = 1:length(outputs)
210+
y = [plugin_v{outIdx}(t), PID2{outIdx}(t), PID4{outIdx}(t)];
211+
p = polyfit(x_extrap, y, 2);
212+
corrected_v{outIdx}(1,t) = corrected_v{outIdx}(t) + p(3)/xtrp;
213+
end
132214
end
133215
elseif strcmp(opts.bias,'le')
134216
for t = 1:nTimepoints
135-
y = [I1_plugin(t), I1_2(t)];
136-
p = polyfit(x_extrap, y, 1);
137-
I1_corrected(t) = I1_corrected(t) + p(2)/xtrp;
138-
y = [I2_plugin(t), I2_2(t)];
139-
p = polyfit(x_extrap, y, 1);
140-
I2_corrected(t) = I2_corrected(t) + p(2)/xtrp;
141-
y = [I12_plugin(t), I12_2(t)];
142-
p = polyfit(x_extrap, y, 1);
143-
I12_corrected(t) = I12_corrected(t) + p(2)/xtrp;
217+
for outIdx = 1:length(outputs)
218+
y = [plugin_v{outIdx}(t), PID2{outIdx}(t)];
219+
p = polyfit(x_extrap, y, 1);
220+
corrected_v{outIdx}(1,t) = corrected_v{outIdx}(t) + p(2)/xtrp;
221+
end
222+
end
223+
end
224+
if opts.pid_constrained
225+
if strcmp(opts.bias,'qe')
226+
for t = 1:nTimepoints
227+
y = [I1_plugin(t), I1_2(t), I1_4(t)];
228+
p = polyfit(x_extrap, y, 2);
229+
I1_corrected(t) = I1_corrected(t) + p(3)/xtrp;
230+
y = [I2_plugin(t), I2_2(t), I2_4(t)];
231+
p = polyfit(x_extrap, y, 2);
232+
I2_corrected(t) = I2_corrected(t) + p(3)/xtrp;
233+
y = [I12_plugin(t), I12_2(t), I12_4(t)];
234+
p = polyfit(x_extrap, y, 2);
235+
I12_corrected(t) = I12_corrected(t) + p(3)/xtrp;
236+
end
237+
elseif strcmp(opts.bias,'le')
238+
for t = 1:nTimepoints
239+
y = [I1_plugin(t), I1_2(t)];
240+
p = polyfit(x_extrap, y, 1);
241+
I1_corrected(t) = I1_corrected(t) + p(2)/xtrp;
242+
y = [I2_plugin(t), I2_2(t)];
243+
p = polyfit(x_extrap, y, 1);
244+
I2_corrected(t) = I2_corrected(t) + p(2)/xtrp;
245+
y = [I12_plugin(t), I12_2(t)];
246+
p = polyfit(x_extrap, y, 1);
247+
I12_corrected(t) = I12_corrected(t) + p(2)/xtrp;
248+
end
144249
end
145250
end
146251
end
252+
147253
end
148254
if opts.pid_constrained
149255
if ~isfield(opts, 'chosen_atom')
@@ -232,6 +338,10 @@
232338
plugin_opts.bias = 'plugin';
233339
plugin_opts.recall = true;
234340
nTrials = size(inputs{1},length(size(inputs{1})));
341+
342+
if opts.parallel
343+
else
344+
end
235345
for i=1:opts.xtrp
236346
randidx = randperm(nTrials, nTrials);
237347
npartition = [1 2 4];

0 commit comments

Comments
 (0)