Re: Use of transmat matrix in full stochastic unit commitment

[Ray Daniel Zimmerman]

I’m afraid there are no video descriptions. At the moment, the available 
documentation consists of the MOST User’s 
Manual and the two MOST papers 
(see refs 4 and 5 here).

   Ray


On Jan 18, 2020, at 4:50 AM, sebaa...@yahoo.fr wrote:


Dear pr zemmerman
I thank you very much Can you send to me a link to some vedios that explain 
Most tool as sransmat constitution for a network or other associated files

--
Envoyé depuis l'application Yahoo Email App pour Android

vendredi, 17 janvier 2020, 11:50PM +01:00 de Ray Daniel Zimmerman 
r...@cornell.edu:

Hi Carlos,

1. You need not use ex_transmat() at all. It was simply a convenience function 
used to create the transmat cell array of transmission probability matrices for 
each period. You can write your own function to create your own general cell 
array of matrices which can certainly vary by period.

2. From Table 5-3 in the MOST User’s Manual, you’ll see that the 3rd dimension 
of the values field of a profile corresponds to the indices specified by the 
rows field. In the example rows = 1 and the 3rd dimension of values is a 
singleton. If you set rows to a vector, say rows  = [1; 3; 4], then values(:, 
:, 1), then values(:, :, 2) and then values(:, :, 3) would correspond to the 
values for wind units 1, 3, and 4 respectively.

3. I’m not sure I follow the question. In neither case are we optimizing 
explicit individual trajectories. We are optimizing the cost of a set of 
probability weighted scenarios (and transitions) with certain costs and 
constraints on the envelope of trajectories.

4. Not exactly. It is actually an approximation of a multi-stage problem, where 
the full multi-stage decision tree is approximated with a Markovian decision 
process. Reference [5] in the 
manual (also ref 5 at https://matpower.org/publications/) attempts to explain 
this.

Hope this helps,

Ray

[5] A. J. Lamadrid, D. Munoz-Alvarez, C. E. Murillo-Sanchez, R. D. Zimmerman, 
H. D. Shin and R. J. Thomas, “Using the MATPOWER Optimal Scheduling Tool to 
Test Power System Operation Methodologies Under Uncertainty,” Sustainable 
Energy, IEEE Transactions on, vol. 10, no. 3, pp. 1280–1289, July 2019. DOI: 
10.1109/TSTE.2018.2865454


On Jan 15, 2020, at 11:04 AM, Carlos Ferrandon Cervantes 
mailto:ferrand...@gmail.com>> wrote:

Hello everyone:

I've been working with MOST for some time now; specifically with the unit 
commitment problem. I have been able to add some linearised constraints to it, 
but now I face the challenge of using the stochastic programming option in 
MOST. I've followed the example "most_ex7_suc" for these cases: Stochastic Unit 
Commitment - Individual Trajectories and Stochastic Unit Commitment - Full 
Transition Probabilities and my questions are the next ones:

1. I've realised that in the matrix "transmat", only the first column is full 
with the three scenarios' probabilities, but for the rest of the columns we 
have the identity matrix size 3x3 for each time step.Then the "scenario_probs" 
variable is computed/updated with the operation:  mdi.tstep(t).TransMat * 
mdi.CostWeights(1, 1:mdi.idx.nj(t-1), t-1). My question is: Can these scenario 
probabilities change through time? I mean, that for every column we could have 
different scenario probabilities in "transmat". And if this was the case, what 
changes would be needed for the function "ex_transmat" to reflect these changes?

2. For the wind input in the example  "most_ex7_suc"  we have three different 
values of wind. I am assuming they belong to the three scenarios specified but 
only for that wind input. If we had different wind inputs from different buses, 
what would the arrangement of windprofiles.values(:,:,:) be?

3. How can we see the difference betwen the Individual Trajectories and Full 
Transition Probabilities case? I've understood from the manual that the system 
"stays" in one path in the case of Individual Trajectories, but for the Full 
Transition Probabilities how can we see the transition between the three 
scenarios' possible paths?

4. Can we that MOST is a Two-Stage model in Full Stochastic Programming then?

Right now Im only working with the full system, i.e. no contingencies.

As usual, thank you so much in advance,

--
Carlos Ferrandon

Re[2]: Use of transmat matrix in full stochastic unit commitment

[sebaa_73]

Dear pr zemmerman 
I thank you very much Can you send to me a link to some vedios that explain 
Most tool as sransmat constitution for a network or other associated files 
--
Envoyé depuis l'application Yahoo Email App pour Android vendredi, 17 janvier 
2020, 11:50PM +01:00 de Ray Daniel Zimmerman  r...@cornell.edu :

>Hi Carlos,
>
>1. You need not use  ex_transmat()  at all. It was simply a convenience 
>function used to create the transmat cell array of transmission probability 
>matrices for each period. You can write your own function to create your own 
>general cell array of matrices which can certainly vary by period.
>
>2. From Table 5-3 in the MOST User’s Manual, you’ll see that the 3rd dimension 
>of the values field of a profile corresponds to the indices specified by the 
>rows field. In the example rows = 1 and the 3rd dimension of values is a 
>singleton. If you set rows to a vector, say rows  = [1; 3; 4] , then  
>values(:, :, 1) , then  values(:, :, 2) and then values(:, :, 3) would 
>correspond to the values for wind units 1, 3, and 4 respectively.
>
>3. I’m not sure I follow the question. In neither case are we optimizing 
>explicit individual trajectories. We are optimizing the cost of a set of 
>probability weighted scenarios (and transitions) with certain costs and 
>constraints on the envelope
> of trajectories.
>
>4. Not exactly. It is actually an approximation of a multi-stage problem, 
>where the full multi-stage decision tree is approximated with a Markovian 
>decision process. Reference  [5]  in
> the manual (also ref 5 at  https://matpower.org/publications/ ) attempts to 
> explain this.
>
>Hope this helps,
>
>    Ray
>
>[5] A. J. Lamadrid, D. Munoz-Alvarez, C. E. Murillo-Sanchez, R. D. Zimmerman, 
>H. D. Shin and R. J. Thomas, “Using the MATPOWER Optimal Scheduling Tool to 
>Test Power System Operation Methodologies Under Uncertainty,” Sustainable 
>Energy, IEEE Transactions on,
> vol. 10, no. 3, pp. 1280–1289, July 2019.  DOI:   10.1109/TSTE.2018.2865454
>
>
>>On Jan 15, 2020, at 11:04 AM, Carlos Ferrandon Cervantes < 
>>ferrand...@gmail.com> wrote:
>>Hello everyone:
>>
>>I've been working with MOST for some time now; specifically with the unit 
>>commitment problem. I have been able to add some linearised constraints to 
>>it, but now I face the challenge of using the stochastic programming option 
>>in MOST. I've followed
>> the example "most_ex7_suc" for these cases: Stochastic Unit Commitment - 
>> Individual Trajectories and Stochastic Unit Commitment - Full Transition 
>> Probabilities and my questions are the next ones:
>>
>>1. I've realised that in the matrix "transmat", only the first column is full 
>>with the three scenarios' probabilities, but for the rest of the columns we 
>>have the identity matrix size 3x3 for each time step.Then the 
>>"scenario_probs" variable is
>> computed/updated with the operation:  mdi.tstep(t).TransMat * 
>> mdi.CostWeights(1, 1:mdi.idx.nj(t-1), t-1). My question is: Can these 
>> scenario probabilities change through time? I mean, that for every column we 
>> could have different scenario probabilities in
>> "transmat". And if this was the case, what changes would be needed for the 
>> function "ex_transmat" to reflect these changes?
>>
>>2. For the wind input in the example  "most_ex7_suc"  we have three different 
>>values of wind. I am assuming they belong to the three scenarios specified 
>>but only for that wind input. If we had different wind inputs from different 
>>buses, what would
>> the arrangement of windprofiles.values(:,:,:) be?
>>
>>3. How can we see the difference betwen the Individual Trajectories and Full 
>>Transition Probabilities case? I've understood from the manual that the 
>>system "stays" in one path in the case of Individual Trajectories, but for 
>>the Full Transition
>> Probabilities how can we see the transition between the three scenarios' 
>> possible paths?
>>
>>4. Can we that MOST is a Two-Stage model in Full Stochastic Programming then?
>>
>>Right now Im only working with the full system, i.e. no contingencies.
>>
>>As usual, thank you so much in advance,
>>
>>-- 
>>Carlos Ferrandon

Re[2]: Use of transmat matrix in full stochastic unit commitment

[sebaa_73]

I just finishing and obtain m'y doctorante degree and now i start with 
stochastic UC problem that i love to do a paper in this way
--
Envoyé depuis l'application Yahoo Email App pour Android vendredi, 17 janvier 
2020, 11:50PM +01:00 de Ray Daniel Zimmerman  r...@cornell.edu :

>Hi Carlos,
>
>1. You need not use  ex_transmat()  at all. It was simply a convenience 
>function used to create the transmat cell array of transmission probability 
>matrices for each period. You can write your own function to create your own 
>general cell array of matrices which can certainly vary by period.
>
>2. From Table 5-3 in the MOST User’s Manual, you’ll see that the 3rd dimension 
>of the values field of a profile corresponds to the indices specified by the 
>rows field. In the example rows = 1 and the 3rd dimension of values is a 
>singleton. If you set rows to a vector, say rows  = [1; 3; 4] , then  
>values(:, :, 1) , then  values(:, :, 2) and then values(:, :, 3) would 
>correspond to the values for wind units 1, 3, and 4 respectively.
>
>3. I’m not sure I follow the question. In neither case are we optimizing 
>explicit individual trajectories. We are optimizing the cost of a set of 
>probability weighted scenarios (and transitions) with certain costs and 
>constraints on the envelope
> of trajectories.
>
>4. Not exactly. It is actually an approximation of a multi-stage problem, 
>where the full multi-stage decision tree is approximated with a Markovian 
>decision process. Reference  [5]  in
> the manual (also ref 5 at  https://matpower.org/publications/ ) attempts to 
> explain this.
>
>Hope this helps,
>
>    Ray
>
>[5] A. J. Lamadrid, D. Munoz-Alvarez, C. E. Murillo-Sanchez, R. D. Zimmerman, 
>H. D. Shin and R. J. Thomas, “Using the MATPOWER Optimal Scheduling Tool to 
>Test Power System Operation Methodologies Under Uncertainty,” Sustainable 
>Energy, IEEE Transactions on,
> vol. 10, no. 3, pp. 1280–1289, July 2019.  DOI:   10.1109/TSTE.2018.2865454
>
>
>>On Jan 15, 2020, at 11:04 AM, Carlos Ferrandon Cervantes < 
>>ferrand...@gmail.com> wrote:
>>Hello everyone:
>>
>>I've been working with MOST for some time now; specifically with the unit 
>>commitment problem. I have been able to add some linearised constraints to 
>>it, but now I face the challenge of using the stochastic programming option 
>>in MOST. I've followed
>> the example "most_ex7_suc" for these cases: Stochastic Unit Commitment - 
>> Individual Trajectories and Stochastic Unit Commitment - Full Transition 
>> Probabilities and my questions are the next ones:
>>
>>1. I've realised that in the matrix "transmat", only the first column is full 
>>with the three scenarios' probabilities, but for the rest of the columns we 
>>have the identity matrix size 3x3 for each time step.Then the 
>>"scenario_probs" variable is
>> computed/updated with the operation:  mdi.tstep(t).TransMat * 
>> mdi.CostWeights(1, 1:mdi.idx.nj(t-1), t-1). My question is: Can these 
>> scenario probabilities change through time? I mean, that for every column we 
>> could have different scenario probabilities in
>> "transmat". And if this was the case, what changes would be needed for the 
>> function "ex_transmat" to reflect these changes?
>>
>>2. For the wind input in the example  "most_ex7_suc"  we have three different 
>>values of wind. I am assuming they belong to the three scenarios specified 
>>but only for that wind input. If we had different wind inputs from different 
>>buses, what would
>> the arrangement of windprofiles.values(:,:,:) be?
>>
>>3. How can we see the difference betwen the Individual Trajectories and Full 
>>Transition Probabilities case? I've understood from the manual that the 
>>system "stays" in one path in the case of Individual Trajectories, but for 
>>the Full Transition
>> Probabilities how can we see the transition between the three scenarios' 
>> possible paths?
>>
>>4. Can we that MOST is a Two-Stage model in Full Stochastic Programming then?
>>
>>Right now Im only working with the full system, i.e. no contingencies.
>>
>>As usual, thank you so much in advance,
>>
>>-- 
>>Carlos Ferrandon

Re: Use of transmat matrix in full stochastic unit commitment

[Ray Daniel Zimmerman]

Hi Carlos,

1. You need not use ex_transmat() at all. It was simply a convenience function 
used to create the transmat cell array of transmission probability matrices for 
each period. You can write your own function to create your own general cell 
array of matrices which can certainly vary by period.

2. From Table 5-3 in the MOST User’s Manual, you’ll see that the 3rd dimension 
of the values field of a profile corresponds to the indices specified by the 
rows field. In the example rows = 1 and the 3rd dimension of values is a 
singleton. If you set rows to a vector, say rows  = [1; 3; 4], then values(:, 
:, 1), then values(:, :, 2) and then values(:, :, 3) would correspond to the 
values for wind units 1, 3, and 4 respectively.

3. I’m not sure I follow the question. In neither case are we optimizing 
explicit individual trajectories. We are optimizing the cost of a set of 
probability weighted scenarios (and transitions) with certain costs and 
constraints on the envelope of trajectories.

4. Not exactly. It is actually an approximation of a multi-stage problem, where 
the full multi-stage decision tree is approximated with a Markovian decision 
process. Reference [5] in the 
manual (also ref 5 at https://matpower.org/publications/) attempts to explain 
this.

Hope this helps,

Ray

[5] A. J. Lamadrid, D. Munoz-Alvarez, C. E. Murillo-Sanchez, R. D. Zimmerman, 
H. D. Shin and R. J. Thomas, “Using the MATPOWER Optimal Scheduling Tool to 
Test Power System Operation Methodologies Under Uncertainty,” Sustainable 
Energy, IEEE Transactions on, vol. 10, no. 3, pp. 1280–1289, July 2019. DOI: 
10.1109/TSTE.2018.2865454


On Jan 15, 2020, at 11:04 AM, Carlos Ferrandon Cervantes 
mailto:ferrand...@gmail.com>> wrote:

Hello everyone:

I've been working with MOST for some time now; specifically with the unit 
commitment problem. I have been able to add some linearised constraints to it, 
but now I face the challenge of using the stochastic programming option in 
MOST. I've followed the example "most_ex7_suc" for these cases: Stochastic Unit 
Commitment - Individual Trajectories and Stochastic Unit Commitment - Full 
Transition Probabilities and my questions are the next ones:

1. I've realised that in the matrix "transmat", only the first column is full 
with the three scenarios' probabilities, but for the rest of the columns we 
have the identity matrix size 3x3 for each time step.Then the "scenario_probs" 
variable is computed/updated with the operation:  mdi.tstep(t).TransMat * 
mdi.CostWeights(1, 1:mdi.idx.nj(t-1), t-1). My question is: Can these scenario 
probabilities change through time? I mean, that for every column we could have 
different scenario probabilities in "transmat". And if this was the case, what 
changes would be needed for the function "ex_transmat" to reflect these changes?

2. For the wind input in the example  "most_ex7_suc"  we have three different 
values of wind. I am assuming they belong to the three scenarios specified but 
only for that wind input. If we had different wind inputs from different buses, 
what would the arrangement of windprofiles.values(:,:,:) be?

3. How can we see the difference betwen the Individual Trajectories and Full 
Transition Probabilities case? I've understood from the manual that the system 
"stays" in one path in the case of Individual Trajectories, but for the Full 
Transition Probabilities how can we see the transition between the three 
scenarios' possible paths?

4. Can we that MOST is a Two-Stage model in Full Stochastic Programming then?

Right now Im only working with the full system, i.e. no contingencies.

As usual, thank you so much in advance,

--
Carlos Ferrandon

Use of transmat matrix in full stochastic unit commitment

[Carlos Ferrandon Cervantes]

Hello everyone:

I've been working with MOST for some time now; specifically with the unit
commitment problem. I have been able to add some linearised constraints to
it, but now I face the challenge of using the stochastic programming option
in MOST. I've followed the example "most_ex7_suc" for these
cases: Stochastic Unit Commitment - Individual Trajectories and Stochastic
Unit Commitment - Full Transition Probabilities and my questions are the
next ones:

1. I've realised that in the matrix "transmat", only the first column is
full with the three scenarios' probabilities, but for the rest of the
columns we have the identity matrix size 3x3 for each time step.Then the
"scenario_probs" variable is computed/updated with the
operation:  mdi.tstep(t).TransMat * mdi.CostWeights(1, 1:mdi.idx.nj(t-1),
t-1). My question is: Can these scenario probabilities change through time?
I mean, that for every column we could have different scenario
probabilities in "transmat". And if this was the case, what changes would
be needed for the function "ex_transmat" to reflect these changes?

2. For the wind input in the example  "most_ex7_suc"  we have three
different values of wind. I am assuming they belong to the three scenarios
specified but only for that wind input. If we had different wind inputs
from different buses, what would the arrangement of
windprofiles.values(:,:,:) be?

3. How can we see the difference betwen the Individual Trajectories and
Full Transition Probabilities case? I've understood from the manual that
the system "stays" in one path in the case of Individual Trajectories, but
for the Full Transition Probabilities how can we see the transition between
the three scenarios' possible paths?

4. Can we that MOST is a Two-Stage model in Full Stochastic Programming
then?

Right now Im only working with the full system, i.e. no contingencies.

As usual, thank you so much in advance,

-- 
Carlos Ferrandon