How do I evaluate interpolated data at unique points provided by two other matrices?

I have calculated thermocouple data as [174x1] vectors. I then used

Int_Thermocouple=triscatteredinterp(Engine_Speed,Engine_Torque,Thermocouple)

to make the thermocouple data a linear function of speed and torque.

I have a matrix Torque_Map [25x23] that has unique torque values in each cell, each row is calculated with respect to a different engine speed.

I used Speed_Map = meshgrid(Engine_Speed) and then resized it to [25x23] so that Torque_Map would have a matrix of the same size containing the corresponding engine speeds for each point within Torque_Map.

Now, here's where I'm having the problem.

I want to evaluate the thermocouple data at the unique speed and load points provided by Torque_Map and Speed_Map.

I've tried

output = Int_Thermocouple(Speed_Map,Torque_Map);

but this fails. How can I evaluate the thermocouple data at every point within Torque_Map and Speed_Map?

[dpb]

Would seem what you need is

[t,r]=meshgrid(torque,rpm);  % over the range of torques and speeds desired
T=int_Thermocouple(t,r);     % the interpolated temps

[Ross]

Unfortunately, meshgrid doesn't seem to be doing what I need. It works well for the speed map, as the rows are constant speed, but the torque map doesn't work that way. Each (i,j) within the torque map is unique. Also, I need the data to remain the same size. Using meshgrid(TQMAP,SPDMAP) I get a [575,575] output.

[dpb]

I guess I'm cornfoozed...if I understand correctly you've got a table of rpm and torque independent variables w/ observed temp's at those locations.

If I make up a smaller example like

>> sp=1000:250:1500   % three speeds
sp =
      1000        1250        1500
>> t=1300:125:1700      % four torques
t =
      1300        1425        1550        1675
>> [S,T]=meshgrid(sp',t)
S =
      1000        1250        1500
      1000        1250        1500
      1000        1250        1500
      1000        1250        1500
T =
      1300        1300        1300
      1425        1425        1425
      1550        1550        1550
      1675        1675        1675
>>

then the resulting mesgrid arrays are the combination of those 3-by-4 points. Wouldn't those be the points at which you want the interpolant?

ADDENDUM

Well, it dawns on me that isn't right, either but that I don't understand the 25x23 table. What is the variable w/ dimension 23?

ADDENDUM 2:

How about posting a very small subset of the data as an example for to use for experimentation/visualization.

[Ross]

>> TQMAP(1:5,1:5)

ans =

   12.8081   38.4337   64.0589   89.6844  115.3099
   13.5579   40.6834   67.8087   94.9342  122.0598
   14.4951   43.4956   72.4959  101.4965  130.4971
   15.3073   45.9329   76.5582  107.1838  137.8094
   16.0570   48.1827   80.3080  112.4336  144.5593

>> SPDMAP(1:5,1:5)

ans =

        2450        2450        2450        2450        2450
        2350        2350        2350        2350        2350
        2250        2250        2250        2250        2250
        2150        2150        2150        2150        2150
        2050        2050        2050        2050        2050

TQMAP has unique values because the columns are based on torque percentage, 100% torque at 1600 RPM is much greater than 100% torque at 800 RPM. The rows of TQMAP are with respect to RPM, which is constant across the rows. For example, TQMAP(1,1) would be 2.5% of the maximum torque at 2450 RPM, and TQMAP(2,1) would be 2.5% of the maximum torque at 2350 RPM. Wheras TQMAP(1,2) would be 7.5% of the maximum torque at 2450 RPM. There are 23 columns of torque percentages, and 25 rows of speeds.

SPDMAP is created via meshgrid because the the row values are constant, but in the case of TQMAP, I already have every value in matrix format. I need to evaluate my triscatteredinterp data (which is based on speed and load) to this different set of speed and load values contained within TQMAP(i,j) and SPDMAP(i,j).

[dpb]

OK, it's so much easier when can visualize the data to think about it....could you also add a sample of the TC data that can fit--I've not used the triscattered data fitter previously so need an example to play with -- might as well be actual numbers/sample space commensurate w/ your actual problem.