Predict a parabola with LSTM deep learning

Hi all,

I'm testing deep learning LSTM networks and I'm trying to predict a parabola, which is a non periodic function, after having succeeded in predicting periodic ones.

I obtained the first two curves with these settings:

numFeatures = 1;
numResponses = 1;
numHiddenUnits = 200;

layers = [ ...
    sequenceInputLayer(numFeatures)
    bilstmLayer(numHiddenUnits)
    fullyConnectedLayer(numResponses)
    regressionLayer];

options = trainingOptions('adam', ...
    'MaxEpochs',500, ...
    'GradientThreshold',1, ...
    'InitialLearnRate',0.005, ...
    'LearnRateSchedule','piecewise', ...
    'SequenceLength','longest',...
    'LearnRateDropPeriod',1250, ...
    'LearnRateDropFactor',0.2, ...
    'Verbose',0, ...
    'Plots','training-progress');

The third one is what I would like to obtain, and this is possible using LSTM as explained here, where some people has done it in python.

I'm wondering how to obtain the same with LSTM layer in MATLAB